Advertisement
Guest User

ted_en-es_training

a guest
Feb 5th, 2019
257
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 266.96 KB | None | 0 0
  1.  
  2. The following have been reloaded with a version change:
  3. 1) cuDNN/7.4.1-CUDA-9.0.176 => cuDNN/7.0.5-CUDA-9.0.176
  4.  
  5.  
  6. The following have been reloaded with a version change:
  7. 1) CUDA/9.0.176 => CUDA/9.1.85
  8. 2) cuDNN/7.0.5-CUDA-9.0.176 => cuDNN/7.0.5-CUDA-9.1.85
  9.  
  10. [2019-02-04 17:50:15,386 INFO] * src vocab size = 17138
  11. [2019-02-04 17:50:15,387 INFO] * tgt vocab size = 21763
  12. [2019-02-04 17:50:15,388 INFO] Building model...
  13. [2019-02-04 17:50:30,601 INFO] NMTModel(
  14. (encoder): RNNEncoder(
  15. (embeddings): Embeddings(
  16. (make_embedding): Sequential(
  17. (emb_luts): Elementwise(
  18. (0): Embedding(17138, 512, padding_idx=1)
  19. )
  20. )
  21. )
  22. (rnn): LSTM(512, 256, num_layers=3, dropout=0.25, bidirectional=True)
  23. )
  24. (decoder): InputFeedRNNDecoder(
  25. (embeddings): Embeddings(
  26. (make_embedding): Sequential(
  27. (emb_luts): Elementwise(
  28. (0): Embedding(21763, 512, padding_idx=1)
  29. )
  30. )
  31. )
  32. (dropout): Dropout(p=0.25)
  33. (rnn): StackedLSTM(
  34. (dropout): Dropout(p=0.25)
  35. (layers): ModuleList(
  36. (0): LSTMCell(1024, 512)
  37. (1): LSTMCell(512, 512)
  38. (2): LSTMCell(512, 512)
  39. )
  40. )
  41. (attn): GlobalAttention(
  42. (linear_context): Linear(in_features=512, out_features=512, bias=False)
  43. (linear_query): Linear(in_features=512, out_features=512, bias=True)
  44. (v): Linear(in_features=512, out_features=1, bias=False)
  45. (linear_out): Linear(in_features=1024, out_features=512, bias=True)
  46. (linear_cover): Linear(in_features=1, out_features=512, bias=False)
  47. )
  48. )
  49. (generator): Sequential(
  50. (0): Linear(in_features=512, out_features=21763, bias=True)
  51. (1): LogSoftmax()
  52. )
  53. )
  54. [2019-02-04 17:50:30,603 INFO] encoder: 13505536
  55. [2019-02-04 17:50:30,603 INFO] decoder: 30710019
  56. [2019-02-04 17:50:30,604 INFO] * number of parameters: 44215555
  57. [2019-02-04 17:50:30,610 INFO] Starting training on GPU: [0]
  58. [2019-02-04 17:50:30,611 INFO] Start training...
  59. [2019-02-04 17:50:35,984 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  60. [2019-02-04 17:50:42,208 INFO] Step 50/102583; acc: 22.29; ppl: 2244.55; xent: 7.72; lr: 0.00010; 909/1133 tok/s; 12 sec
  61. [2019-02-04 17:50:47,054 INFO] Step 100/102583; acc: 23.00; ppl: 51.86; xent: 3.95; lr: 0.00010; 2014/2635 tok/s; 16 sec
  62. [2019-02-04 17:50:52,272 INFO] Step 150/102583; acc: 21.82; ppl: 24.18; xent: 3.19; lr: 0.00010; 2002/2624 tok/s; 22 sec
  63. [2019-02-04 17:50:57,151 INFO] Step 200/102583; acc: 26.59; ppl: 17.16; xent: 2.84; lr: 0.00010; 2006/2555 tok/s; 27 sec
  64. [2019-02-04 17:51:02,174 INFO] Step 250/102583; acc: 23.90; ppl: 17.22; xent: 2.85; lr: 0.00010; 2006/2672 tok/s; 32 sec
  65. [2019-02-04 17:51:07,262 INFO] Step 300/102583; acc: 27.05; ppl: 15.16; xent: 2.72; lr: 0.00010; 2033/2526 tok/s; 37 sec
  66. [2019-02-04 17:51:12,237 INFO] Step 350/102583; acc: 26.49; ppl: 16.15; xent: 2.78; lr: 0.00010; 2085/2628 tok/s; 42 sec
  67. [2019-02-04 17:51:17,271 INFO] Step 400/102583; acc: 28.90; ppl: 13.62; xent: 2.61; lr: 0.00010; 2002/2614 tok/s; 47 sec
  68. [2019-02-04 17:51:22,496 INFO] Step 450/102583; acc: 27.29; ppl: 15.19; xent: 2.72; lr: 0.00010; 1957/2524 tok/s; 52 sec
  69. [2019-02-04 17:51:27,372 INFO] Step 500/102583; acc: 33.66; ppl: 12.59; xent: 2.53; lr: 0.00010; 2074/2603 tok/s; 57 sec
  70. [2019-02-04 17:51:32,590 INFO] Step 550/102583; acc: 30.82; ppl: 13.15; xent: 2.58; lr: 0.00010; 1889/2633 tok/s; 62 sec
  71. [2019-02-04 17:51:37,543 INFO] Step 600/102583; acc: 37.45; ppl: 10.44; xent: 2.35; lr: 0.00010; 2151/2526 tok/s; 67 sec
  72. [2019-02-04 17:51:42,645 INFO] Step 650/102583; acc: 37.38; ppl: 10.46; xent: 2.35; lr: 0.00010; 2028/2532 tok/s; 72 sec
  73. [2019-02-04 17:51:47,611 INFO] Step 700/102583; acc: 36.88; ppl: 11.36; xent: 2.43; lr: 0.00010; 1997/2606 tok/s; 77 sec
  74. [2019-02-04 17:51:52,782 INFO] Step 750/102583; acc: 39.15; ppl: 9.41; xent: 2.24; lr: 0.00010; 2129/2609 tok/s; 82 sec
  75. [2019-02-04 17:51:57,713 INFO] Step 800/102583; acc: 42.00; ppl: 8.45; xent: 2.13; lr: 0.00010; 1953/2596 tok/s; 87 sec
  76. [2019-02-04 17:52:02,916 INFO] Step 850/102583; acc: 44.82; ppl: 7.38; xent: 2.00; lr: 0.00010; 1957/2467 tok/s; 92 sec
  77. [2019-02-04 17:52:08,053 INFO] Step 900/102583; acc: 45.57; ppl: 7.43; xent: 2.01; lr: 0.00010; 1964/2591 tok/s; 97 sec
  78. [2019-02-04 17:52:13,079 INFO] Step 950/102583; acc: 52.08; ppl: 5.78; xent: 1.75; lr: 0.00010; 2065/2411 tok/s; 102 sec
  79. [2019-02-04 17:52:18,112 INFO] Step 1000/102583; acc: 45.86; ppl: 6.88; xent: 1.93; lr: 0.00010; 1967/2734 tok/s; 108 sec
  80. [2019-02-04 17:52:23,449 INFO] Step 1050/102583; acc: 47.03; ppl: 6.43; xent: 1.86; lr: 0.00010; 1885/2573 tok/s; 113 sec
  81. [2019-02-04 17:52:28,272 INFO] Step 1100/102583; acc: 55.66; ppl: 5.08; xent: 1.62; lr: 0.00010; 2107/2556 tok/s; 118 sec
  82. [2019-02-04 17:52:33,504 INFO] Step 1150/102583; acc: 50.02; ppl: 5.70; xent: 1.74; lr: 0.00010; 2075/2497 tok/s; 123 sec
  83. [2019-02-04 17:52:38,365 INFO] Step 1200/102583; acc: 54.90; ppl: 5.00; xent: 1.61; lr: 0.00010; 1961/2639 tok/s; 128 sec
  84. [2019-02-04 17:52:43,057 INFO] Step 1250/102583; acc: 55.88; ppl: 4.84; xent: 1.58; lr: 0.00010; 2056/2644 tok/s; 132 sec
  85. [2019-02-04 17:52:48,423 INFO] Step 1300/102583; acc: 53.18; ppl: 5.24; xent: 1.66; lr: 0.00010; 1987/2524 tok/s; 138 sec
  86. [2019-02-04 17:52:53,820 INFO] Step 1350/102583; acc: 51.94; ppl: 5.23; xent: 1.66; lr: 0.00010; 2071/2480 tok/s; 143 sec
  87. [2019-02-04 17:52:58,611 INFO] Step 1400/102583; acc: 57.14; ppl: 4.48; xent: 1.50; lr: 0.00010; 1891/2668 tok/s; 148 sec
  88. [2019-02-04 17:53:03,815 INFO] Step 1450/102583; acc: 54.11; ppl: 4.90; xent: 1.59; lr: 0.00010; 2057/2630 tok/s; 153 sec
  89. [2019-02-04 17:53:08,635 INFO] Step 1500/102583; acc: 62.25; ppl: 3.99; xent: 1.38; lr: 0.00010; 1983/2510 tok/s; 158 sec
  90. [2019-02-04 17:53:13,493 INFO] Step 1550/102583; acc: 58.04; ppl: 4.39; xent: 1.48; lr: 0.00010; 1916/2629 tok/s; 163 sec
  91. [2019-02-04 17:53:18,709 INFO] Step 1600/102583; acc: 58.65; ppl: 4.17; xent: 1.43; lr: 0.00010; 2067/2504 tok/s; 168 sec
  92. [2019-02-04 17:53:23,662 INFO] Step 1650/102583; acc: 55.29; ppl: 4.58; xent: 1.52; lr: 0.00010; 2055/2769 tok/s; 173 sec
  93. [2019-02-04 17:53:28,697 INFO] Step 1700/102583; acc: 63.21; ppl: 3.72; xent: 1.31; lr: 0.00010; 2035/2478 tok/s; 178 sec
  94. [2019-02-04 17:53:33,790 INFO] Step 1750/102583; acc: 59.42; ppl: 4.12; xent: 1.42; lr: 0.00010; 1877/2713 tok/s; 183 sec
  95. [2019-02-04 17:53:38,694 INFO] Step 1800/102583; acc: 61.78; ppl: 3.76; xent: 1.32; lr: 0.00010; 2176/2483 tok/s; 188 sec
  96. [2019-02-04 17:53:43,834 INFO] Step 1850/102583; acc: 57.93; ppl: 4.27; xent: 1.45; lr: 0.00010; 1867/2648 tok/s; 193 sec
  97. [2019-02-04 17:53:48,773 INFO] Step 1900/102583; acc: 63.80; ppl: 3.47; xent: 1.24; lr: 0.00010; 2143/2505 tok/s; 198 sec
  98. [2019-02-04 17:53:53,971 INFO] Step 1950/102583; acc: 61.64; ppl: 3.82; xent: 1.34; lr: 0.00010; 2121/2602 tok/s; 203 sec
  99. [2019-02-04 17:53:58,951 INFO] Step 2000/102583; acc: 61.96; ppl: 3.73; xent: 1.32; lr: 0.00010; 1880/2533 tok/s; 208 sec
  100. [2019-02-04 17:54:03,869 INFO] Step 2050/102583; acc: 63.81; ppl: 3.57; xent: 1.27; lr: 0.00010; 1931/2618 tok/s; 213 sec
  101. [2019-02-04 17:54:08,976 INFO] Step 2100/102583; acc: 63.32; ppl: 3.56; xent: 1.27; lr: 0.00010; 2110/2528 tok/s; 218 sec
  102. [2019-02-04 17:54:13,788 INFO] Step 2150/102583; acc: 65.30; ppl: 3.48; xent: 1.25; lr: 0.00010; 1983/2676 tok/s; 223 sec
  103. [2019-02-04 17:54:18,872 INFO] Step 2200/102583; acc: 63.72; ppl: 3.46; xent: 1.24; lr: 0.00010; 2090/2552 tok/s; 228 sec
  104. [2019-02-04 17:54:23,653 INFO] Step 2250/102583; acc: 67.05; ppl: 3.14; xent: 1.14; lr: 0.00010; 2118/2578 tok/s; 233 sec
  105. [2019-02-04 17:54:28,954 INFO] Step 2300/102583; acc: 62.97; ppl: 3.51; xent: 1.26; lr: 0.00010; 1900/2598 tok/s; 238 sec
  106. [2019-02-04 17:54:34,018 INFO] Step 2350/102583; acc: 65.63; ppl: 3.23; xent: 1.17; lr: 0.00010; 2041/2646 tok/s; 243 sec
  107. [2019-02-04 17:54:38,901 INFO] Step 2400/102583; acc: 64.94; ppl: 3.42; xent: 1.23; lr: 0.00010; 2024/2603 tok/s; 248 sec
  108. [2019-02-04 17:54:43,995 INFO] Step 2450/102583; acc: 67.50; ppl: 3.16; xent: 1.15; lr: 0.00010; 2055/2572 tok/s; 253 sec
  109. [2019-02-04 17:54:48,881 INFO] Step 2500/102583; acc: 65.04; ppl: 3.33; xent: 1.20; lr: 0.00010; 2019/2659 tok/s; 258 sec
  110. [2019-02-04 17:54:53,544 INFO] Step 2550/102583; acc: 69.14; ppl: 2.89; xent: 1.06; lr: 0.00010; 2085/2576 tok/s; 263 sec
  111. [2019-02-04 17:54:58,945 INFO] Step 2600/102583; acc: 65.39; ppl: 3.37; xent: 1.22; lr: 0.00010; 1972/2632 tok/s; 268 sec
  112. [2019-02-04 17:55:03,881 INFO] Step 2650/102583; acc: 69.14; ppl: 2.88; xent: 1.06; lr: 0.00010; 2192/2504 tok/s; 273 sec
  113. [2019-02-04 17:55:08,949 INFO] Step 2700/102583; acc: 66.56; ppl: 3.29; xent: 1.19; lr: 0.00010; 1846/2700 tok/s; 278 sec
  114. [2019-02-04 17:55:14,115 INFO] Step 2750/102583; acc: 64.83; ppl: 3.38; xent: 1.22; lr: 0.00010; 1905/2532 tok/s; 284 sec
  115. [2019-02-04 17:55:19,022 INFO] Step 2800/102583; acc: 71.06; ppl: 2.71; xent: 1.00; lr: 0.00010; 2140/2657 tok/s; 288 sec
  116. [2019-02-04 17:55:23,876 INFO] Step 2850/102583; acc: 71.02; ppl: 2.76; xent: 1.01; lr: 0.00010; 1878/2701 tok/s; 293 sec
  117. [2019-02-04 17:55:29,154 INFO] Step 2900/102583; acc: 67.34; ppl: 3.11; xent: 1.14; lr: 0.00010; 2123/2482 tok/s; 299 sec
  118. [2019-02-04 17:55:34,035 INFO] Step 2950/102583; acc: 70.86; ppl: 2.81; xent: 1.03; lr: 0.00010; 2013/2682 tok/s; 303 sec
  119. [2019-02-04 17:55:39,171 INFO] Step 3000/102583; acc: 67.21; ppl: 3.04; xent: 1.11; lr: 0.00010; 2067/2480 tok/s; 309 sec
  120. [2019-02-04 17:55:44,325 INFO] Step 3050/102583; acc: 68.09; ppl: 3.00; xent: 1.10; lr: 0.00010; 2116/2585 tok/s; 314 sec
  121. [2019-02-04 17:55:49,150 INFO] Step 3100/102583; acc: 71.07; ppl: 2.72; xent: 1.00; lr: 0.00010; 1962/2655 tok/s; 319 sec
  122. [2019-02-04 17:55:53,813 INFO] Step 3150/102583; acc: 72.80; ppl: 2.59; xent: 0.95; lr: 0.00010; 2109/2564 tok/s; 323 sec
  123. [2019-02-04 17:55:59,247 INFO] Step 3200/102583; acc: 68.24; ppl: 2.99; xent: 1.10; lr: 0.00010; 1914/2608 tok/s; 329 sec
  124. [2019-02-04 17:56:08,691 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  125. [2019-02-04 17:56:09,927 INFO] Step 3250/102583; acc: 71.12; ppl: 2.79; xent: 1.03; lr: 0.00010; 922/1178 tok/s; 339 sec
  126. [2019-02-04 17:56:15,038 INFO] Step 3300/102583; acc: 71.52; ppl: 2.69; xent: 0.99; lr: 0.00010; 2126/2531 tok/s; 344 sec
  127. [2019-02-04 17:56:19,809 INFO] Step 3350/102583; acc: 71.30; ppl: 2.68; xent: 0.98; lr: 0.00010; 1952/2681 tok/s; 349 sec
  128. [2019-02-04 17:56:25,104 INFO] Step 3400/102583; acc: 71.99; ppl: 2.62; xent: 0.96; lr: 0.00010; 2033/2637 tok/s; 354 sec
  129. [2019-02-04 17:56:29,953 INFO] Step 3450/102583; acc: 71.08; ppl: 2.69; xent: 0.99; lr: 0.00010; 2073/2600 tok/s; 359 sec
  130. [2019-02-04 17:56:35,001 INFO] Step 3500/102583; acc: 72.46; ppl: 2.62; xent: 0.96; lr: 0.00010; 2074/2637 tok/s; 364 sec
  131. [2019-02-04 17:56:40,046 INFO] Step 3550/102583; acc: 69.78; ppl: 2.77; xent: 1.02; lr: 0.00010; 2007/2597 tok/s; 369 sec
  132. [2019-02-04 17:56:44,837 INFO] Step 3600/102583; acc: 73.71; ppl: 2.47; xent: 0.90; lr: 0.00010; 2113/2745 tok/s; 374 sec
  133. [2019-02-04 17:56:49,852 INFO] Step 3650/102583; acc: 70.85; ppl: 2.71; xent: 1.00; lr: 0.00010; 1907/2675 tok/s; 379 sec
  134. [2019-02-04 17:56:55,015 INFO] Step 3700/102583; acc: 71.95; ppl: 2.67; xent: 0.98; lr: 0.00010; 2076/2465 tok/s; 384 sec
  135. [2019-02-04 17:56:59,726 INFO] Step 3750/102583; acc: 72.81; ppl: 2.52; xent: 0.93; lr: 0.00010; 2070/2666 tok/s; 389 sec
  136. [2019-02-04 17:57:04,979 INFO] Step 3800/102583; acc: 70.89; ppl: 2.72; xent: 1.00; lr: 0.00010; 1972/2642 tok/s; 394 sec
  137. [2019-02-04 17:57:09,688 INFO] Step 3850/102583; acc: 75.06; ppl: 2.37; xent: 0.86; lr: 0.00010; 2170/2522 tok/s; 399 sec
  138. [2019-02-04 17:57:14,764 INFO] Step 3900/102583; acc: 71.22; ppl: 2.64; xent: 0.97; lr: 0.00010; 2026/2585 tok/s; 404 sec
  139. [2019-02-04 17:57:19,725 INFO] Step 3950/102583; acc: 74.64; ppl: 2.45; xent: 0.90; lr: 0.00010; 2038/2636 tok/s; 409 sec
  140. [2019-02-04 17:57:24,985 INFO] Step 4000/102583; acc: 72.62; ppl: 2.53; xent: 0.93; lr: 0.00010; 2134/2582 tok/s; 414 sec
  141. [2019-02-04 17:57:29,779 INFO] Step 4050/102583; acc: 73.47; ppl: 2.50; xent: 0.92; lr: 0.00010; 1990/2623 tok/s; 419 sec
  142. [2019-02-04 17:57:34,869 INFO] Step 4100/102583; acc: 73.40; ppl: 2.48; xent: 0.91; lr: 0.00010; 1989/2556 tok/s; 424 sec
  143. [2019-02-04 17:57:40,017 INFO] Step 4150/102583; acc: 75.85; ppl: 2.33; xent: 0.84; lr: 0.00010; 1972/2624 tok/s; 429 sec
  144. [2019-02-04 17:57:44,991 INFO] Step 4200/102583; acc: 72.46; ppl: 2.57; xent: 0.94; lr: 0.00010; 2049/2456 tok/s; 434 sec
  145. [2019-02-04 17:57:50,066 INFO] Step 4250/102583; acc: 73.46; ppl: 2.48; xent: 0.91; lr: 0.00010; 1986/2714 tok/s; 439 sec
  146. [2019-02-04 17:57:55,363 INFO] Step 4300/102583; acc: 75.00; ppl: 2.39; xent: 0.87; lr: 0.00010; 1916/2528 tok/s; 445 sec
  147. [2019-02-04 17:58:00,420 INFO] Step 4350/102583; acc: 74.95; ppl: 2.37; xent: 0.86; lr: 0.00010; 2160/2599 tok/s; 450 sec
  148. [2019-02-04 17:58:05,244 INFO] Step 4400/102583; acc: 73.52; ppl: 2.48; xent: 0.91; lr: 0.00010; 2071/2570 tok/s; 455 sec
  149. [2019-02-04 17:58:10,107 INFO] Step 4450/102583; acc: 72.24; ppl: 2.53; xent: 0.93; lr: 0.00010; 1932/2692 tok/s; 459 sec
  150. [2019-02-04 17:58:14,885 INFO] Step 4500/102583; acc: 76.38; ppl: 2.30; xent: 0.83; lr: 0.00010; 2093/2611 tok/s; 464 sec
  151. [2019-02-04 17:58:20,251 INFO] Step 4550/102583; acc: 74.49; ppl: 2.40; xent: 0.87; lr: 0.00010; 2050/2464 tok/s; 470 sec
  152. [2019-02-04 17:58:25,401 INFO] Step 4600/102583; acc: 74.40; ppl: 2.39; xent: 0.87; lr: 0.00010; 2049/2574 tok/s; 475 sec
  153. [2019-02-04 17:58:30,362 INFO] Step 4650/102583; acc: 74.07; ppl: 2.42; xent: 0.88; lr: 0.00010; 1811/2706 tok/s; 480 sec
  154. [2019-02-04 17:58:35,372 INFO] Step 4700/102583; acc: 74.73; ppl: 2.37; xent: 0.86; lr: 0.00010; 2152/2652 tok/s; 485 sec
  155. [2019-02-04 17:58:40,258 INFO] Step 4750/102583; acc: 76.38; ppl: 2.29; xent: 0.83; lr: 0.00010; 1961/2492 tok/s; 490 sec
  156. [2019-02-04 17:58:44,977 INFO] Step 4800/102583; acc: 75.65; ppl: 2.38; xent: 0.87; lr: 0.00010; 1961/2607 tok/s; 494 sec
  157. [2019-02-04 17:58:50,278 INFO] Step 4850/102583; acc: 74.95; ppl: 2.30; xent: 0.83; lr: 0.00010; 2094/2544 tok/s; 500 sec
  158. [2019-02-04 17:58:55,205 INFO] Step 4900/102583; acc: 72.58; ppl: 2.49; xent: 0.91; lr: 0.00010; 1976/2755 tok/s; 505 sec
  159. [2019-02-04 17:59:00,258 INFO] Step 4950/102583; acc: 76.83; ppl: 2.21; xent: 0.79; lr: 0.00010; 2084/2570 tok/s; 510 sec
  160. [2019-02-04 17:59:05,386 INFO] Step 5000/102583; acc: 76.24; ppl: 2.26; xent: 0.81; lr: 0.00010; 1995/2649 tok/s; 515 sec
  161. [2019-02-04 17:59:05,388 INFO] Saving checkpoint models/ted1_step_5000.pt
  162. [2019-02-04 17:59:15,964 INFO] Step 5050/102583; acc: 74.93; ppl: 2.33; xent: 0.85; lr: 0.00010; 890/1136 tok/s; 525 sec
  163. [2019-02-04 17:59:21,080 INFO] Step 5100/102583; acc: 75.06; ppl: 2.36; xent: 0.86; lr: 0.00010; 1976/2614 tok/s; 530 sec
  164. [2019-02-04 17:59:26,010 INFO] Step 5150/102583; acc: 76.50; ppl: 2.21; xent: 0.79; lr: 0.00010; 2121/2543 tok/s; 535 sec
  165. [2019-02-04 17:59:31,329 INFO] Step 5200/102583; acc: 76.07; ppl: 2.29; xent: 0.83; lr: 0.00010; 2142/2523 tok/s; 541 sec
  166. [2019-02-04 17:59:36,076 INFO] Step 5250/102583; acc: 76.86; ppl: 2.24; xent: 0.81; lr: 0.00010; 1843/2617 tok/s; 545 sec
  167. [2019-02-04 17:59:41,041 INFO] Step 5300/102583; acc: 74.85; ppl: 2.37; xent: 0.86; lr: 0.00010; 1941/2644 tok/s; 550 sec
  168. [2019-02-04 17:59:46,178 INFO] Step 5350/102583; acc: 75.97; ppl: 2.28; xent: 0.82; lr: 0.00010; 2072/2528 tok/s; 556 sec
  169. [2019-02-04 17:59:51,002 INFO] Step 5400/102583; acc: 77.61; ppl: 2.15; xent: 0.76; lr: 0.00010; 2077/2705 tok/s; 560 sec
  170. [2019-02-04 17:59:56,070 INFO] Step 5450/102583; acc: 75.00; ppl: 2.30; xent: 0.83; lr: 0.00010; 2097/2457 tok/s; 565 sec
  171. [2019-02-04 18:00:00,830 INFO] Step 5500/102583; acc: 77.01; ppl: 2.18; xent: 0.78; lr: 0.00010; 2099/2705 tok/s; 570 sec
  172. [2019-02-04 18:00:06,045 INFO] Step 5550/102583; acc: 76.62; ppl: 2.23; xent: 0.80; lr: 0.00010; 1938/2623 tok/s; 575 sec
  173. [2019-02-04 18:00:11,035 INFO] Step 5600/102583; acc: 76.24; ppl: 2.19; xent: 0.78; lr: 0.00010; 2033/2642 tok/s; 580 sec
  174. [2019-02-04 18:00:15,993 INFO] Step 5650/102583; acc: 76.74; ppl: 2.23; xent: 0.80; lr: 0.00010; 2073/2496 tok/s; 585 sec
  175. [2019-02-04 18:00:20,970 INFO] Step 5700/102583; acc: 77.77; ppl: 2.13; xent: 0.76; lr: 0.00010; 2011/2698 tok/s; 590 sec
  176. [2019-02-04 18:00:25,876 INFO] Step 5750/102583; acc: 74.28; ppl: 2.34; xent: 0.85; lr: 0.00010; 2037/2681 tok/s; 595 sec
  177. [2019-02-04 18:00:30,733 INFO] Step 5800/102583; acc: 77.20; ppl: 2.17; xent: 0.78; lr: 0.00010; 2031/2521 tok/s; 600 sec
  178. [2019-02-04 18:00:35,853 INFO] Step 5850/102583; acc: 76.95; ppl: 2.20; xent: 0.79; lr: 0.00010; 2028/2652 tok/s; 605 sec
  179. [2019-02-04 18:00:40,903 INFO] Step 5900/102583; acc: 77.43; ppl: 2.15; xent: 0.77; lr: 0.00010; 2169/2533 tok/s; 610 sec
  180. [2019-02-04 18:00:46,094 INFO] Step 5950/102583; acc: 75.64; ppl: 2.28; xent: 0.82; lr: 0.00010; 1754/2766 tok/s; 615 sec
  181. [2019-02-04 18:00:50,983 INFO] Step 6000/102583; acc: 75.06; ppl: 2.37; xent: 0.86; lr: 0.00010; 2062/2456 tok/s; 620 sec
  182. [2019-02-04 18:00:55,907 INFO] Step 6050/102583; acc: 78.94; ppl: 2.04; xent: 0.71; lr: 0.00010; 2093/2686 tok/s; 625 sec
  183. [2019-02-04 18:01:00,747 INFO] Step 6100/102583; acc: 77.20; ppl: 2.15; xent: 0.77; lr: 0.00010; 1919/2747 tok/s; 630 sec
  184. [2019-02-04 18:01:06,081 INFO] Step 6150/102583; acc: 75.65; ppl: 2.28; xent: 0.83; lr: 0.00010; 2118/2451 tok/s; 635 sec
  185. [2019-02-04 18:01:10,919 INFO] Step 6200/102583; acc: 78.35; ppl: 2.09; xent: 0.73; lr: 0.00010; 2081/2675 tok/s; 640 sec
  186. [2019-02-04 18:01:16,071 INFO] Step 6250/102583; acc: 76.69; ppl: 2.17; xent: 0.78; lr: 0.00010; 2114/2477 tok/s; 645 sec
  187. [2019-02-04 18:01:21,214 INFO] Step 6300/102583; acc: 75.80; ppl: 2.28; xent: 0.82; lr: 0.00010; 1984/2628 tok/s; 651 sec
  188. [2019-02-04 18:01:25,902 INFO] Step 6350/102583; acc: 78.64; ppl: 2.07; xent: 0.73; lr: 0.00010; 2010/2689 tok/s; 655 sec
  189. [2019-02-04 18:01:30,553 INFO] Step 6400/102583; acc: 78.21; ppl: 2.08; xent: 0.73; lr: 0.00010; 2128/2574 tok/s; 660 sec
  190. [2019-02-04 18:01:35,920 INFO] Step 6450/102583; acc: 76.03; ppl: 2.23; xent: 0.80; lr: 0.00010; 1942/2587 tok/s; 665 sec
  191. [2019-02-04 18:01:45,215 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  192. [2019-02-04 18:01:47,009 INFO] Step 6500/102583; acc: 76.37; ppl: 2.24; xent: 0.81; lr: 0.00010; 889/1160 tok/s; 676 sec
  193. [2019-02-04 18:01:52,132 INFO] Step 6550/102583; acc: 78.30; ppl: 2.11; xent: 0.75; lr: 0.00010; 2149/2532 tok/s; 682 sec
  194. [2019-02-04 18:01:56,882 INFO] Step 6600/102583; acc: 78.17; ppl: 2.10; xent: 0.74; lr: 0.00010; 1909/2708 tok/s; 686 sec
  195. [2019-02-04 18:02:02,061 INFO] Step 6650/102583; acc: 79.22; ppl: 2.02; xent: 0.70; lr: 0.00010; 2061/2631 tok/s; 691 sec
  196. [2019-02-04 18:02:06,992 INFO] Step 6700/102583; acc: 77.14; ppl: 2.17; xent: 0.78; lr: 0.00010; 2143/2653 tok/s; 696 sec
  197. [2019-02-04 18:02:12,062 INFO] Step 6750/102583; acc: 78.50; ppl: 2.07; xent: 0.73; lr: 0.00010; 2077/2654 tok/s; 701 sec
  198. [2019-02-04 18:02:16,948 INFO] Step 6800/102583; acc: 76.57; ppl: 2.18; xent: 0.78; lr: 0.00010; 1994/2527 tok/s; 706 sec
  199. [2019-02-04 18:02:21,890 INFO] Step 6850/102583; acc: 77.54; ppl: 2.11; xent: 0.75; lr: 0.00010; 2069/2768 tok/s; 711 sec
  200. [2019-02-04 18:02:26,753 INFO] Step 6900/102583; acc: 76.60; ppl: 2.15; xent: 0.77; lr: 0.00010; 1985/2689 tok/s; 716 sec
  201. [2019-02-04 18:02:31,985 INFO] Step 6950/102583; acc: 75.94; ppl: 2.27; xent: 0.82; lr: 0.00010; 1987/2507 tok/s; 721 sec
  202. [2019-02-04 18:02:36,690 INFO] Step 7000/102583; acc: 79.45; ppl: 1.98; xent: 0.68; lr: 0.00010; 2208/2641 tok/s; 726 sec
  203. [2019-02-04 18:02:42,041 INFO] Step 7050/102583; acc: 76.48; ppl: 2.20; xent: 0.79; lr: 0.00010; 1956/2590 tok/s; 731 sec
  204. [2019-02-04 18:02:46,540 INFO] Step 7100/102583; acc: 77.86; ppl: 2.13; xent: 0.76; lr: 0.00010; 2105/2599 tok/s; 736 sec
  205. [2019-02-04 18:02:51,777 INFO] Step 7150/102583; acc: 76.74; ppl: 2.19; xent: 0.78; lr: 0.00010; 1976/2661 tok/s; 741 sec
  206. [2019-02-04 18:02:56,592 INFO] Step 7200/102583; acc: 79.80; ppl: 2.01; xent: 0.70; lr: 0.00010; 2126/2585 tok/s; 746 sec
  207. [2019-02-04 18:03:01,817 INFO] Step 7250/102583; acc: 78.44; ppl: 2.05; xent: 0.72; lr: 0.00010; 2111/2605 tok/s; 751 sec
  208. [2019-02-04 18:03:06,526 INFO] Step 7300/102583; acc: 78.02; ppl: 2.11; xent: 0.75; lr: 0.00010; 2054/2577 tok/s; 756 sec
  209. [2019-02-04 18:03:11,642 INFO] Step 7350/102583; acc: 77.96; ppl: 2.07; xent: 0.73; lr: 0.00010; 1966/2592 tok/s; 761 sec
  210. [2019-02-04 18:03:16,800 INFO] Step 7400/102583; acc: 78.92; ppl: 2.04; xent: 0.71; lr: 0.00010; 2005/2631 tok/s; 766 sec
  211. [2019-02-04 18:03:21,826 INFO] Step 7450/102583; acc: 78.28; ppl: 2.10; xent: 0.74; lr: 0.00010; 2068/2421 tok/s; 771 sec
  212. [2019-02-04 18:03:26,847 INFO] Step 7500/102583; acc: 76.10; ppl: 2.18; xent: 0.78; lr: 0.00010; 1930/2767 tok/s; 776 sec
  213. [2019-02-04 18:03:32,101 INFO] Step 7550/102583; acc: 80.03; ppl: 1.98; xent: 0.68; lr: 0.00010; 1956/2500 tok/s; 781 sec
  214. [2019-02-04 18:03:37,146 INFO] Step 7600/102583; acc: 78.67; ppl: 2.06; xent: 0.72; lr: 0.00010; 2166/2635 tok/s; 787 sec
  215. [2019-02-04 18:03:41,948 INFO] Step 7650/102583; acc: 77.05; ppl: 2.16; xent: 0.77; lr: 0.00010; 2094/2558 tok/s; 791 sec
  216. [2019-02-04 18:03:46,910 INFO] Step 7700/102583; acc: 76.89; ppl: 2.16; xent: 0.77; lr: 0.00010; 1879/2678 tok/s; 796 sec
  217. [2019-02-04 18:03:51,847 INFO] Step 7750/102583; acc: 78.73; ppl: 2.06; xent: 0.73; lr: 0.00010; 2152/2686 tok/s; 801 sec
  218. [2019-02-04 18:03:56,905 INFO] Step 7800/102583; acc: 79.56; ppl: 2.01; xent: 0.70; lr: 0.00010; 2092/2400 tok/s; 806 sec
  219. [2019-02-04 18:04:02,176 INFO] Step 7850/102583; acc: 78.29; ppl: 2.06; xent: 0.72; lr: 0.00010; 1957/2648 tok/s; 812 sec
  220. [2019-02-04 18:04:06,971 INFO] Step 7900/102583; acc: 77.98; ppl: 2.08; xent: 0.73; lr: 0.00010; 1834/2663 tok/s; 816 sec
  221. [2019-02-04 18:04:12,100 INFO] Step 7950/102583; acc: 77.95; ppl: 2.09; xent: 0.74; lr: 0.00010; 2126/2644 tok/s; 821 sec
  222. [2019-02-04 18:04:16,824 INFO] Step 8000/102583; acc: 79.33; ppl: 2.04; xent: 0.72; lr: 0.00010; 2029/2504 tok/s; 826 sec
  223. [2019-02-04 18:04:21,607 INFO] Step 8050/102583; acc: 78.64; ppl: 2.08; xent: 0.73; lr: 0.00010; 1975/2649 tok/s; 831 sec
  224. [2019-02-04 18:04:26,962 INFO] Step 8100/102583; acc: 78.51; ppl: 2.01; xent: 0.70; lr: 0.00010; 2073/2593 tok/s; 836 sec
  225. [2019-02-04 18:04:31,910 INFO] Step 8150/102583; acc: 76.97; ppl: 2.14; xent: 0.76; lr: 0.00010; 2049/2640 tok/s; 841 sec
  226. [2019-02-04 18:04:36,860 INFO] Step 8200/102583; acc: 79.74; ppl: 1.99; xent: 0.69; lr: 0.00010; 2007/2659 tok/s; 846 sec
  227. [2019-02-04 18:04:41,880 INFO] Step 8250/102583; acc: 79.16; ppl: 2.02; xent: 0.70; lr: 0.00010; 2063/2613 tok/s; 851 sec
  228. [2019-02-04 18:04:46,674 INFO] Step 8300/102583; acc: 77.07; ppl: 2.14; xent: 0.76; lr: 0.00010; 1936/2536 tok/s; 856 sec
  229. [2019-02-04 18:04:51,823 INFO] Step 8350/102583; acc: 78.96; ppl: 2.01; xent: 0.70; lr: 0.00010; 2044/2674 tok/s; 861 sec
  230. [2019-02-04 18:04:56,772 INFO] Step 8400/102583; acc: 78.37; ppl: 2.06; xent: 0.72; lr: 0.00010; 2090/2555 tok/s; 866 sec
  231. [2019-02-04 18:05:02,098 INFO] Step 8450/102583; acc: 79.00; ppl: 2.02; xent: 0.70; lr: 0.00010; 2129/2458 tok/s; 871 sec
  232. [2019-02-04 18:05:06,683 INFO] Step 8500/102583; acc: 79.68; ppl: 2.00; xent: 0.69; lr: 0.00010; 1924/2724 tok/s; 876 sec
  233. [2019-02-04 18:05:11,658 INFO] Step 8550/102583; acc: 76.34; ppl: 2.21; xent: 0.79; lr: 0.00010; 1937/2624 tok/s; 881 sec
  234. [2019-02-04 18:05:16,763 INFO] Step 8600/102583; acc: 79.71; ppl: 1.99; xent: 0.69; lr: 0.00010; 1997/2574 tok/s; 886 sec
  235. [2019-02-04 18:05:21,524 INFO] Step 8650/102583; acc: 79.90; ppl: 1.95; xent: 0.67; lr: 0.00010; 2105/2677 tok/s; 891 sec
  236. [2019-02-04 18:05:26,570 INFO] Step 8700/102583; acc: 77.60; ppl: 2.11; xent: 0.75; lr: 0.00010; 2118/2462 tok/s; 896 sec
  237. [2019-02-04 18:05:31,693 INFO] Step 8750/102583; acc: 80.29; ppl: 1.91; xent: 0.65; lr: 0.00010; 2138/2590 tok/s; 901 sec
  238. [2019-02-04 18:05:36,495 INFO] Step 8800/102583; acc: 78.02; ppl: 2.08; xent: 0.73; lr: 0.00010; 1853/2780 tok/s; 906 sec
  239. [2019-02-04 18:05:41,602 INFO] Step 8850/102583; acc: 78.33; ppl: 2.04; xent: 0.71; lr: 0.00010; 2051/2655 tok/s; 911 sec
  240. [2019-02-04 18:05:46,583 INFO] Step 8900/102583; acc: 80.58; ppl: 1.93; xent: 0.66; lr: 0.00010; 2138/2522 tok/s; 916 sec
  241. [2019-02-04 18:05:51,505 INFO] Step 8950/102583; acc: 79.49; ppl: 2.00; xent: 0.69; lr: 0.00010; 2006/2618 tok/s; 921 sec
  242. [2019-02-04 18:05:56,544 INFO] Step 9000/102583; acc: 77.81; ppl: 2.05; xent: 0.72; lr: 0.00010; 2068/2652 tok/s; 926 sec
  243. [2019-02-04 18:06:01,241 INFO] Step 9050/102583; acc: 78.59; ppl: 2.06; xent: 0.72; lr: 0.00010; 2011/2618 tok/s; 931 sec
  244. [2019-02-04 18:06:06,298 INFO] Step 9100/102583; acc: 79.34; ppl: 2.00; xent: 0.69; lr: 0.00010; 2091/2615 tok/s; 936 sec
  245. [2019-02-04 18:06:11,430 INFO] Step 9150/102583; acc: 79.83; ppl: 1.95; xent: 0.67; lr: 0.00010; 2085/2544 tok/s; 941 sec
  246. [2019-02-04 18:06:16,509 INFO] Step 9200/102583; acc: 77.93; ppl: 2.09; xent: 0.74; lr: 0.00010; 1743/2830 tok/s; 946 sec
  247. [2019-02-04 18:06:21,397 INFO] Step 9250/102583; acc: 77.01; ppl: 2.19; xent: 0.78; lr: 0.00010; 2010/2444 tok/s; 951 sec
  248. [2019-02-04 18:06:26,346 INFO] Step 9300/102583; acc: 81.44; ppl: 1.86; xent: 0.62; lr: 0.00010; 2173/2724 tok/s; 956 sec
  249. [2019-02-04 18:06:31,126 INFO] Step 9350/102583; acc: 79.46; ppl: 1.99; xent: 0.69; lr: 0.00010; 2010/2670 tok/s; 961 sec
  250. [2019-02-04 18:06:36,498 INFO] Step 9400/102583; acc: 79.04; ppl: 2.03; xent: 0.71; lr: 0.00010; 2043/2523 tok/s; 966 sec
  251. [2019-02-04 18:06:41,299 INFO] Step 9450/102583; acc: 78.68; ppl: 2.04; xent: 0.71; lr: 0.00010; 2057/2678 tok/s; 971 sec
  252. [2019-02-04 18:06:46,454 INFO] Step 9500/102583; acc: 80.20; ppl: 1.94; xent: 0.66; lr: 0.00010; 2152/2466 tok/s; 976 sec
  253. [2019-02-04 18:06:51,531 INFO] Step 9550/102583; acc: 77.93; ppl: 2.11; xent: 0.74; lr: 0.00010; 2035/2593 tok/s; 981 sec
  254. [2019-02-04 18:06:56,188 INFO] Step 9600/102583; acc: 79.44; ppl: 1.99; xent: 0.69; lr: 0.00010; 1941/2698 tok/s; 986 sec
  255. [2019-02-04 18:07:01,021 INFO] Step 9650/102583; acc: 79.67; ppl: 1.96; xent: 0.67; lr: 0.00010; 2101/2565 tok/s; 990 sec
  256. [2019-02-04 18:07:06,308 INFO] Step 9700/102583; acc: 78.58; ppl: 2.01; xent: 0.70; lr: 0.00010; 1999/2607 tok/s; 996 sec
  257. [2019-02-04 18:07:15,279 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  258. [2019-02-04 18:07:17,550 INFO] Step 9750/102583; acc: 77.72; ppl: 2.11; xent: 0.75; lr: 0.00010; 861/1131 tok/s; 1007 sec
  259. [2019-02-04 18:07:22,633 INFO] Step 9800/102583; acc: 80.48; ppl: 1.96; xent: 0.67; lr: 0.00010; 2166/2508 tok/s; 1012 sec
  260. [2019-02-04 18:07:27,427 INFO] Step 9850/102583; acc: 79.88; ppl: 1.97; xent: 0.68; lr: 0.00010; 1878/2747 tok/s; 1017 sec
  261. [2019-02-04 18:07:32,638 INFO] Step 9900/102583; acc: 81.32; ppl: 1.88; xent: 0.63; lr: 0.00010; 2074/2632 tok/s; 1022 sec
  262. [2019-02-04 18:07:37,522 INFO] Step 9950/102583; acc: 78.55; ppl: 2.03; xent: 0.71; lr: 0.00010; 2098/2693 tok/s; 1027 sec
  263. [2019-02-04 18:07:42,740 INFO] Step 10000/102583; acc: 79.83; ppl: 1.94; xent: 0.66; lr: 0.00010; 2047/2677 tok/s; 1032 sec
  264. [2019-02-04 18:07:42,809 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  265. [2019-02-04 18:07:44,105 INFO] Validation perplexity: 1.81916
  266. [2019-02-04 18:07:44,107 INFO] Validation accuracy: 81.297
  267. [2019-02-04 18:07:44,109 INFO] Saving checkpoint models/ted1_step_10000.pt
  268. [2019-02-04 18:07:54,789 INFO] Step 10050/102583; acc: 78.44; ppl: 2.05; xent: 0.72; lr: 0.00010; 830/1027 tok/s; 1044 sec
  269. [2019-02-04 18:07:59,516 INFO] Step 10100/102583; acc: 80.19; ppl: 1.96; xent: 0.67; lr: 0.00010; 2184/2708 tok/s; 1049 sec
  270. [2019-02-04 18:08:04,625 INFO] Step 10150/102583; acc: 77.24; ppl: 2.08; xent: 0.73; lr: 0.00010; 1972/2586 tok/s; 1054 sec
  271. [2019-02-04 18:08:09,559 INFO] Step 10200/102583; acc: 78.14; ppl: 2.09; xent: 0.74; lr: 0.00010; 1931/2627 tok/s; 1059 sec
  272. [2019-02-04 18:08:14,492 INFO] Step 10250/102583; acc: 81.04; ppl: 1.86; xent: 0.62; lr: 0.00010; 2214/2681 tok/s; 1064 sec
  273. [2019-02-04 18:08:19,598 INFO] Step 10300/102583; acc: 78.10; ppl: 2.09; xent: 0.74; lr: 0.00010; 1920/2546 tok/s; 1069 sec
  274. [2019-02-04 18:08:24,334 INFO] Step 10350/102583; acc: 79.07; ppl: 2.04; xent: 0.71; lr: 0.00010; 2117/2562 tok/s; 1074 sec
  275. [2019-02-04 18:08:29,361 INFO] Step 10400/102583; acc: 78.63; ppl: 2.02; xent: 0.70; lr: 0.00010; 2024/2735 tok/s; 1079 sec
  276. [2019-02-04 18:08:34,355 INFO] Step 10450/102583; acc: 81.81; ppl: 1.87; xent: 0.63; lr: 0.00010; 2083/2617 tok/s; 1084 sec
  277. [2019-02-04 18:08:39,439 INFO] Step 10500/102583; acc: 79.05; ppl: 2.01; xent: 0.70; lr: 0.00010; 2125/2604 tok/s; 1089 sec
  278. [2019-02-04 18:08:44,152 INFO] Step 10550/102583; acc: 80.09; ppl: 1.95; xent: 0.67; lr: 0.00010; 2156/2461 tok/s; 1094 sec
  279. [2019-02-04 18:08:49,264 INFO] Step 10600/102583; acc: 79.61; ppl: 1.96; xent: 0.67; lr: 0.00010; 1859/2722 tok/s; 1099 sec
  280. [2019-02-04 18:08:54,321 INFO] Step 10650/102583; acc: 80.22; ppl: 1.96; xent: 0.67; lr: 0.00010; 2046/2585 tok/s; 1104 sec
  281. [2019-02-04 18:08:59,575 INFO] Step 10700/102583; acc: 78.86; ppl: 2.01; xent: 0.70; lr: 0.00010; 2014/2476 tok/s; 1109 sec
  282. [2019-02-04 18:09:04,614 INFO] Step 10750/102583; acc: 77.63; ppl: 2.07; xent: 0.73; lr: 0.00010; 1897/2688 tok/s; 1114 sec
  283. [2019-02-04 18:09:09,719 INFO] Step 10800/102583; acc: 81.54; ppl: 1.87; xent: 0.62; lr: 0.00010; 1963/2620 tok/s; 1119 sec
  284. [2019-02-04 18:09:14,762 INFO] Step 10850/102583; acc: 79.41; ppl: 2.00; xent: 0.69; lr: 0.00010; 2215/2503 tok/s; 1124 sec
  285. [2019-02-04 18:09:19,394 INFO] Step 10900/102583; acc: 78.49; ppl: 2.05; xent: 0.72; lr: 0.00010; 2090/2617 tok/s; 1129 sec
  286. [2019-02-04 18:09:24,416 INFO] Step 10950/102583; acc: 79.05; ppl: 2.01; xent: 0.70; lr: 0.00010; 1997/2692 tok/s; 1134 sec
  287. [2019-02-04 18:09:29,525 INFO] Step 11000/102583; acc: 80.17; ppl: 1.95; xent: 0.67; lr: 0.00010; 2085/2638 tok/s; 1139 sec
  288. [2019-02-04 18:09:34,525 INFO] Step 11050/102583; acc: 80.87; ppl: 1.91; xent: 0.65; lr: 0.00010; 2158/2369 tok/s; 1144 sec
  289. [2019-02-04 18:09:39,601 INFO] Step 11100/102583; acc: 78.55; ppl: 2.02; xent: 0.70; lr: 0.00010; 1892/2719 tok/s; 1149 sec
  290. [2019-02-04 18:09:44,485 INFO] Step 11150/102583; acc: 79.00; ppl: 2.00; xent: 0.69; lr: 0.00010; 1827/2731 tok/s; 1154 sec
  291. [2019-02-04 18:09:49,479 INFO] Step 11200/102583; acc: 80.35; ppl: 1.91; xent: 0.65; lr: 0.00010; 2171/2577 tok/s; 1159 sec
  292. [2019-02-04 18:09:54,306 INFO] Step 11250/102583; acc: 79.64; ppl: 1.98; xent: 0.69; lr: 0.00010; 1973/2583 tok/s; 1164 sec
  293. [2019-02-04 18:09:59,388 INFO] Step 11300/102583; acc: 79.87; ppl: 1.97; xent: 0.68; lr: 0.00010; 2045/2538 tok/s; 1169 sec
  294. [2019-02-04 18:10:04,494 INFO] Step 11350/102583; acc: 78.81; ppl: 1.98; xent: 0.68; lr: 0.00010; 2029/2666 tok/s; 1174 sec
  295. [2019-02-04 18:10:09,306 INFO] Step 11400/102583; acc: 80.04; ppl: 1.94; xent: 0.66; lr: 0.00010; 2130/2660 tok/s; 1179 sec
  296. [2019-02-04 18:10:14,374 INFO] Step 11450/102583; acc: 79.93; ppl: 1.95; xent: 0.67; lr: 0.00010; 1967/2676 tok/s; 1184 sec
  297. [2019-02-04 18:10:19,284 INFO] Step 11500/102583; acc: 79.86; ppl: 1.96; xent: 0.68; lr: 0.00010; 2025/2624 tok/s; 1189 sec
  298. [2019-02-04 18:10:24,250 INFO] Step 11550/102583; acc: 78.77; ppl: 2.01; xent: 0.70; lr: 0.00010; 1967/2522 tok/s; 1194 sec
  299. [2019-02-04 18:10:29,207 INFO] Step 11600/102583; acc: 79.75; ppl: 1.94; xent: 0.66; lr: 0.00010; 2064/2695 tok/s; 1199 sec
  300. [2019-02-04 18:10:34,318 INFO] Step 11650/102583; acc: 79.29; ppl: 1.98; xent: 0.68; lr: 0.00010; 2073/2537 tok/s; 1204 sec
  301. [2019-02-04 18:10:39,496 INFO] Step 11700/102583; acc: 80.60; ppl: 1.93; xent: 0.66; lr: 0.00010; 2153/2474 tok/s; 1209 sec
  302. [2019-02-04 18:10:44,082 INFO] Step 11750/102583; acc: 79.23; ppl: 2.02; xent: 0.70; lr: 0.00010; 1826/2761 tok/s; 1213 sec
  303. [2019-02-04 18:10:49,028 INFO] Step 11800/102583; acc: 78.62; ppl: 2.01; xent: 0.70; lr: 0.00010; 2077/2597 tok/s; 1218 sec
  304. [2019-02-04 18:10:54,094 INFO] Step 11850/102583; acc: 79.93; ppl: 1.98; xent: 0.68; lr: 0.00010; 2015/2548 tok/s; 1223 sec
  305. [2019-02-04 18:10:58,906 INFO] Step 11900/102583; acc: 81.31; ppl: 1.86; xent: 0.62; lr: 0.00010; 2020/2676 tok/s; 1228 sec
  306. [2019-02-04 18:11:03,976 INFO] Step 11950/102583; acc: 78.85; ppl: 2.01; xent: 0.70; lr: 0.00010; 2193/2484 tok/s; 1233 sec
  307. [2019-02-04 18:11:08,967 INFO] Step 12000/102583; acc: 80.82; ppl: 1.87; xent: 0.63; lr: 0.00010; 2142/2579 tok/s; 1238 sec
  308. [2019-02-04 18:11:14,072 INFO] Step 12050/102583; acc: 79.48; ppl: 1.95; xent: 0.67; lr: 0.00010; 1797/2799 tok/s; 1243 sec
  309. [2019-02-04 18:11:19,144 INFO] Step 12100/102583; acc: 79.86; ppl: 1.95; xent: 0.67; lr: 0.00010; 2024/2603 tok/s; 1249 sec
  310. [2019-02-04 18:11:23,859 INFO] Step 12150/102583; acc: 80.27; ppl: 1.94; xent: 0.66; lr: 0.00010; 2175/2556 tok/s; 1253 sec
  311. [2019-02-04 18:11:28,770 INFO] Step 12200/102583; acc: 80.86; ppl: 1.91; xent: 0.65; lr: 0.00010; 2010/2633 tok/s; 1258 sec
  312. [2019-02-04 18:11:33,887 INFO] Step 12250/102583; acc: 78.54; ppl: 1.99; xent: 0.69; lr: 0.00010; 2078/2628 tok/s; 1263 sec
  313. [2019-02-04 18:11:38,602 INFO] Step 12300/102583; acc: 78.76; ppl: 2.03; xent: 0.71; lr: 0.00010; 1985/2652 tok/s; 1268 sec
  314. [2019-02-04 18:11:43,666 INFO] Step 12350/102583; acc: 81.71; ppl: 1.87; xent: 0.62; lr: 0.00010; 2089/2634 tok/s; 1273 sec
  315. [2019-02-04 18:11:48,742 INFO] Step 12400/102583; acc: 81.06; ppl: 1.88; xent: 0.63; lr: 0.00010; 2107/2557 tok/s; 1278 sec
  316. [2019-02-04 18:11:53,821 INFO] Step 12450/102583; acc: 78.90; ppl: 2.02; xent: 0.70; lr: 0.00010; 1768/2831 tok/s; 1283 sec
  317. [2019-02-04 18:11:58,701 INFO] Step 12500/102583; acc: 78.60; ppl: 2.08; xent: 0.73; lr: 0.00010; 2000/2456 tok/s; 1288 sec
  318. [2019-02-04 18:12:03,577 INFO] Step 12550/102583; acc: 81.55; ppl: 1.82; xent: 0.60; lr: 0.00010; 2180/2706 tok/s; 1293 sec
  319. [2019-02-04 18:12:08,603 INFO] Step 12600/102583; acc: 79.71; ppl: 1.95; xent: 0.67; lr: 0.00010; 1976/2510 tok/s; 1298 sec
  320. [2019-02-04 18:12:13,788 INFO] Step 12650/102583; acc: 80.65; ppl: 1.92; xent: 0.65; lr: 0.00010; 2103/2701 tok/s; 1303 sec
  321. [2019-02-04 18:12:18,843 INFO] Step 12700/102583; acc: 78.94; ppl: 2.01; xent: 0.70; lr: 0.00010; 2113/2550 tok/s; 1308 sec
  322. [2019-02-04 18:12:23,832 INFO] Step 12750/102583; acc: 80.62; ppl: 1.90; xent: 0.64; lr: 0.00010; 2028/2581 tok/s; 1313 sec
  323. [2019-02-04 18:12:28,677 INFO] Step 12800/102583; acc: 78.69; ppl: 2.04; xent: 0.71; lr: 0.00010; 2114/2550 tok/s; 1318 sec
  324. [2019-02-04 18:12:33,515 INFO] Step 12850/102583; acc: 80.16; ppl: 1.92; xent: 0.65; lr: 0.00010; 1998/2643 tok/s; 1323 sec
  325. [2019-02-04 18:12:38,432 INFO] Step 12900/102583; acc: 80.00; ppl: 1.94; xent: 0.66; lr: 0.00010; 1964/2727 tok/s; 1328 sec
  326. [2019-02-04 18:12:43,712 INFO] Step 12950/102583; acc: 79.73; ppl: 1.93; xent: 0.66; lr: 0.00010; 2003/2572 tok/s; 1333 sec
  327. [2019-02-04 18:12:52,119 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  328. [2019-02-04 18:12:55,158 INFO] Step 13000/102583; acc: 80.04; ppl: 1.95; xent: 0.67; lr: 0.00010; 889/1130 tok/s; 1345 sec
  329. [2019-02-04 18:13:00,091 INFO] Step 13050/102583; acc: 80.08; ppl: 1.99; xent: 0.69; lr: 0.00010; 2168/2422 tok/s; 1349 sec
  330. [2019-02-04 18:13:04,937 INFO] Step 13100/102583; acc: 80.39; ppl: 1.90; xent: 0.64; lr: 0.00010; 1938/2801 tok/s; 1354 sec
  331. [2019-02-04 18:13:10,138 INFO] Step 13150/102583; acc: 81.24; ppl: 1.87; xent: 0.63; lr: 0.00010; 1990/2636 tok/s; 1360 sec
  332. [2019-02-04 18:13:14,973 INFO] Step 13200/102583; acc: 79.78; ppl: 1.98; xent: 0.68; lr: 0.00010; 2153/2741 tok/s; 1364 sec
  333. [2019-02-04 18:13:20,073 INFO] Step 13250/102583; acc: 80.30; ppl: 1.88; xent: 0.63; lr: 0.00010; 2019/2657 tok/s; 1369 sec
  334. [2019-02-04 18:13:24,939 INFO] Step 13300/102583; acc: 79.26; ppl: 1.97; xent: 0.68; lr: 0.00010; 2175/2501 tok/s; 1374 sec
  335. [2019-02-04 18:13:29,652 INFO] Step 13350/102583; acc: 80.61; ppl: 1.92; xent: 0.65; lr: 0.00010; 2096/2729 tok/s; 1379 sec
  336. [2019-02-04 18:13:34,884 INFO] Step 13400/102583; acc: 78.31; ppl: 2.02; xent: 0.70; lr: 0.00010; 1944/2586 tok/s; 1384 sec
  337. [2019-02-04 18:13:39,839 INFO] Step 13450/102583; acc: 79.79; ppl: 1.98; xent: 0.68; lr: 0.00010; 1999/2624 tok/s; 1389 sec
  338. [2019-02-04 18:13:44,677 INFO] Step 13500/102583; acc: 81.27; ppl: 1.81; xent: 0.60; lr: 0.00010; 2160/2744 tok/s; 1394 sec
  339. [2019-02-04 18:13:49,906 INFO] Step 13550/102583; acc: 78.08; ppl: 2.07; xent: 0.73; lr: 0.00010; 1947/2486 tok/s; 1399 sec
  340. [2019-02-04 18:13:54,612 INFO] Step 13600/102583; acc: 80.42; ppl: 1.95; xent: 0.67; lr: 0.00010; 2180/2562 tok/s; 1404 sec
  341. [2019-02-04 18:13:59,597 INFO] Step 13650/102583; acc: 79.25; ppl: 1.95; xent: 0.67; lr: 0.00010; 1939/2746 tok/s; 1409 sec
  342. [2019-02-04 18:14:04,683 INFO] Step 13700/102583; acc: 81.73; ppl: 1.85; xent: 0.62; lr: 0.00010; 2171/2555 tok/s; 1414 sec
  343. [2019-02-04 18:14:09,677 INFO] Step 13750/102583; acc: 79.20; ppl: 1.98; xent: 0.68; lr: 0.00010; 2038/2663 tok/s; 1419 sec
  344. [2019-02-04 18:14:14,381 INFO] Step 13800/102583; acc: 80.81; ppl: 1.92; xent: 0.65; lr: 0.00010; 2208/2481 tok/s; 1424 sec
  345. [2019-02-04 18:14:19,499 INFO] Step 13850/102583; acc: 80.34; ppl: 1.86; xent: 0.62; lr: 0.00010; 1853/2760 tok/s; 1429 sec
  346. [2019-02-04 18:14:24,679 INFO] Step 13900/102583; acc: 79.71; ppl: 1.96; xent: 0.67; lr: 0.00010; 2029/2461 tok/s; 1434 sec
  347. [2019-02-04 18:14:29,733 INFO] Step 13950/102583; acc: 80.76; ppl: 1.90; xent: 0.64; lr: 0.00010; 2099/2578 tok/s; 1439 sec
  348. [2019-02-04 18:14:34,773 INFO] Step 14000/102583; acc: 77.56; ppl: 2.07; xent: 0.73; lr: 0.00010; 1820/2704 tok/s; 1444 sec
  349. [2019-02-04 18:14:39,900 INFO] Step 14050/102583; acc: 82.89; ppl: 1.79; xent: 0.58; lr: 0.00010; 2017/2608 tok/s; 1449 sec
  350. [2019-02-04 18:14:44,831 INFO] Step 14100/102583; acc: 78.72; ppl: 2.01; xent: 0.70; lr: 0.00010; 2123/2495 tok/s; 1454 sec
  351. [2019-02-04 18:14:49,436 INFO] Step 14150/102583; acc: 78.63; ppl: 2.02; xent: 0.70; lr: 0.00010; 2172/2612 tok/s; 1459 sec
  352. [2019-02-04 18:14:54,521 INFO] Step 14200/102583; acc: 79.74; ppl: 1.94; xent: 0.66; lr: 0.00010; 1985/2732 tok/s; 1464 sec
  353. [2019-02-04 18:14:59,662 INFO] Step 14250/102583; acc: 81.50; ppl: 1.89; xent: 0.64; lr: 0.00010; 2060/2659 tok/s; 1469 sec
  354. [2019-02-04 18:15:04,845 INFO] Step 14300/102583; acc: 81.28; ppl: 1.86; xent: 0.62; lr: 0.00010; 2158/2321 tok/s; 1474 sec
  355. [2019-02-04 18:15:09,761 INFO] Step 14350/102583; acc: 78.44; ppl: 1.99; xent: 0.69; lr: 0.00010; 1846/2819 tok/s; 1479 sec
  356. [2019-02-04 18:15:14,618 INFO] Step 14400/102583; acc: 79.09; ppl: 1.97; xent: 0.68; lr: 0.00010; 1850/2706 tok/s; 1484 sec
  357. [2019-02-04 18:15:19,653 INFO] Step 14450/102583; acc: 82.39; ppl: 1.81; xent: 0.59; lr: 0.00010; 2253/2503 tok/s; 1489 sec
  358. [2019-02-04 18:15:24,400 INFO] Step 14500/102583; acc: 79.92; ppl: 1.98; xent: 0.68; lr: 0.00010; 1943/2638 tok/s; 1494 sec
  359. [2019-02-04 18:15:29,517 INFO] Step 14550/102583; acc: 79.75; ppl: 1.93; xent: 0.66; lr: 0.00010; 2074/2503 tok/s; 1499 sec
  360. [2019-02-04 18:15:34,663 INFO] Step 14600/102583; acc: 80.37; ppl: 1.90; xent: 0.64; lr: 0.00010; 2045/2746 tok/s; 1504 sec
  361. [2019-02-04 18:15:39,415 INFO] Step 14650/102583; acc: 79.16; ppl: 1.98; xent: 0.68; lr: 0.00010; 2001/2586 tok/s; 1509 sec
  362. [2019-02-04 18:15:44,449 INFO] Step 14700/102583; acc: 80.43; ppl: 1.90; xent: 0.64; lr: 0.00010; 1988/2734 tok/s; 1514 sec
  363. [2019-02-04 18:15:49,500 INFO] Step 14750/102583; acc: 81.52; ppl: 1.86; xent: 0.62; lr: 0.00010; 2108/2586 tok/s; 1519 sec
  364. [2019-02-04 18:15:54,491 INFO] Step 14800/102583; acc: 78.64; ppl: 2.02; xent: 0.70; lr: 0.00010; 1891/2505 tok/s; 1524 sec
  365. [2019-02-04 18:15:59,275 INFO] Step 14850/102583; acc: 80.66; ppl: 1.88; xent: 0.63; lr: 0.00010; 2167/2744 tok/s; 1529 sec
  366. [2019-02-04 18:16:04,334 INFO] Step 14900/102583; acc: 79.52; ppl: 1.97; xent: 0.68; lr: 0.00010; 1981/2516 tok/s; 1534 sec
  367. [2019-02-04 18:16:09,474 INFO] Step 14950/102583; acc: 81.86; ppl: 1.85; xent: 0.62; lr: 0.00010; 2206/2485 tok/s; 1539 sec
  368. [2019-02-04 18:16:14,203 INFO] Step 15000/102583; acc: 78.91; ppl: 2.03; xent: 0.71; lr: 0.00010; 1885/2730 tok/s; 1544 sec
  369. [2019-02-04 18:16:14,205 INFO] Saving checkpoint models/ted1_step_15000.pt
  370. [2019-02-04 18:16:25,284 INFO] Step 15050/102583; acc: 79.62; ppl: 1.94; xent: 0.66; lr: 0.00010; 924/1203 tok/s; 1555 sec
  371. [2019-02-04 18:16:30,152 INFO] Step 15100/102583; acc: 80.14; ppl: 1.95; xent: 0.67; lr: 0.00010; 2030/2573 tok/s; 1560 sec
  372. [2019-02-04 18:16:35,036 INFO] Step 15150/102583; acc: 81.99; ppl: 1.81; xent: 0.59; lr: 0.00010; 2059/2625 tok/s; 1564 sec
  373. [2019-02-04 18:16:40,025 INFO] Step 15200/102583; acc: 79.30; ppl: 2.00; xent: 0.69; lr: 0.00010; 2156/2504 tok/s; 1569 sec
  374. [2019-02-04 18:16:44,956 INFO] Step 15250/102583; acc: 81.92; ppl: 1.81; xent: 0.59; lr: 0.00010; 2173/2590 tok/s; 1574 sec
  375. [2019-02-04 18:16:50,049 INFO] Step 15300/102583; acc: 79.41; ppl: 1.95; xent: 0.67; lr: 0.00010; 1789/2823 tok/s; 1579 sec
  376. [2019-02-04 18:16:55,179 INFO] Step 15350/102583; acc: 80.66; ppl: 1.90; xent: 0.64; lr: 0.00010; 2064/2559 tok/s; 1585 sec
  377. [2019-02-04 18:16:59,878 INFO] Step 15400/102583; acc: 81.69; ppl: 1.85; xent: 0.61; lr: 0.00010; 2118/2600 tok/s; 1589 sec
  378. [2019-02-04 18:17:05,003 INFO] Step 15450/102583; acc: 80.81; ppl: 1.88; xent: 0.63; lr: 0.00010; 2035/2633 tok/s; 1594 sec
  379. [2019-02-04 18:17:09,834 INFO] Step 15500/102583; acc: 78.04; ppl: 2.02; xent: 0.70; lr: 0.00010; 2070/2633 tok/s; 1599 sec
  380. [2019-02-04 18:17:14,739 INFO] Step 15550/102583; acc: 79.99; ppl: 1.97; xent: 0.68; lr: 0.00010; 1906/2663 tok/s; 1604 sec
  381. [2019-02-04 18:17:19,569 INFO] Step 15600/102583; acc: 82.39; ppl: 1.82; xent: 0.60; lr: 0.00010; 2196/2622 tok/s; 1609 sec
  382. [2019-02-04 18:17:24,867 INFO] Step 15650/102583; acc: 80.32; ppl: 1.87; xent: 0.63; lr: 0.00010; 2036/2567 tok/s; 1614 sec
  383. [2019-02-04 18:17:30,088 INFO] Step 15700/102583; acc: 79.10; ppl: 2.01; xent: 0.70; lr: 0.00010; 1827/2745 tok/s; 1619 sec
  384. [2019-02-04 18:17:34,602 INFO] Step 15750/102583; acc: 78.83; ppl: 2.02; xent: 0.70; lr: 0.00010; 2074/2532 tok/s; 1624 sec
  385. [2019-02-04 18:17:39,701 INFO] Step 15800/102583; acc: 81.83; ppl: 1.80; xent: 0.59; lr: 0.00010; 2073/2751 tok/s; 1629 sec
  386. [2019-02-04 18:17:44,716 INFO] Step 15850/102583; acc: 81.16; ppl: 1.89; xent: 0.64; lr: 0.00010; 1991/2493 tok/s; 1634 sec
  387. [2019-02-04 18:17:49,708 INFO] Step 15900/102583; acc: 79.79; ppl: 1.92; xent: 0.65; lr: 0.00010; 2148/2704 tok/s; 1639 sec
  388. [2019-02-04 18:17:54,750 INFO] Step 15950/102583; acc: 80.60; ppl: 1.91; xent: 0.65; lr: 0.00010; 2131/2558 tok/s; 1644 sec
  389. [2019-02-04 18:18:00,059 INFO] Step 16000/102583; acc: 80.75; ppl: 1.89; xent: 0.64; lr: 0.00010; 2053/2509 tok/s; 1649 sec
  390. [2019-02-04 18:18:04,573 INFO] Step 16050/102583; acc: 79.19; ppl: 2.00; xent: 0.69; lr: 0.00010; 2039/2645 tok/s; 1654 sec
  391. [2019-02-04 18:18:09,374 INFO] Step 16100/102583; acc: 80.73; ppl: 1.86; xent: 0.62; lr: 0.00010; 2093/2603 tok/s; 1659 sec
  392. [2019-02-04 18:18:14,414 INFO] Step 16150/102583; acc: 80.85; ppl: 1.90; xent: 0.64; lr: 0.00010; 1964/2662 tok/s; 1664 sec
  393. [2019-02-04 18:18:19,712 INFO] Step 16200/102583; acc: 79.50; ppl: 1.92; xent: 0.65; lr: 0.00010; 1907/2569 tok/s; 1669 sec
  394. [2019-02-04 18:18:27,016 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  395. [2019-02-04 18:18:30,858 INFO] Step 16250/102583; acc: 81.18; ppl: 1.89; xent: 0.64; lr: 0.00010; 969/1197 tok/s; 1680 sec
  396. [2019-02-04 18:18:35,462 INFO] Step 16300/102583; acc: 79.66; ppl: 2.02; xent: 0.70; lr: 0.00010; 2141/2463 tok/s; 1685 sec
  397. [2019-02-04 18:18:40,295 INFO] Step 16350/102583; acc: 81.39; ppl: 1.85; xent: 0.61; lr: 0.00010; 1965/2797 tok/s; 1690 sec
  398. [2019-02-04 18:18:45,598 INFO] Step 16400/102583; acc: 81.15; ppl: 1.87; xent: 0.63; lr: 0.00010; 2017/2613 tok/s; 1695 sec
  399. [2019-02-04 18:18:50,389 INFO] Step 16450/102583; acc: 81.13; ppl: 1.89; xent: 0.64; lr: 0.00010; 2102/2820 tok/s; 1700 sec
  400. [2019-02-04 18:18:55,377 INFO] Step 16500/102583; acc: 79.87; ppl: 1.91; xent: 0.65; lr: 0.00010; 2001/2597 tok/s; 1705 sec
  401. [2019-02-04 18:19:00,428 INFO] Step 16550/102583; acc: 80.56; ppl: 1.88; xent: 0.63; lr: 0.00010; 2218/2521 tok/s; 1710 sec
  402. [2019-02-04 18:19:05,034 INFO] Step 16600/102583; acc: 79.51; ppl: 1.97; xent: 0.68; lr: 0.00010; 2065/2757 tok/s; 1714 sec
  403. [2019-02-04 18:19:10,281 INFO] Step 16650/102583; acc: 79.70; ppl: 1.94; xent: 0.66; lr: 0.00010; 1902/2602 tok/s; 1720 sec
  404. [2019-02-04 18:19:15,341 INFO] Step 16700/102583; acc: 79.99; ppl: 1.95; xent: 0.67; lr: 0.00010; 2158/2563 tok/s; 1725 sec
  405. [2019-02-04 18:19:20,190 INFO] Step 16750/102583; acc: 81.48; ppl: 1.80; xent: 0.59; lr: 0.00010; 2029/2802 tok/s; 1730 sec
  406. [2019-02-04 18:19:25,237 INFO] Step 16800/102583; acc: 78.12; ppl: 2.08; xent: 0.73; lr: 0.00010; 2039/2430 tok/s; 1735 sec
  407. [2019-02-04 18:19:30,181 INFO] Step 16850/102583; acc: 81.89; ppl: 1.86; xent: 0.62; lr: 0.00010; 2066/2628 tok/s; 1740 sec
  408. [2019-02-04 18:19:35,019 INFO] Step 16900/102583; acc: 79.79; ppl: 1.93; xent: 0.66; lr: 0.00010; 1994/2758 tok/s; 1744 sec
  409. [2019-02-04 18:19:39,989 INFO] Step 16950/102583; acc: 82.10; ppl: 1.83; xent: 0.60; lr: 0.00010; 2208/2509 tok/s; 1749 sec
  410. [2019-02-04 18:19:45,236 INFO] Step 17000/102583; acc: 79.32; ppl: 1.95; xent: 0.67; lr: 0.00010; 1956/2689 tok/s; 1755 sec
  411. [2019-02-04 18:19:49,747 INFO] Step 17050/102583; acc: 80.81; ppl: 1.93; xent: 0.66; lr: 0.00010; 2213/2483 tok/s; 1759 sec
  412. [2019-02-04 18:19:54,753 INFO] Step 17100/102583; acc: 81.62; ppl: 1.82; xent: 0.60; lr: 0.00010; 1882/2728 tok/s; 1764 sec
  413. [2019-02-04 18:20:00,089 INFO] Step 17150/102583; acc: 79.95; ppl: 1.93; xent: 0.66; lr: 0.00010; 1990/2473 tok/s; 1769 sec
  414. [2019-02-04 18:20:05,100 INFO] Step 17200/102583; acc: 80.46; ppl: 1.89; xent: 0.64; lr: 0.00010; 2161/2588 tok/s; 1774 sec
  415. [2019-02-04 18:20:10,444 INFO] Step 17250/102583; acc: 79.57; ppl: 1.94; xent: 0.66; lr: 0.00010; 1853/2689 tok/s; 1780 sec
  416. [2019-02-04 18:20:15,234 INFO] Step 17300/102583; acc: 81.48; ppl: 1.86; xent: 0.62; lr: 0.00010; 2047/2630 tok/s; 1785 sec
  417. [2019-02-04 18:20:20,272 INFO] Step 17350/102583; acc: 78.42; ppl: 2.00; xent: 0.69; lr: 0.00010; 2041/2575 tok/s; 1790 sec
  418. [2019-02-04 18:20:24,920 INFO] Step 17400/102583; acc: 80.70; ppl: 1.92; xent: 0.65; lr: 0.00010; 2242/2531 tok/s; 1794 sec
  419. [2019-02-04 18:20:29,887 INFO] Step 17450/102583; acc: 79.03; ppl: 1.97; xent: 0.68; lr: 0.00010; 1960/2755 tok/s; 1799 sec
  420. [2019-02-04 18:20:35,007 INFO] Step 17500/102583; acc: 82.11; ppl: 1.83; xent: 0.60; lr: 0.00010; 2056/2654 tok/s; 1804 sec
  421. [2019-02-04 18:20:40,300 INFO] Step 17550/102583; acc: 81.15; ppl: 1.87; xent: 0.62; lr: 0.00010; 2080/2311 tok/s; 1810 sec
  422. [2019-02-04 18:20:45,130 INFO] Step 17600/102583; acc: 79.41; ppl: 1.92; xent: 0.65; lr: 0.00010; 1942/2851 tok/s; 1815 sec
  423. [2019-02-04 18:20:49,875 INFO] Step 17650/102583; acc: 79.42; ppl: 1.95; xent: 0.67; lr: 0.00010; 1867/2675 tok/s; 1819 sec
  424. [2019-02-04 18:20:54,929 INFO] Step 17700/102583; acc: 82.26; ppl: 1.79; xent: 0.58; lr: 0.00010; 2243/2522 tok/s; 1824 sec
  425. [2019-02-04 18:20:59,748 INFO] Step 17750/102583; acc: 78.59; ppl: 2.02; xent: 0.71; lr: 0.00010; 1833/2676 tok/s; 1829 sec
  426. [2019-02-04 18:21:04,960 INFO] Step 17800/102583; acc: 81.61; ppl: 1.81; xent: 0.59; lr: 0.00010; 2159/2533 tok/s; 1834 sec
  427. [2019-02-04 18:21:09,967 INFO] Step 17850/102583; acc: 80.07; ppl: 1.93; xent: 0.66; lr: 0.00010; 2026/2706 tok/s; 1839 sec
  428. [2019-02-04 18:21:14,834 INFO] Step 17900/102583; acc: 80.24; ppl: 1.91; xent: 0.64; lr: 0.00010; 2047/2594 tok/s; 1844 sec
  429. [2019-02-04 18:21:19,703 INFO] Step 17950/102583; acc: 80.63; ppl: 1.90; xent: 0.64; lr: 0.00010; 1962/2701 tok/s; 1849 sec
  430. [2019-02-04 18:21:24,771 INFO] Step 18000/102583; acc: 81.97; ppl: 1.83; xent: 0.60; lr: 0.00010; 2202/2566 tok/s; 1854 sec
  431. [2019-02-04 18:21:29,751 INFO] Step 18050/102583; acc: 77.89; ppl: 2.07; xent: 0.73; lr: 0.00010; 1818/2528 tok/s; 1859 sec
  432. [2019-02-04 18:21:34,672 INFO] Step 18100/102583; acc: 81.71; ppl: 1.82; xent: 0.60; lr: 0.00010; 2239/2649 tok/s; 1864 sec
  433. [2019-02-04 18:21:39,587 INFO] Step 18150/102583; acc: 79.36; ppl: 1.97; xent: 0.68; lr: 0.00010; 1892/2609 tok/s; 1869 sec
  434. [2019-02-04 18:21:44,950 INFO] Step 18200/102583; acc: 82.31; ppl: 1.81; xent: 0.59; lr: 0.00010; 2174/2494 tok/s; 1874 sec
  435. [2019-02-04 18:21:49,819 INFO] Step 18250/102583; acc: 79.51; ppl: 1.99; xent: 0.69; lr: 0.00010; 1930/2696 tok/s; 1879 sec
  436. [2019-02-04 18:21:54,959 INFO] Step 18300/102583; acc: 81.24; ppl: 1.84; xent: 0.61; lr: 0.00010; 2003/2660 tok/s; 1884 sec
  437. [2019-02-04 18:21:59,686 INFO] Step 18350/102583; acc: 79.55; ppl: 2.01; xent: 0.70; lr: 0.00010; 2054/2603 tok/s; 1889 sec
  438. [2019-02-04 18:22:04,840 INFO] Step 18400/102583; acc: 82.62; ppl: 1.77; xent: 0.57; lr: 0.00010; 1999/2545 tok/s; 1894 sec
  439. [2019-02-04 18:22:09,539 INFO] Step 18450/102583; acc: 79.15; ppl: 1.99; xent: 0.69; lr: 0.00010; 2209/2579 tok/s; 1899 sec
  440. [2019-02-04 18:22:14,640 INFO] Step 18500/102583; acc: 81.77; ppl: 1.81; xent: 0.59; lr: 0.00010; 2105/2585 tok/s; 1904 sec
  441. [2019-02-04 18:22:19,801 INFO] Step 18550/102583; acc: 80.53; ppl: 1.89; xent: 0.63; lr: 0.00010; 1873/2783 tok/s; 1909 sec
  442. [2019-02-04 18:22:24,742 INFO] Step 18600/102583; acc: 80.60; ppl: 1.91; xent: 0.65; lr: 0.00010; 2085/2590 tok/s; 1914 sec
  443. [2019-02-04 18:22:29,491 INFO] Step 18650/102583; acc: 82.46; ppl: 1.80; xent: 0.59; lr: 0.00010; 2157/2567 tok/s; 1919 sec
  444. [2019-02-04 18:22:34,555 INFO] Step 18700/102583; acc: 80.07; ppl: 1.91; xent: 0.65; lr: 0.00010; 1977/2675 tok/s; 1924 sec
  445. [2019-02-04 18:22:39,314 INFO] Step 18750/102583; acc: 78.72; ppl: 1.97; xent: 0.68; lr: 0.00010; 2087/2633 tok/s; 1929 sec
  446. [2019-02-04 18:22:44,362 INFO] Step 18800/102583; acc: 81.24; ppl: 1.87; xent: 0.63; lr: 0.00010; 1968/2651 tok/s; 1934 sec
  447. [2019-02-04 18:22:49,042 INFO] Step 18850/102583; acc: 81.55; ppl: 1.86; xent: 0.62; lr: 0.00010; 2137/2648 tok/s; 1938 sec
  448. [2019-02-04 18:22:54,363 INFO] Step 18900/102583; acc: 81.08; ppl: 1.86; xent: 0.62; lr: 0.00010; 2028/2574 tok/s; 1944 sec
  449. [2019-02-04 18:22:59,528 INFO] Step 18950/102583; acc: 79.29; ppl: 2.00; xent: 0.69; lr: 0.00010; 1872/2696 tok/s; 1949 sec
  450. [2019-02-04 18:23:04,343 INFO] Step 19000/102583; acc: 80.22; ppl: 1.94; xent: 0.66; lr: 0.00010; 2102/2517 tok/s; 1954 sec
  451. [2019-02-04 18:23:09,123 INFO] Step 19050/102583; acc: 81.98; ppl: 1.79; xent: 0.58; lr: 0.00010; 1986/2788 tok/s; 1959 sec
  452. [2019-02-04 18:23:14,426 INFO] Step 19100/102583; acc: 81.32; ppl: 1.86; xent: 0.62; lr: 0.00010; 2043/2521 tok/s; 1964 sec
  453. [2019-02-04 18:23:19,215 INFO] Step 19150/102583; acc: 80.16; ppl: 1.90; xent: 0.64; lr: 0.00010; 2074/2734 tok/s; 1969 sec
  454. [2019-02-04 18:23:24,202 INFO] Step 19200/102583; acc: 80.41; ppl: 1.93; xent: 0.66; lr: 0.00010; 2167/2530 tok/s; 1974 sec
  455. [2019-02-04 18:23:29,658 INFO] Step 19250/102583; acc: 81.81; ppl: 1.82; xent: 0.60; lr: 0.00010; 2058/2514 tok/s; 1979 sec
  456. [2019-02-04 18:23:34,082 INFO] Step 19300/102583; acc: 79.45; ppl: 1.98; xent: 0.69; lr: 0.00010; 2064/2652 tok/s; 1983 sec
  457. [2019-02-04 18:23:38,859 INFO] Step 19350/102583; acc: 80.79; ppl: 1.86; xent: 0.62; lr: 0.00010; 2010/2624 tok/s; 1988 sec
  458. [2019-02-04 18:23:44,218 INFO] Step 19400/102583; acc: 81.01; ppl: 1.88; xent: 0.63; lr: 0.00010; 1992/2564 tok/s; 1994 sec
  459. [2019-02-04 18:23:49,353 INFO] Step 19450/102583; acc: 79.99; ppl: 1.90; xent: 0.64; lr: 0.00010; 1928/2619 tok/s; 1999 sec
  460. [2019-02-04 18:23:56,209 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  461. [2019-02-04 18:24:00,649 INFO] Step 19500/102583; acc: 81.60; ppl: 1.86; xent: 0.62; lr: 0.00010; 956/1154 tok/s; 2010 sec
  462. [2019-02-04 18:24:05,227 INFO] Step 19550/102583; acc: 79.63; ppl: 2.01; xent: 0.70; lr: 0.00010; 2085/2546 tok/s; 2015 sec
  463. [2019-02-04 18:24:10,236 INFO] Step 19600/102583; acc: 82.19; ppl: 1.78; xent: 0.58; lr: 0.00010; 2031/2796 tok/s; 2020 sec
  464. [2019-02-04 18:24:15,419 INFO] Step 19650/102583; acc: 80.31; ppl: 1.91; xent: 0.65; lr: 0.00010; 1960/2616 tok/s; 2025 sec
  465. [2019-02-04 18:24:20,141 INFO] Step 19700/102583; acc: 81.40; ppl: 1.86; xent: 0.62; lr: 0.00010; 2090/2813 tok/s; 2030 sec
  466. [2019-02-04 18:24:25,276 INFO] Step 19750/102583; acc: 80.39; ppl: 1.87; xent: 0.63; lr: 0.00010; 2112/2496 tok/s; 2035 sec
  467. [2019-02-04 18:24:30,189 INFO] Step 19800/102583; acc: 81.19; ppl: 1.85; xent: 0.61; lr: 0.00010; 2181/2631 tok/s; 2040 sec
  468. [2019-02-04 18:24:34,978 INFO] Step 19850/102583; acc: 79.81; ppl: 1.95; xent: 0.67; lr: 0.00010; 2111/2638 tok/s; 2044 sec
  469. [2019-02-04 18:24:40,130 INFO] Step 19900/102583; acc: 79.18; ppl: 1.97; xent: 0.68; lr: 0.00010; 1775/2654 tok/s; 2050 sec
  470. [2019-02-04 18:24:45,074 INFO] Step 19950/102583; acc: 80.89; ppl: 1.89; xent: 0.63; lr: 0.00010; 2194/2635 tok/s; 2054 sec
  471. [2019-02-04 18:24:50,211 INFO] Step 20000/102583; acc: 81.45; ppl: 1.82; xent: 0.60; lr: 0.00010; 1888/2749 tok/s; 2060 sec
  472. [2019-02-04 18:24:50,238 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  473. [2019-02-04 18:24:51,538 INFO] Validation perplexity: 1.76312
  474. [2019-02-04 18:24:51,539 INFO] Validation accuracy: 82.8277
  475. [2019-02-04 18:24:51,540 INFO] Saving checkpoint models/ted1_step_20000.pt
  476. [2019-02-04 18:25:02,098 INFO] Step 20050/102583; acc: 79.34; ppl: 2.00; xent: 0.69; lr: 0.00010; 915/979 tok/s; 2071 sec
  477. [2019-02-04 18:25:06,963 INFO] Step 20100/102583; acc: 81.28; ppl: 1.88; xent: 0.63; lr: 0.00010; 2047/2636 tok/s; 2076 sec
  478. [2019-02-04 18:25:11,902 INFO] Step 20150/102583; acc: 79.36; ppl: 1.95; xent: 0.67; lr: 0.00010; 1910/2684 tok/s; 2081 sec
  479. [2019-02-04 18:25:17,007 INFO] Step 20200/102583; acc: 82.93; ppl: 1.77; xent: 0.57; lr: 0.00010; 2277/2567 tok/s; 2086 sec
  480. [2019-02-04 18:25:22,037 INFO] Step 20250/102583; acc: 79.31; ppl: 1.95; xent: 0.67; lr: 0.00010; 1928/2730 tok/s; 2091 sec
  481. [2019-02-04 18:25:26,598 INFO] Step 20300/102583; acc: 81.26; ppl: 1.90; xent: 0.64; lr: 0.00010; 2132/2524 tok/s; 2096 sec
  482. [2019-02-04 18:25:31,740 INFO] Step 20350/102583; acc: 81.49; ppl: 1.80; xent: 0.59; lr: 0.00010; 1943/2683 tok/s; 2101 sec
  483. [2019-02-04 18:25:36,845 INFO] Step 20400/102583; acc: 80.53; ppl: 1.92; xent: 0.65; lr: 0.00010; 2006/2466 tok/s; 2106 sec
  484. [2019-02-04 18:25:41,904 INFO] Step 20450/102583; acc: 80.73; ppl: 1.88; xent: 0.63; lr: 0.00010; 2101/2609 tok/s; 2111 sec
  485. [2019-02-04 18:25:47,247 INFO] Step 20500/102583; acc: 80.08; ppl: 1.90; xent: 0.64; lr: 0.00010; 1889/2699 tok/s; 2117 sec
  486. [2019-02-04 18:25:51,958 INFO] Step 20550/102583; acc: 81.86; ppl: 1.84; xent: 0.61; lr: 0.00010; 2081/2581 tok/s; 2121 sec
  487. [2019-02-04 18:25:56,967 INFO] Step 20600/102583; acc: 79.00; ppl: 1.98; xent: 0.69; lr: 0.00010; 2141/2537 tok/s; 2126 sec
  488. [2019-02-04 18:26:01,824 INFO] Step 20650/102583; acc: 80.36; ppl: 1.91; xent: 0.65; lr: 0.00010; 2027/2715 tok/s; 2131 sec
  489. [2019-02-04 18:26:06,577 INFO] Step 20700/102583; acc: 79.34; ppl: 1.97; xent: 0.68; lr: 0.00010; 2062/2617 tok/s; 2136 sec
  490. [2019-02-04 18:26:11,898 INFO] Step 20750/102583; acc: 81.72; ppl: 1.83; xent: 0.61; lr: 0.00010; 1926/2584 tok/s; 2141 sec
  491. [2019-02-04 18:26:17,038 INFO] Step 20800/102583; acc: 82.16; ppl: 1.81; xent: 0.60; lr: 0.00010; 2208/2380 tok/s; 2146 sec
  492. [2019-02-04 18:26:21,816 INFO] Step 20850/102583; acc: 79.41; ppl: 1.92; xent: 0.65; lr: 0.00010; 1896/2875 tok/s; 2151 sec
  493. [2019-02-04 18:26:26,712 INFO] Step 20900/102583; acc: 79.85; ppl: 1.90; xent: 0.64; lr: 0.00010; 1979/2702 tok/s; 2156 sec
  494. [2019-02-04 18:26:31,787 INFO] Step 20950/102583; acc: 82.37; ppl: 1.81; xent: 0.60; lr: 0.00010; 2181/2444 tok/s; 2161 sec
  495. [2019-02-04 18:26:36,497 INFO] Step 21000/102583; acc: 79.73; ppl: 1.96; xent: 0.67; lr: 0.00010; 1832/2685 tok/s; 2166 sec
  496. [2019-02-04 18:26:41,712 INFO] Step 21050/102583; acc: 81.03; ppl: 1.81; xent: 0.59; lr: 0.00010; 2151/2601 tok/s; 2171 sec
  497. [2019-02-04 18:26:46,647 INFO] Step 21100/102583; acc: 80.19; ppl: 1.93; xent: 0.66; lr: 0.00010; 2056/2673 tok/s; 2176 sec
  498. [2019-02-04 18:26:51,528 INFO] Step 21150/102583; acc: 80.33; ppl: 1.90; xent: 0.64; lr: 0.00010; 2016/2590 tok/s; 2181 sec
  499. [2019-02-04 18:26:56,593 INFO] Step 21200/102583; acc: 81.52; ppl: 1.84; xent: 0.61; lr: 0.00010; 1876/2718 tok/s; 2186 sec
  500. [2019-02-04 18:27:01,503 INFO] Step 21250/102583; acc: 81.41; ppl: 1.87; xent: 0.62; lr: 0.00010; 2311/2507 tok/s; 2191 sec
  501. [2019-02-04 18:27:06,585 INFO] Step 21300/102583; acc: 78.99; ppl: 1.98; xent: 0.68; lr: 0.00010; 1810/2675 tok/s; 2196 sec
  502. [2019-02-04 18:27:11,438 INFO] Step 21350/102583; acc: 81.72; ppl: 1.84; xent: 0.61; lr: 0.00010; 2242/2529 tok/s; 2201 sec
  503. [2019-02-04 18:27:16,452 INFO] Step 21400/102583; acc: 80.01; ppl: 1.91; xent: 0.65; lr: 0.00010; 1975/2639 tok/s; 2206 sec
  504. [2019-02-04 18:27:21,543 INFO] Step 21450/102583; acc: 82.13; ppl: 1.83; xent: 0.61; lr: 0.00010; 2047/2450 tok/s; 2211 sec
  505. [2019-02-04 18:27:26,328 INFO] Step 21500/102583; acc: 79.39; ppl: 1.97; xent: 0.68; lr: 0.00010; 2024/2752 tok/s; 2216 sec
  506. [2019-02-04 18:27:31,340 INFO] Step 21550/102583; acc: 81.72; ppl: 1.83; xent: 0.60; lr: 0.00010; 2002/2614 tok/s; 2221 sec
  507. [2019-02-04 18:27:36,142 INFO] Step 21600/102583; acc: 80.03; ppl: 1.96; xent: 0.67; lr: 0.00010; 2049/2644 tok/s; 2226 sec
  508. [2019-02-04 18:27:41,355 INFO] Step 21650/102583; acc: 82.58; ppl: 1.77; xent: 0.57; lr: 0.00010; 2075/2548 tok/s; 2231 sec
  509. [2019-02-04 18:27:45,956 INFO] Step 21700/102583; acc: 79.34; ppl: 1.97; xent: 0.68; lr: 0.00010; 2117/2565 tok/s; 2235 sec
  510. [2019-02-04 18:27:51,077 INFO] Step 21750/102583; acc: 81.77; ppl: 1.81; xent: 0.59; lr: 0.00010; 2090/2590 tok/s; 2240 sec
  511. [2019-02-04 18:27:56,410 INFO] Step 21800/102583; acc: 81.43; ppl: 1.82; xent: 0.60; lr: 0.00010; 1904/2756 tok/s; 2246 sec
  512. [2019-02-04 18:28:01,206 INFO] Step 21850/102583; acc: 80.27; ppl: 1.92; xent: 0.65; lr: 0.00010; 2047/2651 tok/s; 2251 sec
  513. [2019-02-04 18:28:05,979 INFO] Step 21900/102583; acc: 82.16; ppl: 1.82; xent: 0.60; lr: 0.00010; 2135/2534 tok/s; 2255 sec
  514. [2019-02-04 18:28:11,012 INFO] Step 21950/102583; acc: 80.15; ppl: 1.90; xent: 0.64; lr: 0.00010; 2076/2699 tok/s; 2260 sec
  515. [2019-02-04 18:28:15,806 INFO] Step 22000/102583; acc: 80.06; ppl: 1.92; xent: 0.65; lr: 0.00010; 2005/2653 tok/s; 2265 sec
  516. [2019-02-04 18:28:20,825 INFO] Step 22050/102583; acc: 81.59; ppl: 1.84; xent: 0.61; lr: 0.00010; 1995/2641 tok/s; 2270 sec
  517. [2019-02-04 18:28:25,588 INFO] Step 22100/102583; acc: 81.89; ppl: 1.82; xent: 0.60; lr: 0.00010; 2162/2626 tok/s; 2275 sec
  518. [2019-02-04 18:28:30,789 INFO] Step 22150/102583; acc: 80.23; ppl: 1.89; xent: 0.64; lr: 0.00010; 1971/2589 tok/s; 2280 sec
  519. [2019-02-04 18:28:36,028 INFO] Step 22200/102583; acc: 79.49; ppl: 1.96; xent: 0.68; lr: 0.00010; 1920/2661 tok/s; 2285 sec
  520. [2019-02-04 18:28:40,737 INFO] Step 22250/102583; acc: 80.40; ppl: 1.93; xent: 0.65; lr: 0.00010; 2067/2553 tok/s; 2290 sec
  521. [2019-02-04 18:28:45,551 INFO] Step 22300/102583; acc: 82.42; ppl: 1.76; xent: 0.57; lr: 0.00010; 2012/2825 tok/s; 2295 sec
  522. [2019-02-04 18:28:50,897 INFO] Step 22350/102583; acc: 82.05; ppl: 1.81; xent: 0.60; lr: 0.00010; 2075/2523 tok/s; 2300 sec
  523. [2019-02-04 18:28:55,731 INFO] Step 22400/102583; acc: 80.62; ppl: 1.88; xent: 0.63; lr: 0.00010; 2078/2725 tok/s; 2305 sec
  524. [2019-02-04 18:29:00,653 INFO] Step 22450/102583; acc: 80.59; ppl: 1.91; xent: 0.65; lr: 0.00010; 2242/2435 tok/s; 2310 sec
  525. [2019-02-04 18:29:06,040 INFO] Step 22500/102583; acc: 81.20; ppl: 1.84; xent: 0.61; lr: 0.00010; 1944/2604 tok/s; 2315 sec
  526. [2019-02-04 18:29:10,639 INFO] Step 22550/102583; acc: 78.93; ppl: 2.00; xent: 0.69; lr: 0.00010; 2036/2674 tok/s; 2320 sec
  527. [2019-02-04 18:29:15,433 INFO] Step 22600/102583; acc: 81.13; ppl: 1.85; xent: 0.62; lr: 0.00010; 2125/2545 tok/s; 2325 sec
  528. [2019-02-04 18:29:20,691 INFO] Step 22650/102583; acc: 80.50; ppl: 1.89; xent: 0.64; lr: 0.00010; 1879/2662 tok/s; 2330 sec
  529. [2019-02-04 18:29:25,873 INFO] Step 22700/102583; acc: 81.15; ppl: 1.85; xent: 0.61; lr: 0.00010; 1921/2524 tok/s; 2335 sec
  530. [2019-02-04 18:29:31,337 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  531. [2019-02-04 18:29:36,359 INFO] Step 22750/102583; acc: 81.21; ppl: 1.85; xent: 0.62; lr: 0.00010; 1043/1244 tok/s; 2346 sec
  532. [2019-02-04 18:29:41,043 INFO] Step 22800/102583; acc: 80.30; ppl: 1.93; xent: 0.66; lr: 0.00010; 2112/2626 tok/s; 2350 sec
  533. [2019-02-04 18:29:46,139 INFO] Step 22850/102583; acc: 82.39; ppl: 1.78; xent: 0.58; lr: 0.00010; 2018/2744 tok/s; 2356 sec
  534. [2019-02-04 18:29:51,027 INFO] Step 22900/102583; acc: 79.99; ppl: 1.94; xent: 0.66; lr: 0.00010; 2023/2597 tok/s; 2360 sec
  535. [2019-02-04 18:29:55,918 INFO] Step 22950/102583; acc: 81.23; ppl: 1.85; xent: 0.62; lr: 0.00010; 1960/2808 tok/s; 2365 sec
  536. [2019-02-04 18:30:00,845 INFO] Step 23000/102583; acc: 81.16; ppl: 1.83; xent: 0.61; lr: 0.00010; 2219/2505 tok/s; 2370 sec
  537. [2019-02-04 18:30:05,789 INFO] Step 23050/102583; acc: 81.15; ppl: 1.85; xent: 0.61; lr: 0.00010; 2180/2640 tok/s; 2375 sec
  538. [2019-02-04 18:30:10,615 INFO] Step 23100/102583; acc: 80.29; ppl: 1.94; xent: 0.66; lr: 0.00010; 2060/2609 tok/s; 2380 sec
  539. [2019-02-04 18:30:15,809 INFO] Step 23150/102583; acc: 79.59; ppl: 1.93; xent: 0.66; lr: 0.00010; 1854/2712 tok/s; 2385 sec
  540. [2019-02-04 18:30:20,623 INFO] Step 23200/102583; acc: 80.82; ppl: 1.89; xent: 0.64; lr: 0.00010; 2106/2622 tok/s; 2390 sec
  541. [2019-02-04 18:30:25,858 INFO] Step 23250/102583; acc: 81.50; ppl: 1.81; xent: 0.59; lr: 0.00010; 1908/2732 tok/s; 2395 sec
  542. [2019-02-04 18:30:30,655 INFO] Step 23300/102583; acc: 80.79; ppl: 1.91; xent: 0.65; lr: 0.00010; 2288/2523 tok/s; 2400 sec
  543. [2019-02-04 18:30:35,548 INFO] Step 23350/102583; acc: 79.89; ppl: 1.94; xent: 0.66; lr: 0.00010; 2061/2632 tok/s; 2405 sec
  544. [2019-02-04 18:30:40,395 INFO] Step 23400/102583; acc: 79.73; ppl: 1.93; xent: 0.66; lr: 0.00010; 1907/2637 tok/s; 2410 sec
  545. [2019-02-04 18:30:45,484 INFO] Step 23450/102583; acc: 83.06; ppl: 1.76; xent: 0.56; lr: 0.00010; 2297/2552 tok/s; 2415 sec
  546. [2019-02-04 18:30:50,530 INFO] Step 23500/102583; acc: 80.54; ppl: 1.88; xent: 0.63; lr: 0.00010; 1928/2763 tok/s; 2420 sec
  547. [2019-02-04 18:30:55,395 INFO] Step 23550/102583; acc: 80.81; ppl: 1.88; xent: 0.63; lr: 0.00010; 2058/2421 tok/s; 2425 sec
  548. [2019-02-04 18:31:00,339 INFO] Step 23600/102583; acc: 81.92; ppl: 1.81; xent: 0.59; lr: 0.00010; 1975/2694 tok/s; 2430 sec
  549. [2019-02-04 18:31:05,475 INFO] Step 23650/102583; acc: 80.53; ppl: 1.90; xent: 0.64; lr: 0.00010; 2121/2470 tok/s; 2435 sec
  550. [2019-02-04 18:31:10,371 INFO] Step 23700/102583; acc: 80.43; ppl: 1.89; xent: 0.64; lr: 0.00010; 1948/2716 tok/s; 2440 sec
  551. [2019-02-04 18:31:15,727 INFO] Step 23750/102583; acc: 80.70; ppl: 1.86; xent: 0.62; lr: 0.00010; 1968/2697 tok/s; 2445 sec
  552. [2019-02-04 18:31:20,434 INFO] Step 23800/102583; acc: 81.25; ppl: 1.86; xent: 0.62; lr: 0.00010; 2110/2545 tok/s; 2450 sec
  553. [2019-02-04 18:31:25,426 INFO] Step 23850/102583; acc: 80.06; ppl: 1.92; xent: 0.65; lr: 0.00010; 2122/2561 tok/s; 2455 sec
  554. [2019-02-04 18:31:30,367 INFO] Step 23900/102583; acc: 79.94; ppl: 1.92; xent: 0.65; lr: 0.00010; 1984/2679 tok/s; 2460 sec
  555. [2019-02-04 18:31:35,082 INFO] Step 23950/102583; acc: 79.71; ppl: 1.95; xent: 0.67; lr: 0.00010; 2098/2584 tok/s; 2464 sec
  556. [2019-02-04 18:31:40,434 INFO] Step 24000/102583; acc: 81.92; ppl: 1.81; xent: 0.59; lr: 0.00010; 1918/2647 tok/s; 2470 sec
  557. [2019-02-04 18:31:45,556 INFO] Step 24050/102583; acc: 82.68; ppl: 1.80; xent: 0.59; lr: 0.00010; 2248/2331 tok/s; 2475 sec
  558. [2019-02-04 18:31:50,370 INFO] Step 24100/102583; acc: 78.91; ppl: 1.95; xent: 0.67; lr: 0.00010; 1804/2870 tok/s; 2480 sec
  559. [2019-02-04 18:31:55,379 INFO] Step 24150/102583; acc: 80.89; ppl: 1.84; xent: 0.61; lr: 0.00010; 2048/2685 tok/s; 2485 sec
  560. [2019-02-04 18:32:00,235 INFO] Step 24200/102583; acc: 81.41; ppl: 1.86; xent: 0.62; lr: 0.00010; 2180/2504 tok/s; 2490 sec
  561. [2019-02-04 18:32:04,926 INFO] Step 24250/102583; acc: 80.19; ppl: 1.94; xent: 0.66; lr: 0.00010; 1825/2664 tok/s; 2494 sec
  562. [2019-02-04 18:32:10,241 INFO] Step 24300/102583; acc: 81.59; ppl: 1.78; xent: 0.58; lr: 0.00010; 2156/2573 tok/s; 2500 sec
  563. [2019-02-04 18:32:15,101 INFO] Step 24350/102583; acc: 79.83; ppl: 1.93; xent: 0.66; lr: 0.00010; 2038/2726 tok/s; 2504 sec
  564. [2019-02-04 18:32:20,057 INFO] Step 24400/102583; acc: 81.18; ppl: 1.86; xent: 0.62; lr: 0.00010; 2031/2524 tok/s; 2509 sec
  565. [2019-02-04 18:32:25,150 INFO] Step 24450/102583; acc: 81.32; ppl: 1.85; xent: 0.62; lr: 0.00010; 1915/2728 tok/s; 2515 sec
  566. [2019-02-04 18:32:30,084 INFO] Step 24500/102583; acc: 81.09; ppl: 1.85; xent: 0.62; lr: 0.00010; 2201/2560 tok/s; 2519 sec
  567. [2019-02-04 18:32:35,057 INFO] Step 24550/102583; acc: 79.87; ppl: 1.93; xent: 0.66; lr: 0.00010; 1871/2686 tok/s; 2524 sec
  568. [2019-02-04 18:32:39,978 INFO] Step 24600/102583; acc: 80.80; ppl: 1.85; xent: 0.62; lr: 0.00010; 2187/2597 tok/s; 2529 sec
  569. [2019-02-04 18:32:44,909 INFO] Step 24650/102583; acc: 81.49; ppl: 1.84; xent: 0.61; lr: 0.00010; 2060/2585 tok/s; 2534 sec
  570. [2019-02-04 18:32:50,047 INFO] Step 24700/102583; acc: 80.84; ppl: 1.88; xent: 0.63; lr: 0.00010; 1943/2502 tok/s; 2539 sec
  571. [2019-02-04 18:32:54,845 INFO] Step 24750/102583; acc: 80.80; ppl: 1.91; xent: 0.65; lr: 0.00010; 2023/2699 tok/s; 2544 sec
  572. [2019-02-04 18:32:59,766 INFO] Step 24800/102583; acc: 81.40; ppl: 1.85; xent: 0.61; lr: 0.00010; 2100/2619 tok/s; 2549 sec
  573. [2019-02-04 18:33:04,733 INFO] Step 24850/102583; acc: 81.01; ppl: 1.89; xent: 0.64; lr: 0.00010; 2019/2604 tok/s; 2554 sec
  574. [2019-02-04 18:33:09,738 INFO] Step 24900/102583; acc: 81.77; ppl: 1.81; xent: 0.59; lr: 0.00010; 2145/2561 tok/s; 2559 sec
  575. [2019-02-04 18:33:14,405 INFO] Step 24950/102583; acc: 80.76; ppl: 1.89; xent: 0.64; lr: 0.00010; 2143/2597 tok/s; 2564 sec
  576. [2019-02-04 18:33:19,581 INFO] Step 25000/102583; acc: 81.66; ppl: 1.81; xent: 0.60; lr: 0.00010; 2018/2627 tok/s; 2569 sec
  577. [2019-02-04 18:33:19,583 INFO] Saving checkpoint models/ted1_step_25000.pt
  578. [2019-02-04 18:33:31,561 INFO] Step 25050/102583; acc: 81.00; ppl: 1.83; xent: 0.60; lr: 0.00010; 842/1202 tok/s; 2581 sec
  579. [2019-02-04 18:33:36,351 INFO] Step 25100/102583; acc: 80.26; ppl: 1.94; xent: 0.66; lr: 0.00010; 2012/2624 tok/s; 2586 sec
  580. [2019-02-04 18:33:41,146 INFO] Step 25150/102583; acc: 83.25; ppl: 1.76; xent: 0.57; lr: 0.00010; 2202/2566 tok/s; 2591 sec
  581. [2019-02-04 18:33:46,221 INFO] Step 25200/102583; acc: 79.60; ppl: 1.93; xent: 0.66; lr: 0.00010; 2062/2686 tok/s; 2596 sec
  582. [2019-02-04 18:33:50,860 INFO] Step 25250/102583; acc: 80.02; ppl: 1.91; xent: 0.65; lr: 0.00010; 1999/2638 tok/s; 2600 sec
  583. [2019-02-04 18:33:55,960 INFO] Step 25300/102583; acc: 81.40; ppl: 1.86; xent: 0.62; lr: 0.00010; 1936/2674 tok/s; 2605 sec
  584. [2019-02-04 18:34:00,677 INFO] Step 25350/102583; acc: 82.36; ppl: 1.79; xent: 0.58; lr: 0.00010; 2234/2604 tok/s; 2610 sec
  585. [2019-02-04 18:34:05,921 INFO] Step 25400/102583; acc: 80.53; ppl: 1.89; xent: 0.64; lr: 0.00010; 1931/2614 tok/s; 2615 sec
  586. [2019-02-04 18:34:11,063 INFO] Step 25450/102583; acc: 79.49; ppl: 1.96; xent: 0.67; lr: 0.00010; 1956/2624 tok/s; 2620 sec
  587. [2019-02-04 18:34:16,011 INFO] Step 25500/102583; acc: 81.83; ppl: 1.84; xent: 0.61; lr: 0.00010; 2112/2567 tok/s; 2625 sec
  588. [2019-02-04 18:34:20,893 INFO] Step 25550/102583; acc: 82.23; ppl: 1.77; xent: 0.57; lr: 0.00010; 1909/2804 tok/s; 2630 sec
  589. [2019-02-04 18:34:26,115 INFO] Step 25600/102583; acc: 81.82; ppl: 1.83; xent: 0.61; lr: 0.00010; 2171/2503 tok/s; 2636 sec
  590. [2019-02-04 18:34:30,855 INFO] Step 25650/102583; acc: 81.09; ppl: 1.85; xent: 0.61; lr: 0.00010; 2048/2768 tok/s; 2640 sec
  591. [2019-02-04 18:34:35,792 INFO] Step 25700/102583; acc: 80.04; ppl: 1.96; xent: 0.67; lr: 0.00010; 2145/2370 tok/s; 2645 sec
  592. [2019-02-04 18:34:41,101 INFO] Step 25750/102583; acc: 81.80; ppl: 1.82; xent: 0.60; lr: 0.00010; 2083/2657 tok/s; 2650 sec
  593. [2019-02-04 18:34:45,685 INFO] Step 25800/102583; acc: 79.45; ppl: 1.96; xent: 0.67; lr: 0.00010; 1948/2691 tok/s; 2655 sec
  594. [2019-02-04 18:34:50,557 INFO] Step 25850/102583; acc: 81.58; ppl: 1.81; xent: 0.60; lr: 0.00010; 2090/2573 tok/s; 2660 sec
  595. [2019-02-04 18:34:55,785 INFO] Step 25900/102583; acc: 80.55; ppl: 1.88; xent: 0.63; lr: 0.00010; 1908/2666 tok/s; 2665 sec
  596. [2019-02-04 18:35:01,027 INFO] Step 25950/102583; acc: 81.22; ppl: 1.86; xent: 0.62; lr: 0.00010; 1931/2440 tok/s; 2670 sec
  597. [2019-02-04 18:35:06,672 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  598. [2019-02-04 18:35:12,224 INFO] Step 26000/102583; acc: 81.15; ppl: 1.85; xent: 0.62; lr: 0.00010; 950/1159 tok/s; 2682 sec
  599. [2019-02-04 18:35:17,023 INFO] Step 26050/102583; acc: 81.47; ppl: 1.86; xent: 0.62; lr: 0.00010; 2101/2684 tok/s; 2686 sec
  600. [2019-02-04 18:35:22,045 INFO] Step 26100/102583; acc: 82.34; ppl: 1.80; xent: 0.59; lr: 0.00010; 2022/2718 tok/s; 2691 sec
  601. [2019-02-04 18:35:26,933 INFO] Step 26150/102583; acc: 80.34; ppl: 1.91; xent: 0.65; lr: 0.00010; 2010/2591 tok/s; 2696 sec
  602. [2019-02-04 18:35:31,765 INFO] Step 26200/102583; acc: 81.81; ppl: 1.83; xent: 0.60; lr: 0.00010; 2050/2766 tok/s; 2701 sec
  603. [2019-02-04 18:35:36,854 INFO] Step 26250/102583; acc: 79.93; ppl: 1.90; xent: 0.64; lr: 0.00010; 2104/2540 tok/s; 2706 sec
  604. [2019-02-04 18:35:41,656 INFO] Step 26300/102583; acc: 81.15; ppl: 1.85; xent: 0.62; lr: 0.00010; 2173/2626 tok/s; 2711 sec
  605. [2019-02-04 18:35:46,655 INFO] Step 26350/102583; acc: 81.21; ppl: 1.86; xent: 0.62; lr: 0.00010; 2003/2710 tok/s; 2716 sec
  606. [2019-02-04 18:35:51,750 INFO] Step 26400/102583; acc: 79.75; ppl: 1.94; xent: 0.67; lr: 0.00010; 1944/2594 tok/s; 2721 sec
  607. [2019-02-04 18:35:56,509 INFO] Step 26450/102583; acc: 80.66; ppl: 1.88; xent: 0.63; lr: 0.00010; 2125/2654 tok/s; 2726 sec
  608. [2019-02-04 18:36:01,675 INFO] Step 26500/102583; acc: 81.83; ppl: 1.79; xent: 0.58; lr: 0.00010; 1945/2711 tok/s; 2731 sec
  609. [2019-02-04 18:36:06,527 INFO] Step 26550/102583; acc: 81.04; ppl: 1.90; xent: 0.64; lr: 0.00010; 2249/2557 tok/s; 2736 sec
  610. [2019-02-04 18:36:11,517 INFO] Step 26600/102583; acc: 79.75; ppl: 1.91; xent: 0.65; lr: 0.00010; 2023/2589 tok/s; 2741 sec
  611. [2019-02-04 18:36:16,252 INFO] Step 26650/102583; acc: 81.02; ppl: 1.88; xent: 0.63; lr: 0.00010; 2031/2654 tok/s; 2746 sec
  612. [2019-02-04 18:36:21,464 INFO] Step 26700/102583; acc: 81.86; ppl: 1.80; xent: 0.59; lr: 0.00010; 2169/2636 tok/s; 2751 sec
  613. [2019-02-04 18:36:26,346 INFO] Step 26750/102583; acc: 80.94; ppl: 1.87; xent: 0.63; lr: 0.00010; 1986/2662 tok/s; 2756 sec
  614. [2019-02-04 18:36:31,403 INFO] Step 26800/102583; acc: 81.80; ppl: 1.81; xent: 0.59; lr: 0.00010; 2051/2458 tok/s; 2761 sec
  615. [2019-02-04 18:36:36,375 INFO] Step 26850/102583; acc: 80.91; ppl: 1.85; xent: 0.62; lr: 0.00010; 1952/2654 tok/s; 2766 sec
  616. [2019-02-04 18:36:41,444 INFO] Step 26900/102583; acc: 80.56; ppl: 1.88; xent: 0.63; lr: 0.00010; 2073/2467 tok/s; 2771 sec
  617. [2019-02-04 18:36:46,391 INFO] Step 26950/102583; acc: 81.54; ppl: 1.83; xent: 0.61; lr: 0.00010; 2027/2755 tok/s; 2776 sec
  618. [2019-02-04 18:36:51,593 INFO] Step 27000/102583; acc: 80.11; ppl: 1.88; xent: 0.63; lr: 0.00010; 1934/2688 tok/s; 2781 sec
  619. [2019-02-04 18:36:56,441 INFO] Step 27050/102583; acc: 80.93; ppl: 1.88; xent: 0.63; lr: 0.00010; 2082/2545 tok/s; 2786 sec
  620. [2019-02-04 18:37:01,491 INFO] Step 27100/102583; acc: 80.74; ppl: 1.88; xent: 0.63; lr: 0.00010; 2091/2546 tok/s; 2791 sec
  621. [2019-02-04 18:37:06,374 INFO] Step 27150/102583; acc: 80.05; ppl: 1.91; xent: 0.64; lr: 0.00010; 2018/2660 tok/s; 2796 sec
  622. [2019-02-04 18:37:10,941 INFO] Step 27200/102583; acc: 80.00; ppl: 1.94; xent: 0.66; lr: 0.00010; 2084/2651 tok/s; 2800 sec
  623. [2019-02-04 18:37:16,211 INFO] Step 27250/102583; acc: 82.31; ppl: 1.79; xent: 0.58; lr: 0.00010; 1972/2604 tok/s; 2806 sec
  624. [2019-02-04 18:37:21,592 INFO] Step 27300/102583; acc: 82.62; ppl: 1.77; xent: 0.57; lr: 0.00010; 2176/2468 tok/s; 2811 sec
  625. [2019-02-04 18:37:26,341 INFO] Step 27350/102583; acc: 79.25; ppl: 1.94; xent: 0.66; lr: 0.00010; 1881/2725 tok/s; 2816 sec
  626. [2019-02-04 18:37:31,437 INFO] Step 27400/102583; acc: 80.37; ppl: 1.85; xent: 0.62; lr: 0.00010; 2076/2671 tok/s; 2821 sec
  627. [2019-02-04 18:37:36,121 INFO] Step 27450/102583; acc: 81.60; ppl: 1.84; xent: 0.61; lr: 0.00010; 2085/2589 tok/s; 2826 sec
  628. [2019-02-04 18:37:40,826 INFO] Step 27500/102583; acc: 80.37; ppl: 1.94; xent: 0.66; lr: 0.00010; 1929/2615 tok/s; 2830 sec
  629. [2019-02-04 18:37:46,131 INFO] Step 27550/102583; acc: 81.43; ppl: 1.79; xent: 0.58; lr: 0.00010; 2049/2546 tok/s; 2836 sec
  630. [2019-02-04 18:37:51,037 INFO] Step 27600/102583; acc: 80.21; ppl: 1.88; xent: 0.63; lr: 0.00010; 2088/2809 tok/s; 2840 sec
  631. [2019-02-04 18:37:55,997 INFO] Step 27650/102583; acc: 81.65; ppl: 1.84; xent: 0.61; lr: 0.00010; 2040/2503 tok/s; 2845 sec
  632. [2019-02-04 18:38:01,027 INFO] Step 27700/102583; acc: 81.91; ppl: 1.83; xent: 0.60; lr: 0.00010; 1925/2737 tok/s; 2850 sec
  633. [2019-02-04 18:38:05,867 INFO] Step 27750/102583; acc: 80.04; ppl: 1.91; xent: 0.65; lr: 0.00010; 2165/2534 tok/s; 2855 sec
  634. [2019-02-04 18:38:10,874 INFO] Step 27800/102583; acc: 80.63; ppl: 1.88; xent: 0.63; lr: 0.00010; 1922/2704 tok/s; 2860 sec
  635. [2019-02-04 18:38:15,857 INFO] Step 27850/102583; acc: 81.34; ppl: 1.84; xent: 0.61; lr: 0.00010; 2157/2514 tok/s; 2865 sec
  636. [2019-02-04 18:38:20,954 INFO] Step 27900/102583; acc: 81.52; ppl: 1.84; xent: 0.61; lr: 0.00010; 2163/2603 tok/s; 2870 sec
  637. [2019-02-04 18:38:25,890 INFO] Step 27950/102583; acc: 81.30; ppl: 1.87; xent: 0.63; lr: 0.00010; 1871/2570 tok/s; 2875 sec
  638. [2019-02-04 18:38:30,675 INFO] Step 28000/102583; acc: 80.64; ppl: 1.90; xent: 0.64; lr: 0.00010; 2029/2682 tok/s; 2880 sec
  639. [2019-02-04 18:38:35,812 INFO] Step 28050/102583; acc: 81.52; ppl: 1.83; xent: 0.61; lr: 0.00010; 2082/2555 tok/s; 2885 sec
  640. [2019-02-04 18:38:40,566 INFO] Step 28100/102583; acc: 80.98; ppl: 1.89; xent: 0.64; lr: 0.00010; 1994/2683 tok/s; 2890 sec
  641. [2019-02-04 18:38:45,593 INFO] Step 28150/102583; acc: 82.18; ppl: 1.80; xent: 0.59; lr: 0.00010; 2113/2580 tok/s; 2895 sec
  642. [2019-02-04 18:38:50,308 INFO] Step 28200/102583; acc: 81.62; ppl: 1.83; xent: 0.61; lr: 0.00010; 2174/2587 tok/s; 2900 sec
  643. [2019-02-04 18:38:55,567 INFO] Step 28250/102583; acc: 80.54; ppl: 1.86; xent: 0.62; lr: 0.00010; 1903/2631 tok/s; 2905 sec
  644. [2019-02-04 18:39:00,609 INFO] Step 28300/102583; acc: 81.47; ppl: 1.83; xent: 0.60; lr: 0.00010; 2063/2684 tok/s; 2910 sec
  645. [2019-02-04 18:39:05,412 INFO] Step 28350/102583; acc: 80.85; ppl: 1.90; xent: 0.64; lr: 0.00010; 2031/2620 tok/s; 2915 sec
  646. [2019-02-04 18:39:10,432 INFO] Step 28400/102583; acc: 82.80; ppl: 1.77; xent: 0.57; lr: 0.00010; 2073/2603 tok/s; 2920 sec
  647. [2019-02-04 18:39:15,314 INFO] Step 28450/102583; acc: 80.20; ppl: 1.89; xent: 0.64; lr: 0.00010; 2086/2694 tok/s; 2925 sec
  648. [2019-02-04 18:39:19,898 INFO] Step 28500/102583; acc: 80.32; ppl: 1.90; xent: 0.64; lr: 0.00010; 2079/2579 tok/s; 2929 sec
  649. [2019-02-04 18:39:25,263 INFO] Step 28550/102583; acc: 81.90; ppl: 1.81; xent: 0.59; lr: 0.00010; 1973/2673 tok/s; 2935 sec
  650. [2019-02-04 18:39:30,013 INFO] Step 28600/102583; acc: 81.43; ppl: 1.84; xent: 0.61; lr: 0.00010; 2218/2551 tok/s; 2939 sec
  651. [2019-02-04 18:39:35,122 INFO] Step 28650/102583; acc: 80.56; ppl: 1.89; xent: 0.63; lr: 0.00010; 1888/2688 tok/s; 2945 sec
  652. [2019-02-04 18:39:40,194 INFO] Step 28700/102583; acc: 80.37; ppl: 1.93; xent: 0.66; lr: 0.00010; 1933/2604 tok/s; 2950 sec
  653. [2019-02-04 18:39:45,113 INFO] Step 28750/102583; acc: 82.41; ppl: 1.79; xent: 0.58; lr: 0.00010; 2154/2651 tok/s; 2955 sec
  654. [2019-02-04 18:39:49,914 INFO] Step 28800/102583; acc: 81.83; ppl: 1.80; xent: 0.59; lr: 0.00010; 1920/2750 tok/s; 2959 sec
  655. [2019-02-04 18:39:55,085 INFO] Step 28850/102583; acc: 81.72; ppl: 1.83; xent: 0.61; lr: 0.00010; 2142/2475 tok/s; 2964 sec
  656. [2019-02-04 18:39:59,877 INFO] Step 28900/102583; acc: 81.17; ppl: 1.86; xent: 0.62; lr: 0.00010; 2027/2761 tok/s; 2969 sec
  657. [2019-02-04 18:40:04,888 INFO] Step 28950/102583; acc: 80.38; ppl: 1.91; xent: 0.65; lr: 0.00010; 2138/2434 tok/s; 2974 sec
  658. [2019-02-04 18:40:10,141 INFO] Step 29000/102583; acc: 81.20; ppl: 1.84; xent: 0.61; lr: 0.00010; 2091/2624 tok/s; 2980 sec
  659. [2019-02-04 18:40:14,822 INFO] Step 29050/102583; acc: 80.51; ppl: 1.88; xent: 0.63; lr: 0.00010; 1907/2761 tok/s; 2984 sec
  660. [2019-02-04 18:40:19,606 INFO] Step 29100/102583; acc: 81.43; ppl: 1.85; xent: 0.61; lr: 0.00010; 2115/2529 tok/s; 2989 sec
  661. [2019-02-04 18:40:25,000 INFO] Step 29150/102583; acc: 80.92; ppl: 1.85; xent: 0.61; lr: 0.00010; 1964/2635 tok/s; 2994 sec
  662. [2019-02-04 18:40:35,122 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  663. [2019-02-04 18:40:36,172 INFO] Step 29200/102583; acc: 80.93; ppl: 1.89; xent: 0.64; lr: 0.00010; 882/1120 tok/s; 3006 sec
  664. [2019-02-04 18:40:41,250 INFO] Step 29250/102583; acc: 81.64; ppl: 1.82; xent: 0.60; lr: 0.00010; 2152/2559 tok/s; 3011 sec
  665. [2019-02-04 18:40:45,982 INFO] Step 29300/102583; acc: 81.16; ppl: 1.87; xent: 0.63; lr: 0.00010; 1954/2702 tok/s; 3015 sec
  666. [2019-02-04 18:40:51,350 INFO] Step 29350/102583; acc: 82.73; ppl: 1.76; xent: 0.57; lr: 0.00010; 2083/2644 tok/s; 3021 sec
  667. [2019-02-04 18:40:56,086 INFO] Step 29400/102583; acc: 80.38; ppl: 1.90; xent: 0.64; lr: 0.00010; 2080/2662 tok/s; 3025 sec
  668. [2019-02-04 18:41:01,084 INFO] Step 29450/102583; acc: 82.18; ppl: 1.80; xent: 0.59; lr: 0.00010; 2086/2680 tok/s; 3030 sec
  669. [2019-02-04 18:41:06,065 INFO] Step 29500/102583; acc: 79.91; ppl: 1.89; xent: 0.64; lr: 0.00010; 2065/2573 tok/s; 3035 sec
  670. [2019-02-04 18:41:10,901 INFO] Step 29550/102583; acc: 81.58; ppl: 1.83; xent: 0.60; lr: 0.00010; 2097/2762 tok/s; 3040 sec
  671. [2019-02-04 18:41:15,884 INFO] Step 29600/102583; acc: 80.45; ppl: 1.90; xent: 0.64; lr: 0.00010; 1945/2696 tok/s; 3045 sec
  672. [2019-02-04 18:41:20,951 INFO] Step 29650/102583; acc: 80.23; ppl: 1.92; xent: 0.65; lr: 0.00010; 2056/2531 tok/s; 3050 sec
  673. [2019-02-04 18:41:25,716 INFO] Step 29700/102583; acc: 81.02; ppl: 1.84; xent: 0.61; lr: 0.00010; 2069/2644 tok/s; 3055 sec
  674. [2019-02-04 18:41:30,934 INFO] Step 29750/102583; acc: 81.53; ppl: 1.82; xent: 0.60; lr: 0.00010; 1985/2672 tok/s; 3060 sec
  675. [2019-02-04 18:41:35,585 INFO] Step 29800/102583; acc: 81.48; ppl: 1.87; xent: 0.62; lr: 0.00010; 2241/2542 tok/s; 3065 sec
  676. [2019-02-04 18:41:40,648 INFO] Step 29850/102583; acc: 79.64; ppl: 1.92; xent: 0.65; lr: 0.00010; 2032/2632 tok/s; 3070 sec
  677. [2019-02-04 18:41:45,563 INFO] Step 29900/102583; acc: 81.95; ppl: 1.81; xent: 0.59; lr: 0.00010; 2044/2648 tok/s; 3075 sec
  678. [2019-02-04 18:41:50,577 INFO] Step 29950/102583; acc: 81.51; ppl: 1.83; xent: 0.60; lr: 0.00010; 2145/2647 tok/s; 3080 sec
  679. [2019-02-04 18:41:55,527 INFO] Step 30000/102583; acc: 80.69; ppl: 1.88; xent: 0.63; lr: 0.00010; 1997/2597 tok/s; 3085 sec
  680. [2019-02-04 18:41:55,559 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  681. [2019-02-04 18:41:56,858 INFO] Validation perplexity: 1.74967
  682. [2019-02-04 18:41:56,859 INFO] Validation accuracy: 82.5772
  683. [2019-02-04 18:41:56,861 INFO] Saving checkpoint models/ted1_step_30000.pt
  684. [2019-02-04 18:42:07,550 INFO] Step 30050/102583; acc: 81.17; ppl: 1.82; xent: 0.60; lr: 0.00010; 852/1085 tok/s; 3097 sec
  685. [2019-02-04 18:42:12,613 INFO] Step 30100/102583; acc: 82.26; ppl: 1.79; xent: 0.58; lr: 0.00010; 1993/2630 tok/s; 3102 sec
  686. [2019-02-04 18:42:17,584 INFO] Step 30150/102583; acc: 80.34; ppl: 1.92; xent: 0.65; lr: 0.00010; 2062/2484 tok/s; 3107 sec
  687. [2019-02-04 18:42:22,567 INFO] Step 30200/102583; acc: 80.61; ppl: 1.86; xent: 0.62; lr: 0.00010; 2023/2691 tok/s; 3112 sec
  688. [2019-02-04 18:42:27,765 INFO] Step 30250/102583; acc: 81.00; ppl: 1.84; xent: 0.61; lr: 0.00010; 1873/2600 tok/s; 3117 sec
  689. [2019-02-04 18:42:32,738 INFO] Step 30300/102583; acc: 81.79; ppl: 1.82; xent: 0.60; lr: 0.00010; 2210/2599 tok/s; 3122 sec
  690. [2019-02-04 18:42:37,640 INFO] Step 30350/102583; acc: 80.09; ppl: 1.90; xent: 0.64; lr: 0.00010; 2070/2593 tok/s; 3127 sec
  691. [2019-02-04 18:42:42,388 INFO] Step 30400/102583; acc: 79.65; ppl: 1.94; xent: 0.66; lr: 0.00010; 2019/2658 tok/s; 3132 sec
  692. [2019-02-04 18:42:47,148 INFO] Step 30450/102583; acc: 81.12; ppl: 1.86; xent: 0.62; lr: 0.00010; 2059/2698 tok/s; 3137 sec
  693. [2019-02-04 18:42:52,498 INFO] Step 30500/102583; acc: 82.12; ppl: 1.80; xent: 0.59; lr: 0.00010; 2034/2474 tok/s; 3142 sec
  694. [2019-02-04 18:42:57,653 INFO] Step 30550/102583; acc: 81.47; ppl: 1.83; xent: 0.60; lr: 0.00010; 2071/2591 tok/s; 3147 sec
  695. [2019-02-04 18:43:02,504 INFO] Step 30600/102583; acc: 80.23; ppl: 1.88; xent: 0.63; lr: 0.00010; 1893/2696 tok/s; 3152 sec
  696. [2019-02-04 18:43:07,558 INFO] Step 30650/102583; acc: 80.93; ppl: 1.83; xent: 0.61; lr: 0.00010; 2122/2713 tok/s; 3157 sec
  697. [2019-02-04 18:43:12,270 INFO] Step 30700/102583; acc: 80.87; ppl: 1.88; xent: 0.63; lr: 0.00010; 2002/2524 tok/s; 3162 sec
  698. [2019-02-04 18:43:17,030 INFO] Step 30750/102583; acc: 81.17; ppl: 1.88; xent: 0.63; lr: 0.00010; 1929/2657 tok/s; 3166 sec
  699. [2019-02-04 18:43:22,322 INFO] Step 30800/102583; acc: 81.65; ppl: 1.79; xent: 0.58; lr: 0.00010; 2141/2511 tok/s; 3172 sec
  700. [2019-02-04 18:43:27,116 INFO] Step 30850/102583; acc: 79.37; ppl: 1.93; xent: 0.66; lr: 0.00010; 2023/2860 tok/s; 3177 sec
  701. [2019-02-04 18:43:32,142 INFO] Step 30900/102583; acc: 82.10; ppl: 1.81; xent: 0.59; lr: 0.00010; 2053/2546 tok/s; 3182 sec
  702. [2019-02-04 18:43:37,256 INFO] Step 30950/102583; acc: 82.56; ppl: 1.80; xent: 0.59; lr: 0.00010; 2001/2626 tok/s; 3187 sec
  703. [2019-02-04 18:43:41,982 INFO] Step 31000/102583; acc: 79.71; ppl: 1.92; xent: 0.65; lr: 0.00010; 2074/2621 tok/s; 3191 sec
  704. [2019-02-04 18:43:47,059 INFO] Step 31050/102583; acc: 81.20; ppl: 1.85; xent: 0.62; lr: 0.00010; 1953/2656 tok/s; 3196 sec
  705. [2019-02-04 18:43:51,902 INFO] Step 31100/102583; acc: 80.94; ppl: 1.84; xent: 0.61; lr: 0.00010; 2172/2514 tok/s; 3201 sec
  706. [2019-02-04 18:43:57,257 INFO] Step 31150/102583; acc: 82.73; ppl: 1.77; xent: 0.57; lr: 0.00010; 2151/2578 tok/s; 3207 sec
  707. [2019-02-04 18:44:01,937 INFO] Step 31200/102583; acc: 80.94; ppl: 1.88; xent: 0.63; lr: 0.00010; 1854/2615 tok/s; 3211 sec
  708. [2019-02-04 18:44:06,873 INFO] Step 31250/102583; acc: 79.88; ppl: 1.95; xent: 0.67; lr: 0.00010; 1939/2667 tok/s; 3216 sec
  709. [2019-02-04 18:44:11,972 INFO] Step 31300/102583; acc: 81.90; ppl: 1.80; xent: 0.59; lr: 0.00010; 2100/2552 tok/s; 3221 sec
  710. [2019-02-04 18:44:16,744 INFO] Step 31350/102583; acc: 81.69; ppl: 1.81; xent: 0.60; lr: 0.00010; 2099/2702 tok/s; 3226 sec
  711. [2019-02-04 18:44:21,713 INFO] Step 31400/102583; acc: 81.49; ppl: 1.84; xent: 0.61; lr: 0.00010; 2115/2536 tok/s; 3231 sec
  712. [2019-02-04 18:44:26,416 INFO] Step 31450/102583; acc: 80.77; ppl: 1.86; xent: 0.62; lr: 0.00010; 2089/2708 tok/s; 3236 sec
  713. [2019-02-04 18:44:31,685 INFO] Step 31500/102583; acc: 81.51; ppl: 1.81; xent: 0.59; lr: 0.00010; 1948/2615 tok/s; 3241 sec
  714. [2019-02-04 18:44:36,692 INFO] Step 31550/102583; acc: 81.19; ppl: 1.82; xent: 0.60; lr: 0.00010; 2039/2676 tok/s; 3246 sec
  715. [2019-02-04 18:44:41,618 INFO] Step 31600/102583; acc: 81.43; ppl: 1.86; xent: 0.62; lr: 0.00010; 2086/2513 tok/s; 3251 sec
  716. [2019-02-04 18:44:46,504 INFO] Step 31650/102583; acc: 82.52; ppl: 1.78; xent: 0.57; lr: 0.00010; 2062/2694 tok/s; 3256 sec
  717. [2019-02-04 18:44:51,362 INFO] Step 31700/102583; acc: 80.00; ppl: 1.89; xent: 0.64; lr: 0.00010; 2084/2664 tok/s; 3261 sec
  718. [2019-02-04 18:44:56,163 INFO] Step 31750/102583; acc: 80.94; ppl: 1.89; xent: 0.63; lr: 0.00010; 2032/2529 tok/s; 3266 sec
  719. [2019-02-04 18:45:01,300 INFO] Step 31800/102583; acc: 81.97; ppl: 1.80; xent: 0.59; lr: 0.00010; 1996/2734 tok/s; 3271 sec
  720. [2019-02-04 18:45:06,266 INFO] Step 31850/102583; acc: 81.76; ppl: 1.82; xent: 0.60; lr: 0.00010; 2206/2529 tok/s; 3276 sec
  721. [2019-02-04 18:45:11,324 INFO] Step 31900/102583; acc: 81.09; ppl: 1.85; xent: 0.62; lr: 0.00010; 1823/2799 tok/s; 3281 sec
  722. [2019-02-04 18:45:16,312 INFO] Step 31950/102583; acc: 79.38; ppl: 1.97; xent: 0.68; lr: 0.00010; 2011/2477 tok/s; 3286 sec
  723. [2019-02-04 18:45:21,210 INFO] Step 32000/102583; acc: 82.43; ppl: 1.77; xent: 0.57; lr: 0.00010; 2104/2710 tok/s; 3291 sec
  724. [2019-02-04 18:45:26,602 INFO] Step 32050/102583; acc: 81.40; ppl: 1.81; xent: 0.59; lr: 0.00010; 1913/2771 tok/s; 3296 sec
  725. [2019-02-04 18:45:31,888 INFO] Step 32100/102583; acc: 81.13; ppl: 1.87; xent: 0.62; lr: 0.00010; 2126/2462 tok/s; 3301 sec
  726. [2019-02-04 18:45:36,691 INFO] Step 32150/102583; acc: 82.21; ppl: 1.80; xent: 0.59; lr: 0.00010; 2070/2705 tok/s; 3306 sec
  727. [2019-02-04 18:45:41,840 INFO] Step 32200/102583; acc: 80.86; ppl: 1.86; xent: 0.62; lr: 0.00010; 2141/2482 tok/s; 3311 sec
  728. [2019-02-04 18:45:46,952 INFO] Step 32250/102583; acc: 80.75; ppl: 1.87; xent: 0.63; lr: 0.00010; 2034/2651 tok/s; 3316 sec
  729. [2019-02-04 18:45:51,583 INFO] Step 32300/102583; acc: 81.50; ppl: 1.83; xent: 0.61; lr: 0.00010; 2035/2672 tok/s; 3321 sec
  730. [2019-02-04 18:45:56,205 INFO] Step 32350/102583; acc: 81.36; ppl: 1.85; xent: 0.61; lr: 0.00010; 2128/2570 tok/s; 3326 sec
  731. [2019-02-04 18:46:01,579 INFO] Step 32400/102583; acc: 81.17; ppl: 1.84; xent: 0.61; lr: 0.00010; 1947/2631 tok/s; 3331 sec
  732. [2019-02-04 18:46:11,247 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  733. [2019-02-04 18:46:12,842 INFO] Step 32450/102583; acc: 80.16; ppl: 1.92; xent: 0.65; lr: 0.00010; 875/1134 tok/s; 3342 sec
  734. [2019-02-04 18:46:17,935 INFO] Step 32500/102583; acc: 81.92; ppl: 1.82; xent: 0.60; lr: 0.00010; 2136/2545 tok/s; 3347 sec
  735. [2019-02-04 18:46:22,683 INFO] Step 32550/102583; acc: 81.94; ppl: 1.82; xent: 0.60; lr: 0.00010; 1910/2728 tok/s; 3352 sec
  736. [2019-02-04 18:46:27,809 INFO] Step 32600/102583; acc: 83.09; ppl: 1.75; xent: 0.56; lr: 0.00010; 2120/2633 tok/s; 3357 sec
  737. [2019-02-04 18:46:32,725 INFO] Step 32650/102583; acc: 80.77; ppl: 1.87; xent: 0.62; lr: 0.00010; 2098/2688 tok/s; 3362 sec
  738. [2019-02-04 18:46:37,698 INFO] Step 32700/102583; acc: 81.18; ppl: 1.83; xent: 0.60; lr: 0.00010; 2079/2652 tok/s; 3367 sec
  739. [2019-02-04 18:46:42,591 INFO] Step 32750/102583; acc: 80.62; ppl: 1.86; xent: 0.62; lr: 0.00010; 2070/2540 tok/s; 3372 sec
  740. [2019-02-04 18:46:47,388 INFO] Step 32800/102583; acc: 81.70; ppl: 1.82; xent: 0.60; lr: 0.00010; 2110/2821 tok/s; 3377 sec
  741. [2019-02-04 18:46:52,370 INFO] Step 32850/102583; acc: 80.45; ppl: 1.89; xent: 0.64; lr: 0.00010; 1919/2692 tok/s; 3382 sec
  742. [2019-02-04 18:46:57,516 INFO] Step 32900/102583; acc: 80.17; ppl: 1.92; xent: 0.65; lr: 0.00010; 2070/2498 tok/s; 3387 sec
  743. [2019-02-04 18:47:02,239 INFO] Step 32950/102583; acc: 82.02; ppl: 1.78; xent: 0.58; lr: 0.00010; 2118/2694 tok/s; 3392 sec
  744. [2019-02-04 18:47:07,544 INFO] Step 33000/102583; acc: 80.60; ppl: 1.88; xent: 0.63; lr: 0.00010; 2001/2574 tok/s; 3397 sec
  745. [2019-02-04 18:47:12,016 INFO] Step 33050/102583; acc: 81.49; ppl: 1.88; xent: 0.63; lr: 0.00010; 2189/2598 tok/s; 3401 sec
  746. [2019-02-04 18:47:17,263 INFO] Step 33100/102583; acc: 79.99; ppl: 1.88; xent: 0.63; lr: 0.00010; 1948/2694 tok/s; 3407 sec
  747. [2019-02-04 18:47:22,020 INFO] Step 33150/102583; acc: 82.15; ppl: 1.82; xent: 0.60; lr: 0.00010; 2125/2590 tok/s; 3411 sec
  748. [2019-02-04 18:47:27,249 INFO] Step 33200/102583; acc: 81.93; ppl: 1.80; xent: 0.59; lr: 0.00010; 2122/2633 tok/s; 3417 sec
  749. [2019-02-04 18:47:31,939 INFO] Step 33250/102583; acc: 81.43; ppl: 1.85; xent: 0.61; lr: 0.00010; 2062/2591 tok/s; 3421 sec
  750. [2019-02-04 18:47:37,051 INFO] Step 33300/102583; acc: 81.32; ppl: 1.82; xent: 0.60; lr: 0.00010; 1992/2610 tok/s; 3426 sec
  751. [2019-02-04 18:47:42,133 INFO] Step 33350/102583; acc: 81.65; ppl: 1.82; xent: 0.60; lr: 0.00010; 2036/2621 tok/s; 3432 sec
  752. [2019-02-04 18:47:47,135 INFO] Step 33400/102583; acc: 81.11; ppl: 1.87; xent: 0.63; lr: 0.00010; 2039/2449 tok/s; 3437 sec
  753. [2019-02-04 18:47:52,131 INFO] Step 33450/102583; acc: 79.45; ppl: 1.90; xent: 0.64; lr: 0.00010; 1952/2788 tok/s; 3442 sec
  754. [2019-02-04 18:47:57,357 INFO] Step 33500/102583; acc: 82.14; ppl: 1.79; xent: 0.58; lr: 0.00010; 1954/2517 tok/s; 3447 sec
  755. [2019-02-04 18:48:02,405 INFO] Step 33550/102583; acc: 81.58; ppl: 1.82; xent: 0.60; lr: 0.00010; 2190/2655 tok/s; 3452 sec
  756. [2019-02-04 18:48:07,113 INFO] Step 33600/102583; acc: 80.25; ppl: 1.91; xent: 0.65; lr: 0.00010; 2109/2547 tok/s; 3457 sec
  757. [2019-02-04 18:48:12,099 INFO] Step 33650/102583; acc: 80.29; ppl: 1.89; xent: 0.64; lr: 0.00010; 1908/2710 tok/s; 3461 sec
  758. [2019-02-04 18:48:16,881 INFO] Step 33700/102583; acc: 81.56; ppl: 1.85; xent: 0.62; lr: 0.00010; 2139/2667 tok/s; 3466 sec
  759. [2019-02-04 18:48:22,004 INFO] Step 33750/102583; acc: 81.89; ppl: 1.80; xent: 0.59; lr: 0.00010; 2091/2438 tok/s; 3471 sec
  760. [2019-02-04 18:48:27,267 INFO] Step 33800/102583; acc: 81.48; ppl: 1.81; xent: 0.60; lr: 0.00010; 1968/2661 tok/s; 3477 sec
  761. [2019-02-04 18:48:32,062 INFO] Step 33850/102583; acc: 80.54; ppl: 1.86; xent: 0.62; lr: 0.00010; 1874/2690 tok/s; 3481 sec
  762. [2019-02-04 18:48:37,057 INFO] Step 33900/102583; acc: 80.82; ppl: 1.85; xent: 0.62; lr: 0.00010; 2172/2703 tok/s; 3486 sec
  763. [2019-02-04 18:48:41,859 INFO] Step 33950/102583; acc: 81.55; ppl: 1.85; xent: 0.62; lr: 0.00010; 1981/2476 tok/s; 3491 sec
  764. [2019-02-04 18:48:46,540 INFO] Step 34000/102583; acc: 80.38; ppl: 1.89; xent: 0.64; lr: 0.00010; 2005/2626 tok/s; 3496 sec
  765. [2019-02-04 18:48:51,883 INFO] Step 34050/102583; acc: 81.87; ppl: 1.76; xent: 0.57; lr: 0.00010; 2113/2586 tok/s; 3501 sec
  766. [2019-02-04 18:48:56,701 INFO] Step 34100/102583; acc: 79.42; ppl: 1.94; xent: 0.66; lr: 0.00010; 1994/2760 tok/s; 3506 sec
  767. [2019-02-04 18:49:01,742 INFO] Step 34150/102583; acc: 82.14; ppl: 1.79; xent: 0.58; lr: 0.00010; 2051/2603 tok/s; 3511 sec
  768. [2019-02-04 18:49:06,800 INFO] Step 34200/102583; acc: 82.18; ppl: 1.81; xent: 0.60; lr: 0.00010; 2048/2645 tok/s; 3516 sec
  769. [2019-02-04 18:49:11,577 INFO] Step 34250/102583; acc: 79.55; ppl: 1.93; xent: 0.66; lr: 0.00010; 1957/2559 tok/s; 3521 sec
  770. [2019-02-04 18:49:16,646 INFO] Step 34300/102583; acc: 82.08; ppl: 1.80; xent: 0.59; lr: 0.00010; 2076/2660 tok/s; 3526 sec
  771. [2019-02-04 18:49:21,492 INFO] Step 34350/102583; acc: 81.18; ppl: 1.85; xent: 0.61; lr: 0.00010; 2060/2615 tok/s; 3531 sec
  772. [2019-02-04 18:49:26,792 INFO] Step 34400/102583; acc: 82.36; ppl: 1.79; xent: 0.58; lr: 0.00010; 2136/2488 tok/s; 3536 sec
  773. [2019-02-04 18:49:31,414 INFO] Step 34450/102583; acc: 81.52; ppl: 1.85; xent: 0.62; lr: 0.00010; 1934/2666 tok/s; 3541 sec
  774. [2019-02-04 18:49:36,373 INFO] Step 34500/102583; acc: 79.50; ppl: 1.95; xent: 0.67; lr: 0.00010; 1995/2633 tok/s; 3546 sec
  775. [2019-02-04 18:49:41,453 INFO] Step 34550/102583; acc: 81.92; ppl: 1.82; xent: 0.60; lr: 0.00010; 2007/2586 tok/s; 3551 sec
  776. [2019-02-04 18:49:46,186 INFO] Step 34600/102583; acc: 81.71; ppl: 1.81; xent: 0.60; lr: 0.00010; 2090/2691 tok/s; 3556 sec
  777. [2019-02-04 18:49:51,217 INFO] Step 34650/102583; acc: 81.61; ppl: 1.84; xent: 0.61; lr: 0.00010; 2163/2462 tok/s; 3561 sec
  778. [2019-02-04 18:49:56,056 INFO] Step 34700/102583; acc: 82.08; ppl: 1.79; xent: 0.58; lr: 0.00010; 2061/2730 tok/s; 3565 sec
  779. [2019-02-04 18:50:01,127 INFO] Step 34750/102583; acc: 81.05; ppl: 1.84; xent: 0.61; lr: 0.00010; 1946/2668 tok/s; 3571 sec
  780. [2019-02-04 18:50:06,126 INFO] Step 34800/102583; acc: 80.91; ppl: 1.84; xent: 0.61; lr: 0.00010; 2067/2663 tok/s; 3576 sec
  781. [2019-02-04 18:50:11,119 INFO] Step 34850/102583; acc: 82.72; ppl: 1.79; xent: 0.58; lr: 0.00010; 2123/2513 tok/s; 3581 sec
  782. [2019-02-04 18:50:16,112 INFO] Step 34900/102583; acc: 82.15; ppl: 1.79; xent: 0.58; lr: 0.00010; 2029/2670 tok/s; 3586 sec
  783. [2019-02-04 18:50:21,102 INFO] Step 34950/102583; acc: 80.56; ppl: 1.87; xent: 0.63; lr: 0.00010; 2050/2663 tok/s; 3590 sec
  784. [2019-02-04 18:50:25,731 INFO] Step 35000/102583; acc: 81.06; ppl: 1.88; xent: 0.63; lr: 0.00010; 2027/2627 tok/s; 3595 sec
  785. [2019-02-04 18:50:25,732 INFO] Saving checkpoint models/ted1_step_35000.pt
  786. [2019-02-04 18:50:36,997 INFO] Step 35050/102583; acc: 81.64; ppl: 1.82; xent: 0.60; lr: 0.00010; 927/1178 tok/s; 3606 sec
  787. [2019-02-04 18:50:42,090 INFO] Step 35100/102583; acc: 81.69; ppl: 1.81; xent: 0.59; lr: 0.00010; 2117/2511 tok/s; 3611 sec
  788. [2019-02-04 18:50:47,210 INFO] Step 35150/102583; acc: 80.93; ppl: 1.85; xent: 0.62; lr: 0.00010; 1754/2882 tok/s; 3617 sec
  789. [2019-02-04 18:50:52,075 INFO] Step 35200/102583; acc: 79.30; ppl: 1.97; xent: 0.68; lr: 0.00010; 2046/2469 tok/s; 3621 sec
  790. [2019-02-04 18:50:56,951 INFO] Step 35250/102583; acc: 83.06; ppl: 1.74; xent: 0.55; lr: 0.00010; 2193/2700 tok/s; 3626 sec
  791. [2019-02-04 18:51:01,709 INFO] Step 35300/102583; acc: 81.05; ppl: 1.82; xent: 0.60; lr: 0.00010; 1951/2716 tok/s; 3631 sec
  792. [2019-02-04 18:51:07,260 INFO] Step 35350/102583; acc: 81.33; ppl: 1.85; xent: 0.61; lr: 0.00010; 2086/2522 tok/s; 3637 sec
  793. [2019-02-04 18:51:11,979 INFO] Step 35400/102583; acc: 82.04; ppl: 1.82; xent: 0.60; lr: 0.00010; 2106/2651 tok/s; 3641 sec
  794. [2019-02-04 18:51:17,087 INFO] Step 35450/102583; acc: 81.65; ppl: 1.82; xent: 0.60; lr: 0.00010; 2170/2465 tok/s; 3646 sec
  795. [2019-02-04 18:51:22,257 INFO] Step 35500/102583; acc: 80.64; ppl: 1.88; xent: 0.63; lr: 0.00010; 1987/2672 tok/s; 3652 sec
  796. [2019-02-04 18:51:26,863 INFO] Step 35550/102583; acc: 81.57; ppl: 1.82; xent: 0.60; lr: 0.00010; 1949/2686 tok/s; 3656 sec
  797. [2019-02-04 18:51:31,704 INFO] Step 35600/102583; acc: 81.56; ppl: 1.83; xent: 0.61; lr: 0.00010; 2110/2573 tok/s; 3661 sec
  798. [2019-02-04 18:51:36,891 INFO] Step 35650/102583; acc: 80.81; ppl: 1.84; xent: 0.61; lr: 0.00010; 2006/2648 tok/s; 3666 sec
  799. [2019-02-04 18:51:46,029 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  800. [2019-02-04 18:51:48,154 INFO] Step 35700/102583; acc: 80.14; ppl: 1.92; xent: 0.65; lr: 0.00010; 869/1117 tok/s; 3678 sec
  801. [2019-02-04 18:51:53,261 INFO] Step 35750/102583; acc: 82.81; ppl: 1.78; xent: 0.58; lr: 0.00010; 2181/2541 tok/s; 3683 sec
  802. [2019-02-04 18:51:58,035 INFO] Step 35800/102583; acc: 81.54; ppl: 1.83; xent: 0.60; lr: 0.00010; 1886/2751 tok/s; 3687 sec
  803. [2019-02-04 18:52:03,228 INFO] Step 35850/102583; acc: 82.98; ppl: 1.75; xent: 0.56; lr: 0.00010; 2080/2640 tok/s; 3693 sec
  804. [2019-02-04 18:52:08,050 INFO] Step 35900/102583; acc: 81.44; ppl: 1.85; xent: 0.61; lr: 0.00010; 2138/2680 tok/s; 3697 sec
  805. [2019-02-04 18:52:13,217 INFO] Step 35950/102583; acc: 81.59; ppl: 1.80; xent: 0.59; lr: 0.00010; 2018/2701 tok/s; 3703 sec
  806. [2019-02-04 18:52:18,063 INFO] Step 36000/102583; acc: 80.51; ppl: 1.87; xent: 0.63; lr: 0.00010; 2078/2510 tok/s; 3707 sec
  807. [2019-02-04 18:52:22,841 INFO] Step 36050/102583; acc: 82.28; ppl: 1.80; xent: 0.59; lr: 0.00010; 2175/2765 tok/s; 3712 sec
  808. [2019-02-04 18:52:27,963 INFO] Step 36100/102583; acc: 79.25; ppl: 1.93; xent: 0.66; lr: 0.00010; 1966/2600 tok/s; 3717 sec
  809. [2019-02-04 18:52:32,865 INFO] Step 36150/102583; acc: 80.76; ppl: 1.90; xent: 0.64; lr: 0.00010; 1983/2630 tok/s; 3722 sec
  810. [2019-02-04 18:52:37,646 INFO] Step 36200/102583; acc: 82.33; ppl: 1.75; xent: 0.56; lr: 0.00010; 2200/2713 tok/s; 3727 sec
  811. [2019-02-04 18:52:42,857 INFO] Step 36250/102583; acc: 81.04; ppl: 1.88; xent: 0.63; lr: 0.00010; 1959/2556 tok/s; 3732 sec
  812. [2019-02-04 18:52:47,541 INFO] Step 36300/102583; acc: 80.96; ppl: 1.89; xent: 0.64; lr: 0.00010; 2141/2540 tok/s; 3737 sec
  813. [2019-02-04 18:52:52,576 INFO] Step 36350/102583; acc: 80.12; ppl: 1.88; xent: 0.63; lr: 0.00010; 1970/2766 tok/s; 3742 sec
  814. [2019-02-04 18:52:57,493 INFO] Step 36400/102583; acc: 82.82; ppl: 1.78; xent: 0.57; lr: 0.00010; 2090/2597 tok/s; 3747 sec
  815. [2019-02-04 18:53:02,562 INFO] Step 36450/102583; acc: 81.56; ppl: 1.83; xent: 0.60; lr: 0.00010; 2169/2622 tok/s; 3752 sec
  816. [2019-02-04 18:53:07,332 INFO] Step 36500/102583; acc: 81.30; ppl: 1.84; xent: 0.61; lr: 0.00010; 2144/2498 tok/s; 3757 sec
  817. [2019-02-04 18:53:12,356 INFO] Step 36550/102583; acc: 81.15; ppl: 1.83; xent: 0.60; lr: 0.00010; 1866/2696 tok/s; 3762 sec
  818. [2019-02-04 18:53:17,590 INFO] Step 36600/102583; acc: 82.37; ppl: 1.78; xent: 0.58; lr: 0.00010; 2051/2636 tok/s; 3767 sec
  819. [2019-02-04 18:53:22,807 INFO] Step 36650/102583; acc: 80.78; ppl: 1.88; xent: 0.63; lr: 0.00010; 2053/2469 tok/s; 3772 sec
  820. [2019-02-04 18:53:27,756 INFO] Step 36700/102583; acc: 79.36; ppl: 1.91; xent: 0.65; lr: 0.00010; 1895/2752 tok/s; 3777 sec
  821. [2019-02-04 18:53:32,918 INFO] Step 36750/102583; acc: 83.13; ppl: 1.74; xent: 0.55; lr: 0.00010; 1988/2584 tok/s; 3782 sec
  822. [2019-02-04 18:53:37,902 INFO] Step 36800/102583; acc: 81.14; ppl: 1.86; xent: 0.62; lr: 0.00010; 2216/2504 tok/s; 3787 sec
  823. [2019-02-04 18:53:42,541 INFO] Step 36850/102583; acc: 80.54; ppl: 1.89; xent: 0.64; lr: 0.00010; 2100/2644 tok/s; 3792 sec
  824. [2019-02-04 18:53:47,564 INFO] Step 36900/102583; acc: 80.70; ppl: 1.87; xent: 0.63; lr: 0.00010; 1984/2710 tok/s; 3797 sec
  825. [2019-02-04 18:53:52,604 INFO] Step 36950/102583; acc: 81.32; ppl: 1.84; xent: 0.61; lr: 0.00010; 2101/2611 tok/s; 3802 sec
  826. [2019-02-04 18:53:57,460 INFO] Step 37000/102583; acc: 82.49; ppl: 1.78; xent: 0.58; lr: 0.00010; 2132/2474 tok/s; 3807 sec
  827. [2019-02-04 18:54:02,694 INFO] Step 37050/102583; acc: 80.73; ppl: 1.85; xent: 0.61; lr: 0.00010; 1917/2665 tok/s; 3812 sec
  828. [2019-02-04 18:54:07,545 INFO] Step 37100/102583; acc: 80.80; ppl: 1.86; xent: 0.62; lr: 0.00010; 1865/2728 tok/s; 3817 sec
  829. [2019-02-04 18:54:12,569 INFO] Step 37150/102583; acc: 81.59; ppl: 1.80; xent: 0.59; lr: 0.00010; 2138/2601 tok/s; 3822 sec
  830. [2019-02-04 18:54:17,313 INFO] Step 37200/102583; acc: 81.47; ppl: 1.87; xent: 0.63; lr: 0.00010; 2021/2558 tok/s; 3827 sec
  831. [2019-02-04 18:54:22,359 INFO] Step 37250/102583; acc: 81.09; ppl: 1.85; xent: 0.62; lr: 0.00010; 2072/2527 tok/s; 3832 sec
  832. [2019-02-04 18:54:27,461 INFO] Step 37300/102583; acc: 81.14; ppl: 1.83; xent: 0.60; lr: 0.00010; 1993/2687 tok/s; 3837 sec
  833. [2019-02-04 18:54:32,324 INFO] Step 37350/102583; acc: 80.88; ppl: 1.86; xent: 0.62; lr: 0.00010; 2134/2698 tok/s; 3842 sec
  834. [2019-02-04 18:54:37,319 INFO] Step 37400/102583; acc: 81.80; ppl: 1.81; xent: 0.59; lr: 0.00010; 1996/2664 tok/s; 3847 sec
  835. [2019-02-04 18:54:42,250 INFO] Step 37450/102583; acc: 81.93; ppl: 1.83; xent: 0.61; lr: 0.00010; 2016/2651 tok/s; 3852 sec
  836. [2019-02-04 18:54:47,093 INFO] Step 37500/102583; acc: 80.08; ppl: 1.91; xent: 0.65; lr: 0.00010; 2017/2476 tok/s; 3856 sec
  837. [2019-02-04 18:54:52,137 INFO] Step 37550/102583; acc: 81.81; ppl: 1.79; xent: 0.58; lr: 0.00010; 2015/2764 tok/s; 3862 sec
  838. [2019-02-04 18:54:57,175 INFO] Step 37600/102583; acc: 81.32; ppl: 1.85; xent: 0.61; lr: 0.00010; 2103/2501 tok/s; 3867 sec
  839. [2019-02-04 18:55:02,326 INFO] Step 37650/102583; acc: 82.21; ppl: 1.81; xent: 0.59; lr: 0.00010; 2127/2495 tok/s; 3872 sec
  840. [2019-02-04 18:55:06,894 INFO] Step 37700/102583; acc: 81.41; ppl: 1.86; xent: 0.62; lr: 0.00010; 1903/2744 tok/s; 3876 sec
  841. [2019-02-04 18:55:11,876 INFO] Step 37750/102583; acc: 79.83; ppl: 1.91; xent: 0.65; lr: 0.00010; 2037/2646 tok/s; 3881 sec
  842. [2019-02-04 18:55:16,904 INFO] Step 37800/102583; acc: 82.21; ppl: 1.81; xent: 0.59; lr: 0.00010; 2002/2558 tok/s; 3886 sec
  843. [2019-02-04 18:55:21,611 INFO] Step 37850/102583; acc: 82.22; ppl: 1.78; xent: 0.58; lr: 0.00010; 2060/2693 tok/s; 3891 sec
  844. [2019-02-04 18:55:26,725 INFO] Step 37900/102583; acc: 81.33; ppl: 1.86; xent: 0.62; lr: 0.00010; 2178/2485 tok/s; 3896 sec
  845. [2019-02-04 18:55:31,762 INFO] Step 37950/102583; acc: 82.63; ppl: 1.75; xent: 0.56; lr: 0.00010; 2135/2603 tok/s; 3901 sec
  846. [2019-02-04 18:55:36,794 INFO] Step 38000/102583; acc: 80.64; ppl: 1.85; xent: 0.61; lr: 0.00010; 1785/2818 tok/s; 3906 sec
  847. [2019-02-04 18:55:41,785 INFO] Step 38050/102583; acc: 81.09; ppl: 1.85; xent: 0.61; lr: 0.00010; 2096/2640 tok/s; 3911 sec
  848. [2019-02-04 18:55:46,561 INFO] Step 38100/102583; acc: 83.04; ppl: 1.77; xent: 0.57; lr: 0.00010; 2200/2494 tok/s; 3916 sec
  849. [2019-02-04 18:55:51,484 INFO] Step 38150/102583; acc: 81.91; ppl: 1.81; xent: 0.59; lr: 0.00010; 1979/2656 tok/s; 3921 sec
  850. [2019-02-04 18:55:56,496 INFO] Step 38200/102583; acc: 80.06; ppl: 1.88; xent: 0.63; lr: 0.00010; 2105/2667 tok/s; 3926 sec
  851. [2019-02-04 18:56:01,176 INFO] Step 38250/102583; acc: 81.06; ppl: 1.88; xent: 0.63; lr: 0.00010; 2059/2625 tok/s; 3931 sec
  852. [2019-02-04 18:56:06,349 INFO] Step 38300/102583; acc: 82.41; ppl: 1.77; xent: 0.57; lr: 0.00010; 2014/2670 tok/s; 3936 sec
  853. [2019-02-04 18:56:11,380 INFO] Step 38350/102583; acc: 81.83; ppl: 1.79; xent: 0.58; lr: 0.00010; 2113/2561 tok/s; 3941 sec
  854. [2019-02-04 18:56:16,354 INFO] Step 38400/102583; acc: 81.21; ppl: 1.89; xent: 0.63; lr: 0.00010; 1793/2805 tok/s; 3946 sec
  855. [2019-02-04 18:56:21,309 INFO] Step 38450/102583; acc: 80.29; ppl: 1.92; xent: 0.65; lr: 0.00010; 1983/2506 tok/s; 3951 sec
  856. [2019-02-04 18:56:26,162 INFO] Step 38500/102583; acc: 82.57; ppl: 1.75; xent: 0.56; lr: 0.00010; 2190/2704 tok/s; 3956 sec
  857. [2019-02-04 18:56:31,201 INFO] Step 38550/102583; acc: 81.43; ppl: 1.81; xent: 0.60; lr: 0.00010; 1997/2537 tok/s; 3961 sec
  858. [2019-02-04 18:56:36,334 INFO] Step 38600/102583; acc: 81.74; ppl: 1.80; xent: 0.59; lr: 0.00010; 2112/2680 tok/s; 3966 sec
  859. [2019-02-04 18:56:41,228 INFO] Step 38650/102583; acc: 81.00; ppl: 1.88; xent: 0.63; lr: 0.00010; 2068/2666 tok/s; 3971 sec
  860. [2019-02-04 18:56:46,310 INFO] Step 38700/102583; acc: 81.83; ppl: 1.80; xent: 0.59; lr: 0.00010; 2076/2473 tok/s; 3976 sec
  861. [2019-02-04 18:56:51,232 INFO] Step 38750/102583; acc: 81.17; ppl: 1.86; xent: 0.62; lr: 0.00010; 2133/2594 tok/s; 3981 sec
  862. [2019-02-04 18:56:56,065 INFO] Step 38800/102583; acc: 81.25; ppl: 1.83; xent: 0.61; lr: 0.00010; 1960/2658 tok/s; 3985 sec
  863. [2019-02-04 18:57:00,827 INFO] Step 38850/102583; acc: 81.37; ppl: 1.85; xent: 0.61; lr: 0.00010; 1988/2690 tok/s; 3990 sec
  864. [2019-02-04 18:57:06,104 INFO] Step 38900/102583; acc: 81.12; ppl: 1.82; xent: 0.60; lr: 0.00010; 1991/2586 tok/s; 3995 sec
  865. [2019-02-04 18:57:13,787 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  866. [2019-02-04 18:57:16,574 INFO] Step 38950/102583; acc: 80.82; ppl: 1.87; xent: 0.63; lr: 0.00010; 972/1219 tok/s; 4006 sec
  867. [2019-02-04 18:57:21,558 INFO] Step 39000/102583; acc: 82.34; ppl: 1.81; xent: 0.60; lr: 0.00010; 2172/2458 tok/s; 4011 sec
  868. [2019-02-04 18:57:26,317 INFO] Step 39050/102583; acc: 81.88; ppl: 1.80; xent: 0.59; lr: 0.00010; 1968/2764 tok/s; 4016 sec
  869. [2019-02-04 18:57:31,427 INFO] Step 39100/102583; acc: 82.69; ppl: 1.76; xent: 0.57; lr: 0.00010; 2008/2692 tok/s; 4021 sec
  870. [2019-02-04 18:57:36,423 INFO] Step 39150/102583; acc: 80.47; ppl: 1.90; xent: 0.64; lr: 0.00010; 2089/2756 tok/s; 4026 sec
  871. [2019-02-04 18:57:41,512 INFO] Step 39200/102583; acc: 81.91; ppl: 1.77; xent: 0.57; lr: 0.00010; 2036/2651 tok/s; 4031 sec
  872. [2019-02-04 18:57:46,350 INFO] Step 39250/102583; acc: 81.62; ppl: 1.83; xent: 0.60; lr: 0.00010; 2161/2502 tok/s; 4036 sec
  873. [2019-02-04 18:57:51,025 INFO] Step 39300/102583; acc: 81.82; ppl: 1.84; xent: 0.61; lr: 0.00010; 2112/2724 tok/s; 4040 sec
  874. [2019-02-04 18:57:56,222 INFO] Step 39350/102583; acc: 79.84; ppl: 1.89; xent: 0.64; lr: 0.00010; 1982/2567 tok/s; 4046 sec
  875. [2019-02-04 18:58:01,233 INFO] Step 39400/102583; acc: 81.62; ppl: 1.85; xent: 0.61; lr: 0.00010; 1977/2671 tok/s; 4051 sec
  876. [2019-02-04 18:58:06,034 INFO] Step 39450/102583; acc: 82.39; ppl: 1.76; xent: 0.56; lr: 0.00010; 2163/2725 tok/s; 4055 sec
  877. [2019-02-04 18:58:11,195 INFO] Step 39500/102583; acc: 80.21; ppl: 1.92; xent: 0.65; lr: 0.00010; 1973/2482 tok/s; 4061 sec
  878. [2019-02-04 18:58:15,822 INFO] Step 39550/102583; acc: 80.87; ppl: 1.90; xent: 0.64; lr: 0.00010; 2146/2625 tok/s; 4065 sec
  879. [2019-02-04 18:58:20,879 INFO] Step 39600/102583; acc: 80.97; ppl: 1.82; xent: 0.60; lr: 0.00010; 2014/2772 tok/s; 4070 sec
  880. [2019-02-04 18:58:25,992 INFO] Step 39650/102583; acc: 83.05; ppl: 1.76; xent: 0.56; lr: 0.00010; 2116/2529 tok/s; 4075 sec
  881. [2019-02-04 18:58:30,940 INFO] Step 39700/102583; acc: 80.80; ppl: 1.86; xent: 0.62; lr: 0.00010; 2077/2711 tok/s; 4080 sec
  882. [2019-02-04 18:58:35,818 INFO] Step 39750/102583; acc: 82.36; ppl: 1.80; xent: 0.59; lr: 0.00010; 2202/2455 tok/s; 4085 sec
  883. [2019-02-04 18:58:40,895 INFO] Step 39800/102583; acc: 81.70; ppl: 1.79; xent: 0.58; lr: 0.00010; 1857/2762 tok/s; 4090 sec
  884. [2019-02-04 18:58:46,147 INFO] Step 39850/102583; acc: 80.99; ppl: 1.85; xent: 0.61; lr: 0.00010; 2050/2479 tok/s; 4096 sec
  885. [2019-02-04 18:58:51,177 INFO] Step 39900/102583; acc: 81.70; ppl: 1.81; xent: 0.60; lr: 0.00010; 2075/2595 tok/s; 4101 sec
  886. [2019-02-04 18:58:56,222 INFO] Step 39950/102583; acc: 78.99; ppl: 1.95; xent: 0.67; lr: 0.00010; 1854/2716 tok/s; 4106 sec
  887. [2019-02-04 18:59:01,278 INFO] Step 40000/102583; acc: 83.44; ppl: 1.72; xent: 0.54; lr: 0.00010; 2033/2579 tok/s; 4111 sec
  888. [2019-02-04 18:59:01,509 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  889. [2019-02-04 18:59:02,811 INFO] Validation perplexity: 1.74125
  890. [2019-02-04 18:59:02,830 INFO] Validation accuracy: 83.1339
  891. [2019-02-04 18:59:02,866 INFO] Saving checkpoint models/ted1_step_40000.pt
  892. [2019-02-04 18:59:16,913 INFO] Step 40050/102583; acc: 80.64; ppl: 1.87; xent: 0.63; lr: 0.00010; 682/809 tok/s; 4126 sec
  893. [2019-02-04 18:59:21,487 INFO] Step 40100/102583; acc: 80.60; ppl: 1.89; xent: 0.64; lr: 0.00010; 2144/2615 tok/s; 4131 sec
  894. [2019-02-04 18:59:26,541 INFO] Step 40150/102583; acc: 80.90; ppl: 1.85; xent: 0.62; lr: 0.00010; 2034/2722 tok/s; 4136 sec
  895. [2019-02-04 18:59:31,709 INFO] Step 40200/102583; acc: 82.18; ppl: 1.80; xent: 0.59; lr: 0.00010; 2052/2678 tok/s; 4141 sec
  896. [2019-02-04 18:59:36,783 INFO] Step 40250/102583; acc: 83.01; ppl: 1.76; xent: 0.56; lr: 0.00010; 2223/2327 tok/s; 4146 sec
  897. [2019-02-04 18:59:41,820 INFO] Step 40300/102583; acc: 79.46; ppl: 1.91; xent: 0.65; lr: 0.00010; 1798/2827 tok/s; 4151 sec
  898. [2019-02-04 18:59:46,659 INFO] Step 40350/102583; acc: 80.50; ppl: 1.85; xent: 0.62; lr: 0.00010; 1870/2718 tok/s; 4156 sec
  899. [2019-02-04 18:59:51,740 INFO] Step 40400/102583; acc: 83.59; ppl: 1.71; xent: 0.54; lr: 0.00010; 2241/2526 tok/s; 4161 sec
  900. [2019-02-04 18:59:56,422 INFO] Step 40450/102583; acc: 80.07; ppl: 1.94; xent: 0.66; lr: 0.00010; 1950/2606 tok/s; 4166 sec
  901. [2019-02-04 19:00:01,581 INFO] Step 40500/102583; acc: 81.50; ppl: 1.81; xent: 0.60; lr: 0.00010; 2057/2537 tok/s; 4171 sec
  902. [2019-02-04 19:00:06,701 INFO] Step 40550/102583; acc: 81.28; ppl: 1.82; xent: 0.60; lr: 0.00010; 2054/2741 tok/s; 4176 sec
  903. [2019-02-04 19:00:11,480 INFO] Step 40600/102583; acc: 80.88; ppl: 1.87; xent: 0.63; lr: 0.00010; 2040/2614 tok/s; 4181 sec
  904. [2019-02-04 19:00:16,483 INFO] Step 40650/102583; acc: 82.13; ppl: 1.79; xent: 0.58; lr: 0.00010; 2013/2743 tok/s; 4186 sec
  905. [2019-02-04 19:00:21,386 INFO] Step 40700/102583; acc: 81.89; ppl: 1.82; xent: 0.60; lr: 0.00010; 2067/2615 tok/s; 4191 sec
  906. [2019-02-04 19:00:26,421 INFO] Step 40750/102583; acc: 80.43; ppl: 1.89; xent: 0.64; lr: 0.00010; 1932/2519 tok/s; 4196 sec
  907. [2019-02-04 19:00:31,287 INFO] Step 40800/102583; acc: 81.26; ppl: 1.82; xent: 0.60; lr: 0.00010; 2140/2706 tok/s; 4201 sec
  908. [2019-02-04 19:00:36,395 INFO] Step 40850/102583; acc: 81.35; ppl: 1.83; xent: 0.61; lr: 0.00010; 2016/2549 tok/s; 4206 sec
  909. [2019-02-04 19:00:41,493 INFO] Step 40900/102583; acc: 82.91; ppl: 1.78; xent: 0.57; lr: 0.00010; 2186/2470 tok/s; 4211 sec
  910. [2019-02-04 19:00:46,150 INFO] Step 40950/102583; acc: 80.46; ppl: 1.92; xent: 0.65; lr: 0.00010; 1846/2798 tok/s; 4216 sec
  911. [2019-02-04 19:00:51,273 INFO] Step 41000/102583; acc: 81.09; ppl: 1.84; xent: 0.61; lr: 0.00010; 2102/2573 tok/s; 4221 sec
  912. [2019-02-04 19:00:56,109 INFO] Step 41050/102583; acc: 81.58; ppl: 1.86; xent: 0.62; lr: 0.00010; 1991/2587 tok/s; 4225 sec
  913. [2019-02-04 19:01:01,014 INFO] Step 41100/102583; acc: 83.15; ppl: 1.72; xent: 0.54; lr: 0.00010; 2090/2640 tok/s; 4230 sec
  914. [2019-02-04 19:01:05,993 INFO] Step 41150/102583; acc: 80.95; ppl: 1.89; xent: 0.63; lr: 0.00010; 2173/2499 tok/s; 4235 sec
  915. [2019-02-04 19:01:10,933 INFO] Step 41200/102583; acc: 82.56; ppl: 1.75; xent: 0.56; lr: 0.00010; 2117/2609 tok/s; 4240 sec
  916. [2019-02-04 19:01:16,017 INFO] Step 41250/102583; acc: 80.83; ppl: 1.84; xent: 0.61; lr: 0.00010; 1793/2830 tok/s; 4245 sec
  917. [2019-02-04 19:01:21,172 INFO] Step 41300/102583; acc: 81.94; ppl: 1.80; xent: 0.59; lr: 0.00010; 2104/2574 tok/s; 4251 sec
  918. [2019-02-04 19:01:25,727 INFO] Step 41350/102583; acc: 82.09; ppl: 1.82; xent: 0.60; lr: 0.00010; 2110/2588 tok/s; 4255 sec
  919. [2019-02-04 19:01:30,784 INFO] Step 41400/102583; acc: 81.96; ppl: 1.79; xent: 0.58; lr: 0.00010; 2052/2637 tok/s; 4260 sec
  920. [2019-02-04 19:01:35,749 INFO] Step 41450/102583; acc: 79.84; ppl: 1.89; xent: 0.64; lr: 0.00010; 2055/2658 tok/s; 4265 sec
  921. [2019-02-04 19:01:40,546 INFO] Step 41500/102583; acc: 81.28; ppl: 1.87; xent: 0.62; lr: 0.00010; 1978/2706 tok/s; 4270 sec
  922. [2019-02-04 19:01:45,463 INFO] Step 41550/102583; acc: 83.21; ppl: 1.74; xent: 0.56; lr: 0.00010; 2138/2586 tok/s; 4275 sec
  923. [2019-02-04 19:01:50,762 INFO] Step 41600/102583; acc: 82.27; ppl: 1.76; xent: 0.56; lr: 0.00010; 2060/2589 tok/s; 4280 sec
  924. [2019-02-04 19:01:55,773 INFO] Step 41650/102583; acc: 80.20; ppl: 1.92; xent: 0.65; lr: 0.00010; 1798/2814 tok/s; 4285 sec
  925. [2019-02-04 19:02:00,415 INFO] Step 41700/102583; acc: 80.53; ppl: 1.90; xent: 0.64; lr: 0.00010; 2090/2455 tok/s; 4290 sec
  926. [2019-02-04 19:02:05,470 INFO] Step 41750/102583; acc: 82.66; ppl: 1.74; xent: 0.55; lr: 0.00010; 2129/2733 tok/s; 4295 sec
  927. [2019-02-04 19:02:10,509 INFO] Step 41800/102583; acc: 81.84; ppl: 1.81; xent: 0.59; lr: 0.00010; 1968/2523 tok/s; 4300 sec
  928. [2019-02-04 19:02:15,471 INFO] Step 41850/102583; acc: 82.30; ppl: 1.78; xent: 0.58; lr: 0.00010; 2161/2705 tok/s; 4305 sec
  929. [2019-02-04 19:02:20,558 INFO] Step 41900/102583; acc: 81.17; ppl: 1.85; xent: 0.62; lr: 0.00010; 2112/2578 tok/s; 4310 sec
  930. [2019-02-04 19:02:25,803 INFO] Step 41950/102583; acc: 82.22; ppl: 1.78; xent: 0.58; lr: 0.00010; 2078/2489 tok/s; 4315 sec
  931. [2019-02-04 19:02:30,331 INFO] Step 42000/102583; acc: 80.75; ppl: 1.90; xent: 0.64; lr: 0.00010; 2047/2655 tok/s; 4320 sec
  932. [2019-02-04 19:02:35,182 INFO] Step 42050/102583; acc: 81.55; ppl: 1.81; xent: 0.59; lr: 0.00010; 2058/2641 tok/s; 4325 sec
  933. [2019-02-04 19:02:40,102 INFO] Step 42100/102583; acc: 81.82; ppl: 1.82; xent: 0.60; lr: 0.00010; 1967/2672 tok/s; 4329 sec
  934. [2019-02-04 19:02:45,421 INFO] Step 42150/102583; acc: 80.81; ppl: 1.83; xent: 0.60; lr: 0.00010; 1928/2534 tok/s; 4335 sec
  935. [2019-02-04 19:02:52,420 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  936. [2019-02-04 19:02:55,957 INFO] Step 42200/102583; acc: 81.88; ppl: 1.82; xent: 0.60; lr: 0.00010; 980/1289 tok/s; 4345 sec
  937. [2019-02-04 19:03:00,697 INFO] Step 42250/102583; acc: 81.10; ppl: 1.89; xent: 0.64; lr: 0.00010; 2221/2401 tok/s; 4350 sec
  938. [2019-02-04 19:03:05,500 INFO] Step 42300/102583; acc: 81.99; ppl: 1.80; xent: 0.59; lr: 0.00010; 1937/2801 tok/s; 4355 sec
  939. [2019-02-04 19:03:10,711 INFO] Step 42350/102583; acc: 82.47; ppl: 1.78; xent: 0.58; lr: 0.00010; 1987/2662 tok/s; 4360 sec
  940. [2019-02-04 19:03:15,559 INFO] Step 42400/102583; acc: 81.87; ppl: 1.84; xent: 0.61; lr: 0.00010; 2134/2776 tok/s; 4365 sec
  941. [2019-02-04 19:03:20,565 INFO] Step 42450/102583; acc: 81.59; ppl: 1.80; xent: 0.59; lr: 0.00010; 2019/2608 tok/s; 4370 sec
  942. [2019-02-04 19:03:25,552 INFO] Step 42500/102583; acc: 81.60; ppl: 1.81; xent: 0.60; lr: 0.00010; 2220/2502 tok/s; 4375 sec
  943. [2019-02-04 19:03:30,156 INFO] Step 42550/102583; acc: 81.27; ppl: 1.86; xent: 0.62; lr: 0.00010; 2080/2776 tok/s; 4380 sec
  944. [2019-02-04 19:03:35,416 INFO] Step 42600/102583; acc: 80.56; ppl: 1.87; xent: 0.63; lr: 0.00010; 1922/2622 tok/s; 4385 sec
  945. [2019-02-04 19:03:40,363 INFO] Step 42650/102583; acc: 81.07; ppl: 1.87; xent: 0.63; lr: 0.00010; 2080/2595 tok/s; 4390 sec
  946. [2019-02-04 19:03:45,265 INFO] Step 42700/102583; acc: 83.07; ppl: 1.71; xent: 0.54; lr: 0.00010; 2123/2713 tok/s; 4395 sec
  947. [2019-02-04 19:03:50,333 INFO] Step 42750/102583; acc: 79.72; ppl: 1.97; xent: 0.68; lr: 0.00010; 1993/2496 tok/s; 4400 sec
  948. [2019-02-04 19:03:55,166 INFO] Step 42800/102583; acc: 81.99; ppl: 1.83; xent: 0.60; lr: 0.00010; 2098/2620 tok/s; 4405 sec
  949. [2019-02-04 19:04:00,086 INFO] Step 42850/102583; acc: 80.73; ppl: 1.84; xent: 0.61; lr: 0.00010; 2016/2753 tok/s; 4409 sec
  950. [2019-02-04 19:04:05,081 INFO] Step 42900/102583; acc: 83.22; ppl: 1.75; xent: 0.56; lr: 0.00010; 2146/2537 tok/s; 4414 sec
  951. [2019-02-04 19:04:10,200 INFO] Step 42950/102583; acc: 80.79; ppl: 1.85; xent: 0.61; lr: 0.00010; 2055/2684 tok/s; 4420 sec
  952. [2019-02-04 19:04:14,837 INFO] Step 43000/102583; acc: 81.86; ppl: 1.83; xent: 0.61; lr: 0.00010; 2153/2521 tok/s; 4424 sec
  953. [2019-02-04 19:04:19,850 INFO] Step 43050/102583; acc: 81.90; ppl: 1.76; xent: 0.57; lr: 0.00010; 1881/2729 tok/s; 4429 sec
  954. [2019-02-04 19:04:25,170 INFO] Step 43100/102583; acc: 80.99; ppl: 1.85; xent: 0.61; lr: 0.00010; 1972/2486 tok/s; 4435 sec
  955. [2019-02-04 19:04:30,167 INFO] Step 43150/102583; acc: 82.26; ppl: 1.79; xent: 0.58; lr: 0.00010; 2179/2576 tok/s; 4440 sec
  956. [2019-02-04 19:04:35,441 INFO] Step 43200/102583; acc: 79.87; ppl: 1.88; xent: 0.63; lr: 0.00010; 1906/2670 tok/s; 4445 sec
  957. [2019-02-04 19:04:40,213 INFO] Step 43250/102583; acc: 82.59; ppl: 1.79; xent: 0.58; lr: 0.00010; 1987/2648 tok/s; 4450 sec
  958. [2019-02-04 19:04:45,229 INFO] Step 43300/102583; acc: 79.43; ppl: 1.92; xent: 0.65; lr: 0.00010; 2062/2562 tok/s; 4455 sec
  959. [2019-02-04 19:04:49,845 INFO] Step 43350/102583; acc: 81.51; ppl: 1.85; xent: 0.62; lr: 0.00010; 2257/2529 tok/s; 4459 sec
  960. [2019-02-04 19:04:54,829 INFO] Step 43400/102583; acc: 80.81; ppl: 1.85; xent: 0.61; lr: 0.00010; 1954/2783 tok/s; 4464 sec
  961. [2019-02-04 19:04:59,971 INFO] Step 43450/102583; acc: 82.52; ppl: 1.77; xent: 0.57; lr: 0.00010; 2060/2661 tok/s; 4469 sec
  962. [2019-02-04 19:05:05,252 INFO] Step 43500/102583; acc: 82.24; ppl: 1.79; xent: 0.58; lr: 0.00010; 2096/2292 tok/s; 4475 sec
  963. [2019-02-04 19:05:10,049 INFO] Step 43550/102583; acc: 80.12; ppl: 1.87; xent: 0.62; lr: 0.00010; 1902/2890 tok/s; 4479 sec
  964. [2019-02-04 19:05:14,857 INFO] Step 43600/102583; acc: 80.86; ppl: 1.86; xent: 0.62; lr: 0.00010; 1891/2700 tok/s; 4484 sec
  965. [2019-02-04 19:05:19,902 INFO] Step 43650/102583; acc: 83.30; ppl: 1.73; xent: 0.55; lr: 0.00010; 2260/2514 tok/s; 4489 sec
  966. [2019-02-04 19:05:24,678 INFO] Step 43700/102583; acc: 79.91; ppl: 1.94; xent: 0.66; lr: 0.00010; 1863/2669 tok/s; 4494 sec
  967. [2019-02-04 19:05:29,855 INFO] Step 43750/102583; acc: 82.23; ppl: 1.77; xent: 0.57; lr: 0.00010; 2124/2522 tok/s; 4499 sec
  968. [2019-02-04 19:05:34,859 INFO] Step 43800/102583; acc: 81.18; ppl: 1.84; xent: 0.61; lr: 0.00010; 2091/2696 tok/s; 4504 sec
  969. [2019-02-04 19:05:39,681 INFO] Step 43850/102583; acc: 81.03; ppl: 1.85; xent: 0.61; lr: 0.00010; 1972/2646 tok/s; 4509 sec
  970. [2019-02-04 19:05:44,650 INFO] Step 43900/102583; acc: 81.79; ppl: 1.81; xent: 0.59; lr: 0.00010; 2001/2719 tok/s; 4514 sec
  971. [2019-02-04 19:05:49,707 INFO] Step 43950/102583; acc: 82.86; ppl: 1.76; xent: 0.56; lr: 0.00010; 2206/2572 tok/s; 4519 sec
  972. [2019-02-04 19:05:54,648 INFO] Step 44000/102583; acc: 79.19; ppl: 1.96; xent: 0.67; lr: 0.00010; 1806/2509 tok/s; 4524 sec
  973. [2019-02-04 19:05:59,576 INFO] Step 44050/102583; acc: 82.29; ppl: 1.77; xent: 0.57; lr: 0.00010; 2223/2665 tok/s; 4529 sec
  974. [2019-02-04 19:06:04,473 INFO] Step 44100/102583; acc: 80.66; ppl: 1.89; xent: 0.64; lr: 0.00010; 1938/2598 tok/s; 4534 sec
  975. [2019-02-04 19:06:09,646 INFO] Step 44150/102583; acc: 83.56; ppl: 1.73; xent: 0.55; lr: 0.00010; 2182/2500 tok/s; 4539 sec
  976. [2019-02-04 19:06:14,353 INFO] Step 44200/102583; acc: 79.86; ppl: 1.94; xent: 0.66; lr: 0.00010; 1881/2731 tok/s; 4544 sec
  977. [2019-02-04 19:06:19,569 INFO] Step 44250/102583; acc: 82.06; ppl: 1.78; xent: 0.58; lr: 0.00010; 2045/2601 tok/s; 4549 sec
  978. [2019-02-04 19:06:24,281 INFO] Step 44300/102583; acc: 80.95; ppl: 1.90; xent: 0.64; lr: 0.00010; 2020/2618 tok/s; 4554 sec
  979. [2019-02-04 19:06:29,368 INFO] Step 44350/102583; acc: 83.44; ppl: 1.70; xent: 0.53; lr: 0.00010; 2043/2593 tok/s; 4559 sec
  980. [2019-02-04 19:06:34,172 INFO] Step 44400/102583; acc: 80.33; ppl: 1.90; xent: 0.64; lr: 0.00010; 2182/2554 tok/s; 4564 sec
  981. [2019-02-04 19:06:39,232 INFO] Step 44450/102583; acc: 82.73; ppl: 1.74; xent: 0.55; lr: 0.00010; 2102/2582 tok/s; 4569 sec
  982. [2019-02-04 19:06:44,326 INFO] Step 44500/102583; acc: 80.82; ppl: 1.82; xent: 0.60; lr: 0.00010; 1910/2760 tok/s; 4574 sec
  983. [2019-02-04 19:06:49,278 INFO] Step 44550/102583; acc: 81.46; ppl: 1.85; xent: 0.62; lr: 0.00010; 2030/2627 tok/s; 4579 sec
  984. [2019-02-04 19:06:54,067 INFO] Step 44600/102583; acc: 83.64; ppl: 1.72; xent: 0.54; lr: 0.00010; 2152/2578 tok/s; 4583 sec
  985. [2019-02-04 19:06:59,038 INFO] Step 44650/102583; acc: 81.24; ppl: 1.83; xent: 0.60; lr: 0.00010; 2039/2650 tok/s; 4588 sec
  986. [2019-02-04 19:07:03,870 INFO] Step 44700/102583; acc: 79.74; ppl: 1.90; xent: 0.64; lr: 0.00010; 2069/2657 tok/s; 4593 sec
  987. [2019-02-04 19:07:08,903 INFO] Step 44750/102583; acc: 82.17; ppl: 1.81; xent: 0.59; lr: 0.00010; 1936/2657 tok/s; 4598 sec
  988. [2019-02-04 19:07:13,578 INFO] Step 44800/102583; acc: 82.19; ppl: 1.79; xent: 0.58; lr: 0.00010; 2125/2639 tok/s; 4603 sec
  989. [2019-02-04 19:07:18,894 INFO] Step 44850/102583; acc: 82.07; ppl: 1.77; xent: 0.57; lr: 0.00010; 2053/2582 tok/s; 4608 sec
  990. [2019-02-04 19:07:24,053 INFO] Step 44900/102583; acc: 80.10; ppl: 1.93; xent: 0.65; lr: 0.00010; 1849/2728 tok/s; 4613 sec
  991. [2019-02-04 19:07:28,744 INFO] Step 44950/102583; acc: 81.30; ppl: 1.86; xent: 0.62; lr: 0.00010; 2139/2550 tok/s; 4618 sec
  992. [2019-02-04 19:07:33,648 INFO] Step 45000/102583; acc: 82.40; ppl: 1.75; xent: 0.56; lr: 0.00010; 2008/2755 tok/s; 4623 sec
  993. [2019-02-04 19:07:33,650 INFO] Saving checkpoint models/ted1_step_45000.pt
  994. [2019-02-04 19:07:44,361 INFO] Step 45050/102583; acc: 82.21; ppl: 1.80; xent: 0.59; lr: 0.00010; 931/1226 tok/s; 4634 sec
  995. [2019-02-04 19:07:49,252 INFO] Step 45100/102583; acc: 81.06; ppl: 1.84; xent: 0.61; lr: 0.00010; 2179/2643 tok/s; 4639 sec
  996. [2019-02-04 19:07:54,297 INFO] Step 45150/102583; acc: 81.70; ppl: 1.83; xent: 0.60; lr: 0.00010; 2155/2565 tok/s; 4644 sec
  997. [2019-02-04 19:07:59,707 INFO] Step 45200/102583; acc: 83.20; ppl: 1.74; xent: 0.55; lr: 0.00010; 2028/2510 tok/s; 4649 sec
  998. [2019-02-04 19:08:04,136 INFO] Step 45250/102583; acc: 80.15; ppl: 1.91; xent: 0.65; lr: 0.00010; 2105/2666 tok/s; 4654 sec
  999. [2019-02-04 19:08:08,877 INFO] Step 45300/102583; acc: 81.03; ppl: 1.83; xent: 0.60; lr: 0.00010; 2052/2587 tok/s; 4658 sec
  1000. [2019-02-04 19:08:14,222 INFO] Step 45350/102583; acc: 82.37; ppl: 1.80; xent: 0.59; lr: 0.00010; 1973/2576 tok/s; 4664 sec
  1001. [2019-02-04 19:08:19,402 INFO] Step 45400/102583; acc: 80.78; ppl: 1.81; xent: 0.60; lr: 0.00010; 1923/2647 tok/s; 4669 sec
  1002. [2019-02-04 19:08:26,453 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1003. [2019-02-04 19:08:30,701 INFO] Step 45450/102583; acc: 81.84; ppl: 1.82; xent: 0.60; lr: 0.00010; 955/1147 tok/s; 4680 sec
  1004. [2019-02-04 19:08:35,245 INFO] Step 45500/102583; acc: 80.33; ppl: 1.93; xent: 0.66; lr: 0.00010; 2100/2545 tok/s; 4685 sec
  1005. [2019-02-04 19:08:40,207 INFO] Step 45550/102583; acc: 82.83; ppl: 1.75; xent: 0.56; lr: 0.00010; 1985/2776 tok/s; 4690 sec
  1006. [2019-02-04 19:08:45,456 INFO] Step 45600/102583; acc: 81.74; ppl: 1.81; xent: 0.59; lr: 0.00010; 1984/2643 tok/s; 4695 sec
  1007. [2019-02-04 19:08:50,165 INFO] Step 45650/102583; acc: 81.75; ppl: 1.82; xent: 0.60; lr: 0.00010; 2109/2820 tok/s; 4700 sec
  1008. [2019-02-04 19:08:55,305 INFO] Step 45700/102583; acc: 81.64; ppl: 1.78; xent: 0.58; lr: 0.00010; 2098/2514 tok/s; 4705 sec
  1009. [2019-02-04 19:09:00,171 INFO] Step 45750/102583; acc: 82.44; ppl: 1.78; xent: 0.58; lr: 0.00010; 2176/2612 tok/s; 4710 sec
  1010. [2019-02-04 19:09:04,986 INFO] Step 45800/102583; acc: 80.77; ppl: 1.88; xent: 0.63; lr: 0.00010; 2138/2646 tok/s; 4714 sec
  1011. [2019-02-04 19:09:10,226 INFO] Step 45850/102583; acc: 80.16; ppl: 1.88; xent: 0.63; lr: 0.00010; 1768/2662 tok/s; 4720 sec
  1012. [2019-02-04 19:09:15,101 INFO] Step 45900/102583; acc: 81.94; ppl: 1.82; xent: 0.60; lr: 0.00010; 2238/2604 tok/s; 4724 sec
  1013. [2019-02-04 19:09:20,218 INFO] Step 45950/102583; acc: 82.56; ppl: 1.72; xent: 0.54; lr: 0.00010; 1908/2738 tok/s; 4730 sec
  1014. [2019-02-04 19:09:25,077 INFO] Step 46000/102583; acc: 80.56; ppl: 1.91; xent: 0.65; lr: 0.00010; 2186/2515 tok/s; 4734 sec
  1015. [2019-02-04 19:09:29,911 INFO] Step 46050/102583; acc: 81.89; ppl: 1.82; xent: 0.60; lr: 0.00010; 2060/2621 tok/s; 4739 sec
  1016. [2019-02-04 19:09:34,732 INFO] Step 46100/102583; acc: 80.00; ppl: 1.89; xent: 0.63; lr: 0.00010; 1987/2755 tok/s; 4744 sec
  1017. [2019-02-04 19:09:39,960 INFO] Step 46150/102583; acc: 83.68; ppl: 1.72; xent: 0.54; lr: 0.00010; 2220/2507 tok/s; 4749 sec
  1018. [2019-02-04 19:09:44,974 INFO] Step 46200/102583; acc: 80.61; ppl: 1.86; xent: 0.62; lr: 0.00010; 1959/2735 tok/s; 4754 sec
  1019. [2019-02-04 19:09:49,510 INFO] Step 46250/102583; acc: 81.72; ppl: 1.84; xent: 0.61; lr: 0.00010; 2116/2517 tok/s; 4759 sec
  1020. [2019-02-04 19:09:54,658 INFO] Step 46300/102583; acc: 82.24; ppl: 1.75; xent: 0.56; lr: 0.00010; 1891/2689 tok/s; 4764 sec
  1021. [2019-02-04 19:09:59,769 INFO] Step 46350/102583; acc: 81.20; ppl: 1.84; xent: 0.61; lr: 0.00010; 2054/2452 tok/s; 4769 sec
  1022. [2019-02-04 19:10:04,832 INFO] Step 46400/102583; acc: 81.51; ppl: 1.82; xent: 0.60; lr: 0.00010; 2112/2618 tok/s; 4774 sec
  1023. [2019-02-04 19:10:10,207 INFO] Step 46450/102583; acc: 81.00; ppl: 1.82; xent: 0.60; lr: 0.00010; 1866/2730 tok/s; 4780 sec
  1024. [2019-02-04 19:10:14,908 INFO] Step 46500/102583; acc: 82.80; ppl: 1.77; xent: 0.57; lr: 0.00010; 2099/2573 tok/s; 4784 sec
  1025. [2019-02-04 19:10:19,944 INFO] Step 46550/102583; acc: 79.70; ppl: 1.91; xent: 0.65; lr: 0.00010; 2113/2559 tok/s; 4789 sec
  1026. [2019-02-04 19:10:24,705 INFO] Step 46600/102583; acc: 81.28; ppl: 1.86; xent: 0.62; lr: 0.00010; 2108/2648 tok/s; 4794 sec
  1027. [2019-02-04 19:10:29,509 INFO] Step 46650/102583; acc: 80.36; ppl: 1.88; xent: 0.63; lr: 0.00010; 2040/2656 tok/s; 4799 sec
  1028. [2019-02-04 19:10:34,837 INFO] Step 46700/102583; acc: 82.70; ppl: 1.77; xent: 0.57; lr: 0.00010; 1911/2592 tok/s; 4804 sec
  1029. [2019-02-04 19:10:39,972 INFO] Step 46750/102583; acc: 82.32; ppl: 1.78; xent: 0.57; lr: 0.00010; 2186/2388 tok/s; 4809 sec
  1030. [2019-02-04 19:10:44,745 INFO] Step 46800/102583; acc: 80.26; ppl: 1.86; xent: 0.62; lr: 0.00010; 1938/2872 tok/s; 4814 sec
  1031. [2019-02-04 19:10:49,587 INFO] Step 46850/102583; acc: 81.01; ppl: 1.84; xent: 0.61; lr: 0.00010; 1948/2701 tok/s; 4819 sec
  1032. [2019-02-04 19:10:54,622 INFO] Step 46900/102583; acc: 83.03; ppl: 1.75; xent: 0.56; lr: 0.00010; 2206/2486 tok/s; 4824 sec
  1033. [2019-02-04 19:10:59,417 INFO] Step 46950/102583; acc: 80.41; ppl: 1.91; xent: 0.65; lr: 0.00010; 1805/2672 tok/s; 4829 sec
  1034. [2019-02-04 19:11:04,541 INFO] Step 47000/102583; acc: 82.78; ppl: 1.73; xent: 0.55; lr: 0.00010; 2176/2574 tok/s; 4834 sec
  1035. [2019-02-04 19:11:09,521 INFO] Step 47050/102583; acc: 80.69; ppl: 1.86; xent: 0.62; lr: 0.00010; 2075/2684 tok/s; 4839 sec
  1036. [2019-02-04 19:11:14,422 INFO] Step 47100/102583; acc: 81.36; ppl: 1.83; xent: 0.61; lr: 0.00010; 2048/2609 tok/s; 4844 sec
  1037. [2019-02-04 19:11:19,367 INFO] Step 47150/102583; acc: 81.23; ppl: 1.84; xent: 0.61; lr: 0.00010; 1856/2710 tok/s; 4849 sec
  1038. [2019-02-04 19:11:24,391 INFO] Step 47200/102583; acc: 82.79; ppl: 1.76; xent: 0.56; lr: 0.00010; 2267/2532 tok/s; 4854 sec
  1039. [2019-02-04 19:11:29,484 INFO] Step 47250/102583; acc: 80.12; ppl: 1.89; xent: 0.64; lr: 0.00010; 1831/2676 tok/s; 4859 sec
  1040. [2019-02-04 19:11:34,306 INFO] Step 47300/102583; acc: 82.16; ppl: 1.79; xent: 0.58; lr: 0.00010; 2255/2510 tok/s; 4864 sec
  1041. [2019-02-04 19:11:39,184 INFO] Step 47350/102583; acc: 80.79; ppl: 1.87; xent: 0.63; lr: 0.00010; 1963/2666 tok/s; 4869 sec
  1042. [2019-02-04 19:11:44,396 INFO] Step 47400/102583; acc: 82.99; ppl: 1.76; xent: 0.57; lr: 0.00010; 2050/2449 tok/s; 4874 sec
  1043. [2019-02-04 19:11:49,177 INFO] Step 47450/102583; acc: 80.80; ppl: 1.87; xent: 0.63; lr: 0.00010; 2040/2755 tok/s; 4879 sec
  1044. [2019-02-04 19:11:54,187 INFO] Step 47500/102583; acc: 82.07; ppl: 1.79; xent: 0.58; lr: 0.00010; 1990/2622 tok/s; 4884 sec
  1045. [2019-02-04 19:11:58,967 INFO] Step 47550/102583; acc: 80.99; ppl: 1.89; xent: 0.64; lr: 0.00010; 2018/2649 tok/s; 4888 sec
  1046. [2019-02-04 19:12:04,172 INFO] Step 47600/102583; acc: 83.83; ppl: 1.68; xent: 0.52; lr: 0.00010; 2090/2539 tok/s; 4894 sec
  1047. [2019-02-04 19:12:08,755 INFO] Step 47650/102583; acc: 79.52; ppl: 1.95; xent: 0.67; lr: 0.00010; 2153/2547 tok/s; 4898 sec
  1048. [2019-02-04 19:12:13,895 INFO] Step 47700/102583; acc: 82.71; ppl: 1.74; xent: 0.55; lr: 0.00010; 2069/2627 tok/s; 4903 sec
  1049. [2019-02-04 19:12:19,065 INFO] Step 47750/102583; acc: 81.34; ppl: 1.79; xent: 0.58; lr: 0.00010; 1884/2765 tok/s; 4908 sec
  1050. [2019-02-04 19:12:23,962 INFO] Step 47800/102583; acc: 81.11; ppl: 1.86; xent: 0.62; lr: 0.00010; 2090/2614 tok/s; 4913 sec
  1051. [2019-02-04 19:12:28,777 INFO] Step 47850/102583; acc: 83.17; ppl: 1.74; xent: 0.55; lr: 0.00010; 2102/2564 tok/s; 4918 sec
  1052. [2019-02-04 19:12:33,781 INFO] Step 47900/102583; acc: 81.29; ppl: 1.82; xent: 0.60; lr: 0.00010; 2087/2689 tok/s; 4923 sec
  1053. [2019-02-04 19:12:38,535 INFO] Step 47950/102583; acc: 80.67; ppl: 1.88; xent: 0.63; lr: 0.00010; 2023/2640 tok/s; 4928 sec
  1054. [2019-02-04 19:12:43,540 INFO] Step 48000/102583; acc: 82.54; ppl: 1.78; xent: 0.58; lr: 0.00010; 1985/2624 tok/s; 4933 sec
  1055. [2019-02-04 19:12:48,317 INFO] Step 48050/102583; acc: 82.00; ppl: 1.79; xent: 0.58; lr: 0.00010; 2151/2632 tok/s; 4938 sec
  1056. [2019-02-04 19:12:53,557 INFO] Step 48100/102583; acc: 81.58; ppl: 1.81; xent: 0.59; lr: 0.00010; 1993/2619 tok/s; 4943 sec
  1057. [2019-02-04 19:12:58,741 INFO] Step 48150/102583; acc: 80.48; ppl: 1.90; xent: 0.64; lr: 0.00010; 1903/2669 tok/s; 4948 sec
  1058. [2019-02-04 19:13:03,486 INFO] Step 48200/102583; acc: 81.29; ppl: 1.85; xent: 0.62; lr: 0.00010; 2092/2555 tok/s; 4953 sec
  1059. [2019-02-04 19:13:08,291 INFO] Step 48250/102583; acc: 82.89; ppl: 1.75; xent: 0.56; lr: 0.00010; 1976/2817 tok/s; 4958 sec
  1060. [2019-02-04 19:13:13,620 INFO] Step 48300/102583; acc: 82.99; ppl: 1.73; xent: 0.55; lr: 0.00010; 2081/2528 tok/s; 4963 sec
  1061. [2019-02-04 19:13:18,415 INFO] Step 48350/102583; acc: 80.83; ppl: 1.84; xent: 0.61; lr: 0.00010; 2102/2688 tok/s; 4968 sec
  1062. [2019-02-04 19:13:23,281 INFO] Step 48400/102583; acc: 81.39; ppl: 1.86; xent: 0.62; lr: 0.00010; 2192/2514 tok/s; 4973 sec
  1063. [2019-02-04 19:13:28,697 INFO] Step 48450/102583; acc: 83.09; ppl: 1.76; xent: 0.56; lr: 0.00010; 1990/2532 tok/s; 4978 sec
  1064. [2019-02-04 19:13:33,265 INFO] Step 48500/102583; acc: 80.12; ppl: 1.91; xent: 0.65; lr: 0.00010; 2080/2659 tok/s; 4983 sec
  1065. [2019-02-04 19:13:38,074 INFO] Step 48550/102583; acc: 81.60; ppl: 1.81; xent: 0.59; lr: 0.00010; 2064/2585 tok/s; 4987 sec
  1066. [2019-02-04 19:13:43,359 INFO] Step 48600/102583; acc: 81.99; ppl: 1.80; xent: 0.59; lr: 0.00010; 1936/2649 tok/s; 4993 sec
  1067. [2019-02-04 19:13:48,566 INFO] Step 48650/102583; acc: 80.81; ppl: 1.83; xent: 0.61; lr: 0.00010; 1875/2547 tok/s; 4998 sec
  1068. [2019-02-04 19:13:54,729 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1069. [2019-02-04 19:13:59,577 INFO] Step 48700/102583; acc: 82.56; ppl: 1.77; xent: 0.57; lr: 0.00010; 999/1167 tok/s; 5009 sec
  1070. [2019-02-04 19:14:04,184 INFO] Step 48750/102583; acc: 80.99; ppl: 1.90; xent: 0.64; lr: 0.00010; 2147/2565 tok/s; 5014 sec
  1071. [2019-02-04 19:14:09,356 INFO] Step 48800/102583; acc: 82.97; ppl: 1.73; xent: 0.55; lr: 0.00010; 1988/2800 tok/s; 5019 sec
  1072. [2019-02-04 19:14:14,278 INFO] Step 48850/102583; acc: 80.55; ppl: 1.88; xent: 0.63; lr: 0.00010; 1983/2640 tok/s; 5024 sec
  1073. [2019-02-04 19:14:19,095 INFO] Step 48900/102583; acc: 82.78; ppl: 1.78; xent: 0.58; lr: 0.00010; 2038/2803 tok/s; 5028 sec
  1074. [2019-02-04 19:14:24,107 INFO] Step 48950/102583; acc: 81.90; ppl: 1.79; xent: 0.58; lr: 0.00010; 2161/2487 tok/s; 5033 sec
  1075. [2019-02-04 19:14:29,059 INFO] Step 49000/102583; acc: 81.99; ppl: 1.79; xent: 0.58; lr: 0.00010; 2177/2642 tok/s; 5038 sec
  1076. [2019-02-04 19:14:33,877 INFO] Step 49050/102583; acc: 80.84; ppl: 1.88; xent: 0.63; lr: 0.00010; 2063/2607 tok/s; 5043 sec
  1077. [2019-02-04 19:14:39,054 INFO] Step 49100/102583; acc: 80.35; ppl: 1.87; xent: 0.62; lr: 0.00010; 1848/2708 tok/s; 5048 sec
  1078. [2019-02-04 19:14:43,895 INFO] Step 49150/102583; acc: 81.35; ppl: 1.85; xent: 0.61; lr: 0.00010; 2095/2634 tok/s; 5053 sec
  1079. [2019-02-04 19:14:49,019 INFO] Step 49200/102583; acc: 82.38; ppl: 1.75; xent: 0.56; lr: 0.00010; 1943/2742 tok/s; 5058 sec
  1080. [2019-02-04 19:14:53,886 INFO] Step 49250/102583; acc: 81.43; ppl: 1.87; xent: 0.62; lr: 0.00010; 2261/2486 tok/s; 5063 sec
  1081. [2019-02-04 19:14:58,717 INFO] Step 49300/102583; acc: 80.68; ppl: 1.88; xent: 0.63; lr: 0.00010; 2035/2639 tok/s; 5068 sec
  1082. [2019-02-04 19:15:03,602 INFO] Step 49350/102583; acc: 80.53; ppl: 1.87; xent: 0.63; lr: 0.00010; 1931/2656 tok/s; 5073 sec
  1083. [2019-02-04 19:15:08,725 INFO] Step 49400/102583; acc: 84.16; ppl: 1.69; xent: 0.53; lr: 0.00010; 2282/2559 tok/s; 5078 sec
  1084. [2019-02-04 19:15:13,729 INFO] Step 49450/102583; acc: 80.70; ppl: 1.86; xent: 0.62; lr: 0.00010; 1925/2742 tok/s; 5083 sec
  1085. [2019-02-04 19:15:18,600 INFO] Step 49500/102583; acc: 82.70; ppl: 1.78; xent: 0.58; lr: 0.00010; 2095/2414 tok/s; 5088 sec
  1086. [2019-02-04 19:15:23,423 INFO] Step 49550/102583; acc: 82.36; ppl: 1.76; xent: 0.56; lr: 0.00010; 1999/2809 tok/s; 5093 sec
  1087. [2019-02-04 19:15:28,520 INFO] Step 49600/102583; acc: 80.75; ppl: 1.87; xent: 0.63; lr: 0.00010; 2062/2451 tok/s; 5098 sec
  1088. [2019-02-04 19:15:33,556 INFO] Step 49650/102583; acc: 81.36; ppl: 1.83; xent: 0.60; lr: 0.00010; 2022/2620 tok/s; 5103 sec
  1089. [2019-02-04 19:15:38,970 INFO] Step 49700/102583; acc: 81.59; ppl: 1.79; xent: 0.58; lr: 0.00010; 1912/2732 tok/s; 5108 sec
  1090. [2019-02-04 19:15:43,618 INFO] Step 49750/102583; acc: 81.53; ppl: 1.83; xent: 0.60; lr: 0.00010; 2096/2542 tok/s; 5113 sec
  1091. [2019-02-04 19:15:48,661 INFO] Step 49800/102583; acc: 80.80; ppl: 1.87; xent: 0.63; lr: 0.00010; 2133/2563 tok/s; 5118 sec
  1092. [2019-02-04 19:15:53,607 INFO] Step 49850/102583; acc: 80.84; ppl: 1.85; xent: 0.61; lr: 0.00010; 1994/2701 tok/s; 5123 sec
  1093. [2019-02-04 19:15:58,309 INFO] Step 49900/102583; acc: 80.20; ppl: 1.89; xent: 0.64; lr: 0.00010; 2118/2579 tok/s; 5128 sec
  1094. [2019-02-04 19:16:03,568 INFO] Step 49950/102583; acc: 83.23; ppl: 1.75; xent: 0.56; lr: 0.00010; 1951/2611 tok/s; 5133 sec
  1095. [2019-02-04 19:16:08,689 INFO] Step 50000/102583; acc: 82.68; ppl: 1.76; xent: 0.57; lr: 0.00005; 2179/2400 tok/s; 5138 sec
  1096. [2019-02-04 19:16:08,709 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  1097. [2019-02-04 19:16:10,015 INFO] Validation perplexity: 1.75152
  1098. [2019-02-04 19:16:10,016 INFO] Validation accuracy: 82.9947
  1099. [2019-02-04 19:16:10,017 INFO] Saving checkpoint models/ted1_step_50000.pt
  1100. [2019-02-04 19:16:20,439 INFO] Step 50050/102583; acc: 80.12; ppl: 1.87; xent: 0.62; lr: 0.00005; 771/1174 tok/s; 5150 sec
  1101. [2019-02-04 19:16:25,497 INFO] Step 50100/102583; acc: 81.55; ppl: 1.80; xent: 0.59; lr: 0.00005; 1987/2623 tok/s; 5155 sec
  1102. [2019-02-04 19:16:30,383 INFO] Step 50150/102583; acc: 82.49; ppl: 1.78; xent: 0.58; lr: 0.00005; 2205/2507 tok/s; 5160 sec
  1103. [2019-02-04 19:16:35,072 INFO] Step 50200/102583; acc: 80.73; ppl: 1.89; xent: 0.64; lr: 0.00005; 1812/2673 tok/s; 5164 sec
  1104. [2019-02-04 19:16:40,405 INFO] Step 50250/102583; acc: 82.94; ppl: 1.71; xent: 0.54; lr: 0.00005; 2195/2618 tok/s; 5170 sec
  1105. [2019-02-04 19:16:45,367 INFO] Step 50300/102583; acc: 80.51; ppl: 1.87; xent: 0.63; lr: 0.00005; 1979/2711 tok/s; 5175 sec
  1106. [2019-02-04 19:16:50,311 INFO] Step 50350/102583; acc: 82.60; ppl: 1.79; xent: 0.58; lr: 0.00005; 2062/2517 tok/s; 5180 sec
  1107. [2019-02-04 19:16:55,368 INFO] Step 50400/102583; acc: 82.41; ppl: 1.78; xent: 0.58; lr: 0.00005; 1928/2719 tok/s; 5185 sec
  1108. [2019-02-04 19:17:00,212 INFO] Step 50450/102583; acc: 82.03; ppl: 1.80; xent: 0.59; lr: 0.00005; 2225/2568 tok/s; 5190 sec
  1109. [2019-02-04 19:17:05,328 INFO] Step 50500/102583; acc: 80.99; ppl: 1.87; xent: 0.62; lr: 0.00005; 1823/2688 tok/s; 5195 sec
  1110. [2019-02-04 19:17:10,174 INFO] Step 50550/102583; acc: 82.23; ppl: 1.77; xent: 0.57; lr: 0.00005; 2231/2541 tok/s; 5200 sec
  1111. [2019-02-04 19:17:15,116 INFO] Step 50600/102583; acc: 81.85; ppl: 1.81; xent: 0.59; lr: 0.00005; 2016/2609 tok/s; 5205 sec
  1112. [2019-02-04 19:17:20,280 INFO] Step 50650/102583; acc: 82.39; ppl: 1.79; xent: 0.58; lr: 0.00005; 1957/2509 tok/s; 5210 sec
  1113. [2019-02-04 19:17:25,033 INFO] Step 50700/102583; acc: 81.12; ppl: 1.85; xent: 0.62; lr: 0.00005; 2042/2680 tok/s; 5214 sec
  1114. [2019-02-04 19:17:30,010 INFO] Step 50750/102583; acc: 82.97; ppl: 1.75; xent: 0.56; lr: 0.00005; 2063/2658 tok/s; 5219 sec
  1115. [2019-02-04 19:17:34,938 INFO] Step 50800/102583; acc: 81.25; ppl: 1.87; xent: 0.63; lr: 0.00005; 2022/2585 tok/s; 5224 sec
  1116. [2019-02-04 19:17:39,955 INFO] Step 50850/102583; acc: 83.41; ppl: 1.71; xent: 0.53; lr: 0.00005; 2143/2567 tok/s; 5229 sec
  1117. [2019-02-04 19:17:44,574 INFO] Step 50900/102583; acc: 81.17; ppl: 1.85; xent: 0.62; lr: 0.00005; 2109/2579 tok/s; 5234 sec
  1118. [2019-02-04 19:17:49,772 INFO] Step 50950/102583; acc: 82.48; ppl: 1.75; xent: 0.56; lr: 0.00005; 2047/2638 tok/s; 5239 sec
  1119. [2019-02-04 19:17:54,994 INFO] Step 51000/102583; acc: 81.76; ppl: 1.77; xent: 0.57; lr: 0.00005; 1944/2713 tok/s; 5244 sec
  1120. [2019-02-04 19:17:59,849 INFO] Step 51050/102583; acc: 81.29; ppl: 1.84; xent: 0.61; lr: 0.00005; 1997/2666 tok/s; 5249 sec
  1121. [2019-02-04 19:18:04,560 INFO] Step 51100/102583; acc: 83.54; ppl: 1.73; xent: 0.55; lr: 0.00005; 2203/2517 tok/s; 5254 sec
  1122. [2019-02-04 19:18:09,673 INFO] Step 51150/102583; acc: 81.12; ppl: 1.82; xent: 0.60; lr: 0.00005; 2060/2705 tok/s; 5259 sec
  1123. [2019-02-04 19:18:14,309 INFO] Step 51200/102583; acc: 81.18; ppl: 1.84; xent: 0.61; lr: 0.00005; 1973/2653 tok/s; 5264 sec
  1124. [2019-02-04 19:18:19,344 INFO] Step 51250/102583; acc: 82.67; ppl: 1.78; xent: 0.57; lr: 0.00005; 1986/2649 tok/s; 5269 sec
  1125. [2019-02-04 19:18:24,114 INFO] Step 51300/102583; acc: 83.40; ppl: 1.72; xent: 0.54; lr: 0.00005; 2222/2624 tok/s; 5274 sec
  1126. [2019-02-04 19:18:29,370 INFO] Step 51350/102583; acc: 81.04; ppl: 1.84; xent: 0.61; lr: 0.00005; 1926/2632 tok/s; 5279 sec
  1127. [2019-02-04 19:18:34,510 INFO] Step 51400/102583; acc: 80.99; ppl: 1.86; xent: 0.62; lr: 0.00005; 1956/2636 tok/s; 5284 sec
  1128. [2019-02-04 19:18:39,382 INFO] Step 51450/102583; acc: 82.59; ppl: 1.77; xent: 0.57; lr: 0.00005; 2119/2533 tok/s; 5289 sec
  1129. [2019-02-04 19:18:44,237 INFO] Step 51500/102583; acc: 83.07; ppl: 1.72; xent: 0.54; lr: 0.00005; 1925/2788 tok/s; 5294 sec
  1130. [2019-02-04 19:18:49,486 INFO] Step 51550/102583; acc: 82.88; ppl: 1.76; xent: 0.56; lr: 0.00005; 2135/2529 tok/s; 5299 sec
  1131. [2019-02-04 19:18:54,294 INFO] Step 51600/102583; acc: 81.62; ppl: 1.80; xent: 0.59; lr: 0.00005; 2060/2784 tok/s; 5304 sec
  1132. [2019-02-04 19:18:59,205 INFO] Step 51650/102583; acc: 81.34; ppl: 1.86; xent: 0.62; lr: 0.00005; 2182/2347 tok/s; 5309 sec
  1133. [2019-02-04 19:19:04,497 INFO] Step 51700/102583; acc: 82.85; ppl: 1.74; xent: 0.55; lr: 0.00005; 2028/2660 tok/s; 5314 sec
  1134. [2019-02-04 19:19:09,081 INFO] Step 51750/102583; acc: 80.40; ppl: 1.89; xent: 0.64; lr: 0.00005; 1989/2690 tok/s; 5318 sec
  1135. [2019-02-04 19:19:13,881 INFO] Step 51800/102583; acc: 82.17; ppl: 1.78; xent: 0.58; lr: 0.00005; 2122/2544 tok/s; 5323 sec
  1136. [2019-02-04 19:19:19,174 INFO] Step 51850/102583; acc: 81.65; ppl: 1.81; xent: 0.59; lr: 0.00005; 1862/2698 tok/s; 5329 sec
  1137. [2019-02-04 19:19:24,337 INFO] Step 51900/102583; acc: 82.27; ppl: 1.80; xent: 0.59; lr: 0.00005; 1953/2469 tok/s; 5334 sec
  1138. [2019-02-04 19:19:29,681 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1139. [2019-02-04 19:19:35,019 INFO] Step 51950/102583; acc: 82.46; ppl: 1.76; xent: 0.57; lr: 0.00005; 1017/1205 tok/s; 5344 sec
  1140. [2019-02-04 19:19:39,827 INFO] Step 52000/102583; acc: 81.70; ppl: 1.83; xent: 0.60; lr: 0.00005; 2057/2700 tok/s; 5349 sec
  1141. [2019-02-04 19:19:44,855 INFO] Step 52050/102583; acc: 83.18; ppl: 1.73; xent: 0.55; lr: 0.00005; 2032/2721 tok/s; 5354 sec
  1142. [2019-02-04 19:19:49,749 INFO] Step 52100/102583; acc: 81.58; ppl: 1.84; xent: 0.61; lr: 0.00005; 2020/2588 tok/s; 5359 sec
  1143. [2019-02-04 19:19:54,593 INFO] Step 52150/102583; acc: 82.44; ppl: 1.78; xent: 0.58; lr: 0.00005; 2018/2762 tok/s; 5364 sec
  1144. [2019-02-04 19:19:59,659 INFO] Step 52200/102583; acc: 81.82; ppl: 1.78; xent: 0.58; lr: 0.00005; 2139/2526 tok/s; 5369 sec
  1145. [2019-02-04 19:20:04,473 INFO] Step 52250/102583; acc: 81.93; ppl: 1.80; xent: 0.59; lr: 0.00005; 2181/2643 tok/s; 5374 sec
  1146. [2019-02-04 19:20:09,898 INFO] Step 52300/102583; acc: 81.79; ppl: 1.82; xent: 0.60; lr: 0.00005; 2026/2662 tok/s; 5379 sec
  1147. [2019-02-04 19:20:15,122 INFO] Step 52350/102583; acc: 80.31; ppl: 1.88; xent: 0.63; lr: 0.00005; 1884/2673 tok/s; 5385 sec
  1148. [2019-02-04 19:20:19,838 INFO] Step 52400/102583; acc: 82.34; ppl: 1.77; xent: 0.57; lr: 0.00005; 2144/2623 tok/s; 5389 sec
  1149. [2019-02-04 19:20:25,034 INFO] Step 52450/102583; acc: 82.27; ppl: 1.75; xent: 0.56; lr: 0.00005; 1934/2721 tok/s; 5394 sec
  1150. [2019-02-04 19:20:29,889 INFO] Step 52500/102583; acc: 81.81; ppl: 1.84; xent: 0.61; lr: 0.00005; 2261/2558 tok/s; 5399 sec
  1151. [2019-02-04 19:20:34,875 INFO] Step 52550/102583; acc: 81.14; ppl: 1.83; xent: 0.60; lr: 0.00005; 2024/2589 tok/s; 5404 sec
  1152. [2019-02-04 19:20:39,558 INFO] Step 52600/102583; acc: 81.44; ppl: 1.84; xent: 0.61; lr: 0.00005; 2013/2644 tok/s; 5409 sec
  1153. [2019-02-04 19:20:44,748 INFO] Step 52650/102583; acc: 83.33; ppl: 1.71; xent: 0.54; lr: 0.00005; 2215/2627 tok/s; 5414 sec
  1154. [2019-02-04 19:20:49,654 INFO] Step 52700/102583; acc: 81.89; ppl: 1.79; xent: 0.58; lr: 0.00005; 1951/2695 tok/s; 5419 sec
  1155. [2019-02-04 19:20:54,697 INFO] Step 52750/102583; acc: 82.00; ppl: 1.78; xent: 0.58; lr: 0.00005; 2057/2453 tok/s; 5424 sec
  1156. [2019-02-04 19:20:59,523 INFO] Step 52800/102583; acc: 82.35; ppl: 1.78; xent: 0.58; lr: 0.00005; 2014/2698 tok/s; 5429 sec
  1157. [2019-02-04 19:21:04,744 INFO] Step 52850/102583; acc: 81.96; ppl: 1.79; xent: 0.58; lr: 0.00005; 2059/2439 tok/s; 5434 sec
  1158. [2019-02-04 19:21:09,660 INFO] Step 52900/102583; acc: 82.28; ppl: 1.78; xent: 0.58; lr: 0.00005; 2014/2747 tok/s; 5439 sec
  1159. [2019-02-04 19:21:14,893 INFO] Step 52950/102583; acc: 81.18; ppl: 1.81; xent: 0.59; lr: 0.00005; 1896/2721 tok/s; 5444 sec
  1160. [2019-02-04 19:21:19,701 INFO] Step 53000/102583; acc: 82.28; ppl: 1.79; xent: 0.58; lr: 0.00005; 2127/2512 tok/s; 5449 sec
  1161. [2019-02-04 19:21:24,675 INFO] Step 53050/102583; acc: 81.17; ppl: 1.84; xent: 0.61; lr: 0.00005; 2107/2554 tok/s; 5454 sec
  1162. [2019-02-04 19:21:29,525 INFO] Step 53100/102583; acc: 80.84; ppl: 1.84; xent: 0.61; lr: 0.00005; 1984/2729 tok/s; 5459 sec
  1163. [2019-02-04 19:21:34,195 INFO] Step 53150/102583; acc: 81.84; ppl: 1.85; xent: 0.61; lr: 0.00005; 2146/2579 tok/s; 5464 sec
  1164. [2019-02-04 19:21:39,550 INFO] Step 53200/102583; acc: 82.80; ppl: 1.73; xent: 0.55; lr: 0.00005; 1918/2640 tok/s; 5469 sec
  1165. [2019-02-04 19:21:44,914 INFO] Step 53250/102583; acc: 83.37; ppl: 1.69; xent: 0.53; lr: 0.00005; 2180/2462 tok/s; 5474 sec
  1166. [2019-02-04 19:21:49,613 INFO] Step 53300/102583; acc: 80.18; ppl: 1.88; xent: 0.63; lr: 0.00005; 1935/2789 tok/s; 5479 sec
  1167. [2019-02-04 19:21:54,786 INFO] Step 53350/102583; acc: 81.81; ppl: 1.79; xent: 0.58; lr: 0.00005; 2023/2617 tok/s; 5484 sec
  1168. [2019-02-04 19:21:59,508 INFO] Step 53400/102583; acc: 81.87; ppl: 1.81; xent: 0.59; lr: 0.00005; 2115/2600 tok/s; 5489 sec
  1169. [2019-02-04 19:22:04,234 INFO] Step 53450/102583; acc: 81.49; ppl: 1.86; xent: 0.62; lr: 0.00005; 1966/2618 tok/s; 5494 sec
  1170. [2019-02-04 19:22:09,587 INFO] Step 53500/102583; acc: 82.27; ppl: 1.73; xent: 0.55; lr: 0.00005; 2027/2562 tok/s; 5499 sec
  1171. [2019-02-04 19:22:14,460 INFO] Step 53550/102583; acc: 80.84; ppl: 1.84; xent: 0.61; lr: 0.00005; 2076/2802 tok/s; 5504 sec
  1172. [2019-02-04 19:22:19,448 INFO] Step 53600/102583; acc: 82.76; ppl: 1.76; xent: 0.57; lr: 0.00005; 2048/2535 tok/s; 5509 sec
  1173. [2019-02-04 19:22:24,536 INFO] Step 53650/102583; acc: 82.45; ppl: 1.78; xent: 0.58; lr: 0.00005; 1924/2748 tok/s; 5514 sec
  1174. [2019-02-04 19:22:29,369 INFO] Step 53700/102583; acc: 81.30; ppl: 1.82; xent: 0.60; lr: 0.00005; 2181/2525 tok/s; 5519 sec
  1175. [2019-02-04 19:22:34,371 INFO] Step 53750/102583; acc: 80.84; ppl: 1.84; xent: 0.61; lr: 0.00005; 1930/2702 tok/s; 5524 sec
  1176. [2019-02-04 19:22:39,226 INFO] Step 53800/102583; acc: 82.18; ppl: 1.78; xent: 0.58; lr: 0.00005; 2196/2561 tok/s; 5529 sec
  1177. [2019-02-04 19:22:44,454 INFO] Step 53850/102583; acc: 82.56; ppl: 1.77; xent: 0.57; lr: 0.00005; 2075/2583 tok/s; 5534 sec
  1178. [2019-02-04 19:22:49,431 INFO] Step 53900/102583; acc: 81.76; ppl: 1.82; xent: 0.60; lr: 0.00005; 1943/2564 tok/s; 5539 sec
  1179. [2019-02-04 19:22:54,255 INFO] Step 53950/102583; acc: 81.81; ppl: 1.82; xent: 0.60; lr: 0.00005; 2024/2693 tok/s; 5544 sec
  1180. [2019-02-04 19:22:59,389 INFO] Step 54000/102583; acc: 82.97; ppl: 1.75; xent: 0.56; lr: 0.00005; 2108/2544 tok/s; 5549 sec
  1181. [2019-02-04 19:23:04,204 INFO] Step 54050/102583; acc: 80.62; ppl: 1.87; xent: 0.63; lr: 0.00005; 1947/2715 tok/s; 5554 sec
  1182. [2019-02-04 19:23:09,240 INFO] Step 54100/102583; acc: 83.29; ppl: 1.72; xent: 0.54; lr: 0.00005; 2140/2577 tok/s; 5559 sec
  1183. [2019-02-04 19:23:13,984 INFO] Step 54150/102583; acc: 82.15; ppl: 1.78; xent: 0.58; lr: 0.00005; 2182/2590 tok/s; 5563 sec
  1184. [2019-02-04 19:23:19,201 INFO] Step 54200/102583; acc: 81.63; ppl: 1.80; xent: 0.59; lr: 0.00005; 1921/2610 tok/s; 5569 sec
  1185. [2019-02-04 19:23:24,281 INFO] Step 54250/102583; acc: 82.20; ppl: 1.77; xent: 0.57; lr: 0.00005; 2085/2697 tok/s; 5574 sec
  1186. [2019-02-04 19:23:29,095 INFO] Step 54300/102583; acc: 81.41; ppl: 1.84; xent: 0.61; lr: 0.00005; 2000/2641 tok/s; 5578 sec
  1187. [2019-02-04 19:23:34,058 INFO] Step 54350/102583; acc: 83.54; ppl: 1.71; xent: 0.54; lr: 0.00005; 2109/2558 tok/s; 5583 sec
  1188. [2019-02-04 19:23:38,965 INFO] Step 54400/102583; acc: 80.90; ppl: 1.84; xent: 0.61; lr: 0.00005; 2097/2683 tok/s; 5588 sec
  1189. [2019-02-04 19:23:43,739 INFO] Step 54450/102583; acc: 81.66; ppl: 1.82; xent: 0.60; lr: 0.00005; 1997/2653 tok/s; 5593 sec
  1190. [2019-02-04 19:23:49,061 INFO] Step 54500/102583; acc: 82.42; ppl: 1.77; xent: 0.57; lr: 0.00005; 2025/2635 tok/s; 5598 sec
  1191. [2019-02-04 19:23:53,866 INFO] Step 54550/102583; acc: 82.74; ppl: 1.75; xent: 0.56; lr: 0.00005; 2179/2579 tok/s; 5603 sec
  1192. [2019-02-04 19:23:58,833 INFO] Step 54600/102583; acc: 81.71; ppl: 1.83; xent: 0.61; lr: 0.00005; 1908/2688 tok/s; 5608 sec
  1193. [2019-02-04 19:24:03,994 INFO] Step 54650/102583; acc: 80.91; ppl: 1.85; xent: 0.62; lr: 0.00005; 1952/2646 tok/s; 5613 sec
  1194. [2019-02-04 19:24:08,990 INFO] Step 54700/102583; acc: 83.12; ppl: 1.74; xent: 0.55; lr: 0.00005; 2130/2604 tok/s; 5618 sec
  1195. [2019-02-04 19:24:13,806 INFO] Step 54750/102583; acc: 82.75; ppl: 1.73; xent: 0.55; lr: 0.00005; 1941/2753 tok/s; 5623 sec
  1196. [2019-02-04 19:24:19,145 INFO] Step 54800/102583; acc: 82.16; ppl: 1.78; xent: 0.58; lr: 0.00005; 2142/2490 tok/s; 5629 sec
  1197. [2019-02-04 19:24:23,945 INFO] Step 54850/102583; acc: 81.70; ppl: 1.79; xent: 0.58; lr: 0.00005; 1997/2770 tok/s; 5633 sec
  1198. [2019-02-04 19:24:28,860 INFO] Step 54900/102583; acc: 81.11; ppl: 1.87; xent: 0.62; lr: 0.00005; 2174/2392 tok/s; 5638 sec
  1199. [2019-02-04 19:24:34,194 INFO] Step 54950/102583; acc: 82.35; ppl: 1.76; xent: 0.56; lr: 0.00005; 2072/2672 tok/s; 5644 sec
  1200. [2019-02-04 19:24:38,834 INFO] Step 55000/102583; acc: 80.60; ppl: 1.87; xent: 0.63; lr: 0.00005; 1924/2719 tok/s; 5648 sec
  1201. [2019-02-04 19:24:38,927 INFO] Saving checkpoint models/ted1_step_55000.pt
  1202. [2019-02-04 19:24:51,121 INFO] Step 55050/102583; acc: 82.81; ppl: 1.76; xent: 0.57; lr: 0.00005; 830/1012 tok/s; 5661 sec
  1203. [2019-02-04 19:24:56,454 INFO] Step 55100/102583; acc: 81.65; ppl: 1.80; xent: 0.59; lr: 0.00005; 1950/2606 tok/s; 5666 sec
  1204. [2019-02-04 19:25:06,706 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1205. [2019-02-04 19:25:07,594 INFO] Step 55150/102583; acc: 81.74; ppl: 1.81; xent: 0.59; lr: 0.00005; 907/1136 tok/s; 5677 sec
  1206. [2019-02-04 19:25:12,667 INFO] Step 55200/102583; acc: 82.46; ppl: 1.77; xent: 0.57; lr: 0.00005; 2106/2575 tok/s; 5682 sec
  1207. [2019-02-04 19:25:17,431 INFO] Step 55250/102583; acc: 82.25; ppl: 1.79; xent: 0.58; lr: 0.00005; 2009/2695 tok/s; 5687 sec
  1208. [2019-02-04 19:25:22,684 INFO] Step 55300/102583; acc: 83.06; ppl: 1.73; xent: 0.55; lr: 0.00005; 2051/2661 tok/s; 5692 sec
  1209. [2019-02-04 19:25:27,449 INFO] Step 55350/102583; acc: 81.27; ppl: 1.86; xent: 0.62; lr: 0.00005; 2094/2642 tok/s; 5697 sec
  1210. [2019-02-04 19:25:32,495 INFO] Step 55400/102583; acc: 82.78; ppl: 1.74; xent: 0.56; lr: 0.00005; 2061/2713 tok/s; 5702 sec
  1211. [2019-02-04 19:25:37,365 INFO] Step 55450/102583; acc: 80.70; ppl: 1.85; xent: 0.62; lr: 0.00005; 2052/2591 tok/s; 5707 sec
  1212. [2019-02-04 19:25:42,226 INFO] Step 55500/102583; acc: 82.25; ppl: 1.79; xent: 0.58; lr: 0.00005; 2107/2690 tok/s; 5712 sec
  1213. [2019-02-04 19:25:47,252 INFO] Step 55550/102583; acc: 81.60; ppl: 1.81; xent: 0.59; lr: 0.00005; 2010/2702 tok/s; 5717 sec
  1214. [2019-02-04 19:25:52,379 INFO] Step 55600/102583; acc: 81.49; ppl: 1.82; xent: 0.60; lr: 0.00005; 2030/2570 tok/s; 5722 sec
  1215. [2019-02-04 19:25:57,169 INFO] Step 55650/102583; acc: 81.73; ppl: 1.79; xent: 0.58; lr: 0.00005; 2085/2655 tok/s; 5727 sec
  1216. [2019-02-04 19:26:02,240 INFO] Step 55700/102583; acc: 82.44; ppl: 1.77; xent: 0.57; lr: 0.00005; 1958/2683 tok/s; 5732 sec
  1217. [2019-02-04 19:26:07,056 INFO] Step 55750/102583; acc: 81.78; ppl: 1.83; xent: 0.60; lr: 0.00005; 2232/2544 tok/s; 5736 sec
  1218. [2019-02-04 19:26:12,072 INFO] Step 55800/102583; acc: 80.17; ppl: 1.86; xent: 0.62; lr: 0.00005; 2063/2592 tok/s; 5741 sec
  1219. [2019-02-04 19:26:16,943 INFO] Step 55850/102583; acc: 83.04; ppl: 1.76; xent: 0.57; lr: 0.00005; 2025/2670 tok/s; 5746 sec
  1220. [2019-02-04 19:26:22,005 INFO] Step 55900/102583; acc: 82.80; ppl: 1.74; xent: 0.55; lr: 0.00005; 2175/2653 tok/s; 5751 sec
  1221. [2019-02-04 19:26:26,981 INFO] Step 55950/102583; acc: 81.33; ppl: 1.82; xent: 0.60; lr: 0.00005; 1983/2601 tok/s; 5756 sec
  1222. [2019-02-04 19:26:32,008 INFO] Step 56000/102583; acc: 82.53; ppl: 1.75; xent: 0.56; lr: 0.00005; 2026/2550 tok/s; 5761 sec
  1223. [2019-02-04 19:26:37,074 INFO] Step 56050/102583; acc: 82.22; ppl: 1.77; xent: 0.57; lr: 0.00005; 1953/2652 tok/s; 5766 sec
  1224. [2019-02-04 19:26:42,095 INFO] Step 56100/102583; acc: 81.55; ppl: 1.83; xent: 0.60; lr: 0.00005; 2093/2486 tok/s; 5771 sec
  1225. [2019-02-04 19:26:46,991 INFO] Step 56150/102583; acc: 81.33; ppl: 1.81; xent: 0.59; lr: 0.00005; 2017/2738 tok/s; 5776 sec
  1226. [2019-02-04 19:26:52,210 INFO] Step 56200/102583; acc: 82.07; ppl: 1.78; xent: 0.57; lr: 0.00005; 1923/2614 tok/s; 5782 sec
  1227. [2019-02-04 19:26:57,170 INFO] Step 56250/102583; acc: 82.34; ppl: 1.79; xent: 0.58; lr: 0.00005; 2148/2542 tok/s; 5787 sec
  1228. [2019-02-04 19:27:02,165 INFO] Step 56300/102583; acc: 81.59; ppl: 1.81; xent: 0.59; lr: 0.00005; 2115/2616 tok/s; 5792 sec
  1229. [2019-02-04 19:27:06,959 INFO] Step 56350/102583; acc: 80.66; ppl: 1.85; xent: 0.61; lr: 0.00005; 1992/2659 tok/s; 5796 sec
  1230. [2019-02-04 19:27:11,716 INFO] Step 56400/102583; acc: 81.62; ppl: 1.82; xent: 0.60; lr: 0.00005; 2101/2690 tok/s; 5801 sec
  1231. [2019-02-04 19:27:16,966 INFO] Step 56450/102583; acc: 82.50; ppl: 1.77; xent: 0.57; lr: 0.00005; 2012/2514 tok/s; 5806 sec
  1232. [2019-02-04 19:27:22,203 INFO] Step 56500/102583; acc: 82.80; ppl: 1.74; xent: 0.55; lr: 0.00005; 2087/2559 tok/s; 5812 sec
  1233. [2019-02-04 19:27:26,940 INFO] Step 56550/102583; acc: 80.68; ppl: 1.84; xent: 0.61; lr: 0.00005; 1913/2712 tok/s; 5816 sec
  1234. [2019-02-04 19:27:32,065 INFO] Step 56600/102583; acc: 81.97; ppl: 1.78; xent: 0.58; lr: 0.00005; 2077/2683 tok/s; 5821 sec
  1235. [2019-02-04 19:27:36,813 INFO] Step 56650/102583; acc: 81.62; ppl: 1.83; xent: 0.60; lr: 0.00005; 2013/2535 tok/s; 5826 sec
  1236. [2019-02-04 19:27:41,617 INFO] Step 56700/102583; acc: 81.63; ppl: 1.83; xent: 0.60; lr: 0.00005; 1944/2695 tok/s; 5831 sec
  1237. [2019-02-04 19:27:46,746 INFO] Step 56750/102583; acc: 82.44; ppl: 1.74; xent: 0.56; lr: 0.00005; 2114/2508 tok/s; 5836 sec
  1238. [2019-02-04 19:27:51,670 INFO] Step 56800/102583; acc: 80.76; ppl: 1.84; xent: 0.61; lr: 0.00005; 2029/2843 tok/s; 5841 sec
  1239. [2019-02-04 19:27:56,670 INFO] Step 56850/102583; acc: 83.10; ppl: 1.76; xent: 0.56; lr: 0.00005; 2087/2508 tok/s; 5846 sec
  1240. [2019-02-04 19:28:01,821 INFO] Step 56900/102583; acc: 82.80; ppl: 1.77; xent: 0.57; lr: 0.00005; 1938/2662 tok/s; 5851 sec
  1241. [2019-02-04 19:28:06,473 INFO] Step 56950/102583; acc: 81.12; ppl: 1.85; xent: 0.61; lr: 0.00005; 2189/2565 tok/s; 5856 sec
  1242. [2019-02-04 19:28:11,572 INFO] Step 57000/102583; acc: 81.16; ppl: 1.82; xent: 0.60; lr: 0.00005; 1882/2695 tok/s; 5861 sec
  1243. [2019-02-04 19:28:16,488 INFO] Step 57050/102583; acc: 82.44; ppl: 1.77; xent: 0.57; lr: 0.00005; 2198/2530 tok/s; 5866 sec
  1244. [2019-02-04 19:28:21,731 INFO] Step 57100/102583; acc: 82.91; ppl: 1.75; xent: 0.56; lr: 0.00005; 2136/2593 tok/s; 5871 sec
  1245. [2019-02-04 19:28:26,505 INFO] Step 57150/102583; acc: 81.85; ppl: 1.83; xent: 0.60; lr: 0.00005; 1885/2581 tok/s; 5876 sec
  1246. [2019-02-04 19:28:31,375 INFO] Step 57200/102583; acc: 80.75; ppl: 1.89; xent: 0.63; lr: 0.00005; 1937/2664 tok/s; 5881 sec
  1247. [2019-02-04 19:28:36,399 INFO] Step 57250/102583; acc: 82.64; ppl: 1.75; xent: 0.56; lr: 0.00005; 2145/2550 tok/s; 5886 sec
  1248. [2019-02-04 19:28:41,235 INFO] Step 57300/102583; acc: 82.09; ppl: 1.79; xent: 0.58; lr: 0.00005; 1990/2737 tok/s; 5891 sec
  1249. [2019-02-04 19:28:46,230 INFO] Step 57350/102583; acc: 82.88; ppl: 1.76; xent: 0.57; lr: 0.00005; 2140/2550 tok/s; 5896 sec
  1250. [2019-02-04 19:28:51,001 INFO] Step 57400/102583; acc: 81.76; ppl: 1.80; xent: 0.59; lr: 0.00005; 2157/2622 tok/s; 5900 sec
  1251. [2019-02-04 19:28:56,258 INFO] Step 57450/102583; acc: 82.10; ppl: 1.76; xent: 0.57; lr: 0.00005; 1916/2627 tok/s; 5906 sec
  1252. [2019-02-04 19:29:01,308 INFO] Step 57500/102583; acc: 82.42; ppl: 1.76; xent: 0.56; lr: 0.00005; 2040/2714 tok/s; 5911 sec
  1253. [2019-02-04 19:29:06,285 INFO] Step 57550/102583; acc: 82.11; ppl: 1.80; xent: 0.59; lr: 0.00005; 2100/2527 tok/s; 5916 sec
  1254. [2019-02-04 19:29:11,164 INFO] Step 57600/102583; acc: 82.97; ppl: 1.73; xent: 0.55; lr: 0.00005; 2052/2686 tok/s; 5921 sec
  1255. [2019-02-04 19:29:16,003 INFO] Step 57650/102583; acc: 80.74; ppl: 1.85; xent: 0.62; lr: 0.00005; 2052/2661 tok/s; 5925 sec
  1256. [2019-02-04 19:29:20,607 INFO] Step 57700/102583; acc: 81.67; ppl: 1.83; xent: 0.60; lr: 0.00005; 2087/2600 tok/s; 5930 sec
  1257. [2019-02-04 19:29:25,956 INFO] Step 57750/102583; acc: 83.03; ppl: 1.73; xent: 0.55; lr: 0.00005; 1993/2657 tok/s; 5935 sec
  1258. [2019-02-04 19:29:30,939 INFO] Step 57800/102583; acc: 82.07; ppl: 1.76; xent: 0.57; lr: 0.00005; 2199/2541 tok/s; 5940 sec
  1259. [2019-02-04 19:29:35,929 INFO] Step 57850/102583; acc: 81.50; ppl: 1.83; xent: 0.60; lr: 0.00005; 1848/2752 tok/s; 5945 sec
  1260. [2019-02-04 19:29:40,996 INFO] Step 57900/102583; acc: 80.51; ppl: 1.88; xent: 0.63; lr: 0.00005; 1967/2540 tok/s; 5950 sec
  1261. [2019-02-04 19:29:45,842 INFO] Step 57950/102583; acc: 83.71; ppl: 1.71; xent: 0.54; lr: 0.00005; 2128/2685 tok/s; 5955 sec
  1262. [2019-02-04 19:29:50,656 INFO] Step 58000/102583; acc: 81.91; ppl: 1.77; xent: 0.57; lr: 0.00005; 1916/2747 tok/s; 5960 sec
  1263. [2019-02-04 19:29:55,895 INFO] Step 58050/102583; acc: 82.17; ppl: 1.79; xent: 0.58; lr: 0.00005; 2144/2504 tok/s; 5965 sec
  1264. [2019-02-04 19:30:00,810 INFO] Step 58100/102583; acc: 83.08; ppl: 1.75; xent: 0.56; lr: 0.00005; 2028/2692 tok/s; 5970 sec
  1265. [2019-02-04 19:30:05,911 INFO] Step 58150/102583; acc: 81.18; ppl: 1.83; xent: 0.61; lr: 0.00005; 2124/2459 tok/s; 5975 sec
  1266. [2019-02-04 19:30:10,955 INFO] Step 58200/102583; acc: 81.83; ppl: 1.79; xent: 0.58; lr: 0.00005; 2079/2684 tok/s; 5980 sec
  1267. [2019-02-04 19:30:15,680 INFO] Step 58250/102583; acc: 81.95; ppl: 1.80; xent: 0.59; lr: 0.00005; 2019/2630 tok/s; 5985 sec
  1268. [2019-02-04 19:30:20,351 INFO] Step 58300/102583; acc: 82.12; ppl: 1.80; xent: 0.59; lr: 0.00005; 2106/2597 tok/s; 5990 sec
  1269. [2019-02-04 19:30:25,748 INFO] Step 58350/102583; acc: 81.58; ppl: 1.80; xent: 0.59; lr: 0.00005; 1927/2644 tok/s; 5995 sec
  1270. [2019-02-04 19:30:35,665 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1271. [2019-02-04 19:30:37,044 INFO] Step 58400/102583; acc: 81.73; ppl: 1.84; xent: 0.61; lr: 0.00005; 899/1127 tok/s; 6006 sec
  1272. [2019-02-04 19:30:42,190 INFO] Step 58450/102583; acc: 82.59; ppl: 1.76; xent: 0.57; lr: 0.00005; 2102/2559 tok/s; 6012 sec
  1273. [2019-02-04 19:30:46,911 INFO] Step 58500/102583; acc: 81.93; ppl: 1.80; xent: 0.59; lr: 0.00005; 1948/2715 tok/s; 6016 sec
  1274. [2019-02-04 19:30:52,061 INFO] Step 58550/102583; acc: 83.67; ppl: 1.70; xent: 0.53; lr: 0.00005; 2110/2649 tok/s; 6021 sec
  1275. [2019-02-04 19:30:56,922 INFO] Step 58600/102583; acc: 81.60; ppl: 1.82; xent: 0.60; lr: 0.00005; 2112/2651 tok/s; 6026 sec
  1276. [2019-02-04 19:31:02,180 INFO] Step 58650/102583; acc: 81.94; ppl: 1.77; xent: 0.57; lr: 0.00005; 2055/2703 tok/s; 6032 sec
  1277. [2019-02-04 19:31:07,092 INFO] Step 58700/102583; acc: 80.98; ppl: 1.83; xent: 0.61; lr: 0.00005; 2062/2556 tok/s; 6036 sec
  1278. [2019-02-04 19:31:11,832 INFO] Step 58750/102583; acc: 82.86; ppl: 1.76; xent: 0.57; lr: 0.00005; 2109/2788 tok/s; 6041 sec
  1279. [2019-02-04 19:31:16,860 INFO] Step 58800/102583; acc: 81.20; ppl: 1.84; xent: 0.61; lr: 0.00005; 1928/2718 tok/s; 6046 sec
  1280. [2019-02-04 19:31:21,995 INFO] Step 58850/102583; acc: 81.26; ppl: 1.85; xent: 0.61; lr: 0.00005; 2087/2479 tok/s; 6051 sec
  1281. [2019-02-04 19:31:26,831 INFO] Step 58900/102583; acc: 82.28; ppl: 1.76; xent: 0.57; lr: 0.00005; 2112/2692 tok/s; 6056 sec
  1282. [2019-02-04 19:31:32,198 INFO] Step 58950/102583; acc: 81.74; ppl: 1.81; xent: 0.59; lr: 0.00005; 1986/2643 tok/s; 6062 sec
  1283. [2019-02-04 19:31:36,850 INFO] Step 59000/102583; acc: 82.56; ppl: 1.79; xent: 0.58; lr: 0.00005; 2218/2543 tok/s; 6066 sec
  1284. [2019-02-04 19:31:42,063 INFO] Step 59050/102583; acc: 80.51; ppl: 1.83; xent: 0.60; lr: 0.00005; 1950/2687 tok/s; 6071 sec
  1285. [2019-02-04 19:31:46,836 INFO] Step 59100/102583; acc: 83.65; ppl: 1.74; xent: 0.56; lr: 0.00005; 2145/2585 tok/s; 6076 sec
  1286. [2019-02-04 19:31:52,114 INFO] Step 59150/102583; acc: 82.36; ppl: 1.74; xent: 0.55; lr: 0.00005; 2102/2653 tok/s; 6082 sec
  1287. [2019-02-04 19:31:56,765 INFO] Step 59200/102583; acc: 81.66; ppl: 1.83; xent: 0.61; lr: 0.00005; 2065/2571 tok/s; 6086 sec
  1288. [2019-02-04 19:32:01,850 INFO] Step 59250/102583; acc: 82.17; ppl: 1.76; xent: 0.56; lr: 0.00005; 1978/2609 tok/s; 6091 sec
  1289. [2019-02-04 19:32:06,952 INFO] Step 59300/102583; acc: 82.46; ppl: 1.76; xent: 0.56; lr: 0.00005; 2015/2636 tok/s; 6096 sec
  1290. [2019-02-04 19:32:11,991 INFO] Step 59350/102583; acc: 81.65; ppl: 1.82; xent: 0.60; lr: 0.00005; 2038/2461 tok/s; 6101 sec
  1291. [2019-02-04 19:32:16,915 INFO] Step 59400/102583; acc: 80.81; ppl: 1.83; xent: 0.60; lr: 0.00005; 2007/2762 tok/s; 6106 sec
  1292. [2019-02-04 19:32:22,214 INFO] Step 59450/102583; acc: 82.48; ppl: 1.76; xent: 0.57; lr: 0.00005; 1927/2544 tok/s; 6112 sec
  1293. [2019-02-04 19:32:27,222 INFO] Step 59500/102583; acc: 83.07; ppl: 1.73; xent: 0.55; lr: 0.00005; 2207/2630 tok/s; 6117 sec
  1294. [2019-02-04 19:32:31,966 INFO] Step 59550/102583; acc: 80.46; ppl: 1.88; xent: 0.63; lr: 0.00005; 2082/2578 tok/s; 6121 sec
  1295. [2019-02-04 19:32:36,908 INFO] Step 59600/102583; acc: 80.32; ppl: 1.86; xent: 0.62; lr: 0.00005; 1926/2676 tok/s; 6126 sec
  1296. [2019-02-04 19:32:41,859 INFO] Step 59650/102583; acc: 82.79; ppl: 1.78; xent: 0.57; lr: 0.00005; 2154/2658 tok/s; 6131 sec
  1297. [2019-02-04 19:32:47,043 INFO] Step 59700/102583; acc: 82.41; ppl: 1.76; xent: 0.57; lr: 0.00005; 2041/2495 tok/s; 6136 sec
  1298. [2019-02-04 19:32:52,240 INFO] Step 59750/102583; acc: 82.36; ppl: 1.76; xent: 0.56; lr: 0.00005; 2042/2624 tok/s; 6142 sec
  1299. [2019-02-04 19:32:57,200 INFO] Step 59800/102583; acc: 80.88; ppl: 1.82; xent: 0.60; lr: 0.00005; 1835/2725 tok/s; 6147 sec
  1300. [2019-02-04 19:33:02,176 INFO] Step 59850/102583; acc: 82.31; ppl: 1.77; xent: 0.57; lr: 0.00005; 2208/2682 tok/s; 6152 sec
  1301. [2019-02-04 19:33:07,017 INFO] Step 59900/102583; acc: 81.64; ppl: 1.83; xent: 0.60; lr: 0.00005; 1939/2502 tok/s; 6156 sec
  1302. [2019-02-04 19:33:11,705 INFO] Step 59950/102583; acc: 81.08; ppl: 1.86; xent: 0.62; lr: 0.00005; 2002/2611 tok/s; 6161 sec
  1303. [2019-02-04 19:33:16,981 INFO] Step 60000/102583; acc: 82.81; ppl: 1.72; xent: 0.54; lr: 0.00003; 2116/2564 tok/s; 6166 sec
  1304. [2019-02-04 19:33:17,171 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  1305. [2019-02-04 19:33:18,475 INFO] Validation perplexity: 1.74624
  1306. [2019-02-04 19:33:18,476 INFO] Validation accuracy: 82.7164
  1307. [2019-02-04 19:33:18,477 INFO] Saving checkpoint models/ted1_step_60000.pt
  1308. [2019-02-04 19:33:29,593 INFO] Step 60050/102583; acc: 80.24; ppl: 1.87; xent: 0.63; lr: 0.00003; 780/1083 tok/s; 6179 sec
  1309. [2019-02-04 19:33:34,645 INFO] Step 60100/102583; acc: 82.97; ppl: 1.73; xent: 0.55; lr: 0.00003; 2071/2597 tok/s; 6184 sec
  1310. [2019-02-04 19:33:39,703 INFO] Step 60150/102583; acc: 82.82; ppl: 1.77; xent: 0.57; lr: 0.00003; 2010/2657 tok/s; 6189 sec
  1311. [2019-02-04 19:33:44,451 INFO] Step 60200/102583; acc: 80.30; ppl: 1.89; xent: 0.63; lr: 0.00003; 2002/2628 tok/s; 6194 sec
  1312. [2019-02-04 19:33:49,533 INFO] Step 60250/102583; acc: 82.55; ppl: 1.75; xent: 0.56; lr: 0.00003; 2027/2623 tok/s; 6199 sec
  1313. [2019-02-04 19:33:54,381 INFO] Step 60300/102583; acc: 82.57; ppl: 1.78; xent: 0.58; lr: 0.00003; 2131/2538 tok/s; 6204 sec
  1314. [2019-02-04 19:33:59,687 INFO] Step 60350/102583; acc: 83.07; ppl: 1.73; xent: 0.55; lr: 0.00003; 2135/2554 tok/s; 6209 sec
  1315. [2019-02-04 19:34:04,390 INFO] Step 60400/102583; acc: 82.05; ppl: 1.80; xent: 0.59; lr: 0.00003; 1886/2634 tok/s; 6214 sec
  1316. [2019-02-04 19:34:09,361 INFO] Step 60450/102583; acc: 80.87; ppl: 1.87; xent: 0.63; lr: 0.00003; 1977/2639 tok/s; 6219 sec
  1317. [2019-02-04 19:34:14,401 INFO] Step 60500/102583; acc: 82.27; ppl: 1.78; xent: 0.58; lr: 0.00003; 2048/2554 tok/s; 6224 sec
  1318. [2019-02-04 19:34:19,179 INFO] Step 60550/102583; acc: 82.82; ppl: 1.75; xent: 0.56; lr: 0.00003; 2097/2721 tok/s; 6229 sec
  1319. [2019-02-04 19:34:24,222 INFO] Step 60600/102583; acc: 82.46; ppl: 1.78; xent: 0.58; lr: 0.00003; 2145/2469 tok/s; 6234 sec
  1320. [2019-02-04 19:34:28,967 INFO] Step 60650/102583; acc: 82.02; ppl: 1.77; xent: 0.57; lr: 0.00003; 2079/2754 tok/s; 6238 sec
  1321. [2019-02-04 19:34:34,137 INFO] Step 60700/102583; acc: 82.42; ppl: 1.76; xent: 0.57; lr: 0.00003; 1942/2633 tok/s; 6244 sec
  1322. [2019-02-04 19:34:39,102 INFO] Step 60750/102583; acc: 81.72; ppl: 1.79; xent: 0.58; lr: 0.00003; 2056/2655 tok/s; 6248 sec
  1323. [2019-02-04 19:34:44,050 INFO] Step 60800/102583; acc: 82.73; ppl: 1.76; xent: 0.57; lr: 0.00003; 2103/2508 tok/s; 6253 sec
  1324. [2019-02-04 19:34:49,069 INFO] Step 60850/102583; acc: 82.73; ppl: 1.74; xent: 0.55; lr: 0.00003; 2020/2670 tok/s; 6258 sec
  1325. [2019-02-04 19:34:54,468 INFO] Step 60900/102583; acc: 81.03; ppl: 1.82; xent: 0.60; lr: 0.00003; 2089/2671 tok/s; 6264 sec
  1326. [2019-02-04 19:34:59,326 INFO] Step 60950/102583; acc: 81.87; ppl: 1.81; xent: 0.59; lr: 0.00003; 2025/2616 tok/s; 6269 sec
  1327. [2019-02-04 19:35:04,483 INFO] Step 61000/102583; acc: 82.49; ppl: 1.76; xent: 0.57; lr: 0.00003; 2049/2693 tok/s; 6274 sec
  1328. [2019-02-04 19:35:09,590 INFO] Step 61050/102583; acc: 82.81; ppl: 1.73; xent: 0.55; lr: 0.00003; 2145/2500 tok/s; 6279 sec
  1329. [2019-02-04 19:35:14,691 INFO] Step 61100/102583; acc: 81.59; ppl: 1.81; xent: 0.59; lr: 0.00003; 1773/2871 tok/s; 6284 sec
  1330. [2019-02-04 19:35:19,556 INFO] Step 61150/102583; acc: 81.17; ppl: 1.86; xent: 0.62; lr: 0.00003; 2075/2455 tok/s; 6289 sec
  1331. [2019-02-04 19:35:24,461 INFO] Step 61200/102583; acc: 83.34; ppl: 1.70; xent: 0.53; lr: 0.00003; 2141/2710 tok/s; 6294 sec
  1332. [2019-02-04 19:35:29,219 INFO] Step 61250/102583; acc: 81.76; ppl: 1.79; xent: 0.58; lr: 0.00003; 1938/2741 tok/s; 6299 sec
  1333. [2019-02-04 19:35:34,572 INFO] Step 61300/102583; acc: 82.04; ppl: 1.78; xent: 0.58; lr: 0.00003; 2111/2492 tok/s; 6304 sec
  1334. [2019-02-04 19:35:39,328 INFO] Step 61350/102583; acc: 83.24; ppl: 1.74; xent: 0.56; lr: 0.00003; 2090/2668 tok/s; 6309 sec
  1335. [2019-02-04 19:35:44,442 INFO] Step 61400/102583; acc: 82.24; ppl: 1.77; xent: 0.57; lr: 0.00003; 2142/2470 tok/s; 6314 sec
  1336. [2019-02-04 19:35:49,564 INFO] Step 61450/102583; acc: 81.39; ppl: 1.83; xent: 0.60; lr: 0.00003; 1980/2663 tok/s; 6319 sec
  1337. [2019-02-04 19:35:54,215 INFO] Step 61500/102583; acc: 82.75; ppl: 1.75; xent: 0.56; lr: 0.00003; 2026/2689 tok/s; 6324 sec
  1338. [2019-02-04 19:35:58,963 INFO] Step 61550/102583; acc: 81.55; ppl: 1.83; xent: 0.60; lr: 0.00003; 2086/2537 tok/s; 6328 sec
  1339. [2019-02-04 19:36:04,228 INFO] Step 61600/102583; acc: 82.17; ppl: 1.77; xent: 0.57; lr: 0.00003; 2000/2669 tok/s; 6334 sec
  1340. [2019-02-04 19:36:13,052 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1341. [2019-02-04 19:36:14,991 INFO] Step 61650/102583; acc: 81.10; ppl: 1.86; xent: 0.62; lr: 0.00003; 893/1186 tok/s; 6344 sec
  1342. [2019-02-04 19:36:20,161 INFO] Step 61700/102583; acc: 83.36; ppl: 1.73; xent: 0.55; lr: 0.00003; 2220/2542 tok/s; 6350 sec
  1343. [2019-02-04 19:36:24,857 INFO] Step 61750/102583; acc: 82.50; ppl: 1.77; xent: 0.57; lr: 0.00003; 1917/2712 tok/s; 6354 sec
  1344. [2019-02-04 19:36:30,084 INFO] Step 61800/102583; acc: 83.94; ppl: 1.70; xent: 0.53; lr: 0.00003; 2067/2658 tok/s; 6359 sec
  1345. [2019-02-04 19:36:34,923 INFO] Step 61850/102583; acc: 81.57; ppl: 1.82; xent: 0.60; lr: 0.00003; 2117/2688 tok/s; 6364 sec
  1346. [2019-02-04 19:36:39,943 INFO] Step 61900/102583; acc: 82.60; ppl: 1.74; xent: 0.56; lr: 0.00003; 2097/2654 tok/s; 6369 sec
  1347. [2019-02-04 19:36:44,952 INFO] Step 61950/102583; acc: 81.80; ppl: 1.80; xent: 0.59; lr: 0.00003; 2002/2554 tok/s; 6374 sec
  1348. [2019-02-04 19:36:49,720 INFO] Step 62000/102583; acc: 82.69; ppl: 1.77; xent: 0.57; lr: 0.00003; 2112/2790 tok/s; 6379 sec
  1349. [2019-02-04 19:36:54,838 INFO] Step 62050/102583; acc: 81.27; ppl: 1.82; xent: 0.60; lr: 0.00003; 2043/2577 tok/s; 6384 sec
  1350. [2019-02-04 19:36:59,791 INFO] Step 62100/102583; acc: 80.96; ppl: 1.86; xent: 0.62; lr: 0.00003; 1963/2638 tok/s; 6389 sec
  1351. [2019-02-04 19:37:04,485 INFO] Step 62150/102583; acc: 83.14; ppl: 1.71; xent: 0.54; lr: 0.00003; 2213/2662 tok/s; 6394 sec
  1352. [2019-02-04 19:37:09,768 INFO] Step 62200/102583; acc: 82.04; ppl: 1.80; xent: 0.59; lr: 0.00003; 1957/2587 tok/s; 6399 sec
  1353. [2019-02-04 19:37:14,468 INFO] Step 62250/102583; acc: 81.86; ppl: 1.83; xent: 0.60; lr: 0.00003; 2134/2549 tok/s; 6404 sec
  1354. [2019-02-04 19:37:19,473 INFO] Step 62300/102583; acc: 80.56; ppl: 1.83; xent: 0.60; lr: 0.00003; 1943/2754 tok/s; 6409 sec
  1355. [2019-02-04 19:37:24,394 INFO] Step 62350/102583; acc: 83.91; ppl: 1.72; xent: 0.54; lr: 0.00003; 2101/2595 tok/s; 6414 sec
  1356. [2019-02-04 19:37:29,478 INFO] Step 62400/102583; acc: 82.64; ppl: 1.75; xent: 0.56; lr: 0.00003; 2162/2626 tok/s; 6419 sec
  1357. [2019-02-04 19:37:34,140 INFO] Step 62450/102583; acc: 82.27; ppl: 1.79; xent: 0.58; lr: 0.00003; 2107/2558 tok/s; 6424 sec
  1358. [2019-02-04 19:37:39,270 INFO] Step 62500/102583; acc: 81.88; ppl: 1.77; xent: 0.57; lr: 0.00003; 1941/2626 tok/s; 6429 sec
  1359. [2019-02-04 19:37:44,362 INFO] Step 62550/102583; acc: 83.20; ppl: 1.73; xent: 0.55; lr: 0.00003; 2044/2632 tok/s; 6434 sec
  1360. [2019-02-04 19:37:49,564 INFO] Step 62600/102583; acc: 81.46; ppl: 1.82; xent: 0.60; lr: 0.00003; 1993/2507 tok/s; 6439 sec
  1361. [2019-02-04 19:37:54,517 INFO] Step 62650/102583; acc: 80.38; ppl: 1.85; xent: 0.61; lr: 0.00003; 1971/2731 tok/s; 6444 sec
  1362. [2019-02-04 19:37:59,641 INFO] Step 62700/102583; acc: 84.22; ppl: 1.69; xent: 0.53; lr: 0.00003; 2016/2548 tok/s; 6449 sec
  1363. [2019-02-04 19:38:04,622 INFO] Step 62750/102583; acc: 82.06; ppl: 1.78; xent: 0.58; lr: 0.00003; 2206/2617 tok/s; 6454 sec
  1364. [2019-02-04 19:38:09,374 INFO] Step 62800/102583; acc: 81.30; ppl: 1.84; xent: 0.61; lr: 0.00003; 2062/2572 tok/s; 6459 sec
  1365. [2019-02-04 19:38:14,379 INFO] Step 62850/102583; acc: 81.11; ppl: 1.84; xent: 0.61; lr: 0.00003; 1947/2696 tok/s; 6464 sec
  1366. [2019-02-04 19:38:19,270 INFO] Step 62900/102583; acc: 82.19; ppl: 1.78; xent: 0.58; lr: 0.00003; 2118/2711 tok/s; 6469 sec
  1367. [2019-02-04 19:38:24,315 INFO] Step 62950/102583; acc: 83.41; ppl: 1.72; xent: 0.54; lr: 0.00003; 2109/2412 tok/s; 6474 sec
  1368. [2019-02-04 19:38:29,552 INFO] Step 63000/102583; acc: 82.13; ppl: 1.77; xent: 0.57; lr: 0.00003; 1941/2664 tok/s; 6479 sec
  1369. [2019-02-04 19:38:34,324 INFO] Step 63050/102583; acc: 81.65; ppl: 1.80; xent: 0.59; lr: 0.00003; 1910/2672 tok/s; 6484 sec
  1370. [2019-02-04 19:38:39,422 INFO] Step 63100/102583; acc: 82.25; ppl: 1.77; xent: 0.57; lr: 0.00003; 2102/2661 tok/s; 6489 sec
  1371. [2019-02-04 19:38:44,270 INFO] Step 63150/102583; acc: 81.58; ppl: 1.83; xent: 0.61; lr: 0.00003; 2025/2553 tok/s; 6494 sec
  1372. [2019-02-04 19:38:49,128 INFO] Step 63200/102583; acc: 82.02; ppl: 1.80; xent: 0.59; lr: 0.00003; 2042/2594 tok/s; 6499 sec
  1373. [2019-02-04 19:38:54,313 INFO] Step 63250/102583; acc: 82.09; ppl: 1.75; xent: 0.56; lr: 0.00003; 2038/2642 tok/s; 6504 sec
  1374. [2019-02-04 19:38:59,208 INFO] Step 63300/102583; acc: 80.80; ppl: 1.85; xent: 0.62; lr: 0.00003; 2084/2642 tok/s; 6509 sec
  1375. [2019-02-04 19:39:04,140 INFO] Step 63350/102583; acc: 82.79; ppl: 1.74; xent: 0.56; lr: 0.00003; 2028/2669 tok/s; 6514 sec
  1376. [2019-02-04 19:39:09,169 INFO] Step 63400/102583; acc: 82.91; ppl: 1.77; xent: 0.57; lr: 0.00003; 2047/2635 tok/s; 6519 sec
  1377. [2019-02-04 19:39:13,974 INFO] Step 63450/102583; acc: 80.47; ppl: 1.88; xent: 0.63; lr: 0.00003; 1944/2546 tok/s; 6523 sec
  1378. [2019-02-04 19:39:19,069 INFO] Step 63500/102583; acc: 82.47; ppl: 1.74; xent: 0.55; lr: 0.00003; 2045/2705 tok/s; 6528 sec
  1379. [2019-02-04 19:39:24,097 INFO] Step 63550/102583; acc: 82.54; ppl: 1.77; xent: 0.57; lr: 0.00003; 2095/2513 tok/s; 6533 sec
  1380. [2019-02-04 19:39:29,308 INFO] Step 63600/102583; acc: 82.88; ppl: 1.76; xent: 0.57; lr: 0.00003; 2139/2521 tok/s; 6539 sec
  1381. [2019-02-04 19:39:33,850 INFO] Step 63650/102583; acc: 82.28; ppl: 1.79; xent: 0.58; lr: 0.00003; 1942/2728 tok/s; 6543 sec
  1382. [2019-02-04 19:39:38,869 INFO] Step 63700/102583; acc: 80.24; ppl: 1.88; xent: 0.63; lr: 0.00003; 1981/2645 tok/s; 6548 sec
  1383. [2019-02-04 19:39:43,938 INFO] Step 63750/102583; acc: 83.03; ppl: 1.75; xent: 0.56; lr: 0.00003; 1986/2580 tok/s; 6553 sec
  1384. [2019-02-04 19:39:48,649 INFO] Step 63800/102583; acc: 83.26; ppl: 1.73; xent: 0.55; lr: 0.00003; 2113/2677 tok/s; 6558 sec
  1385. [2019-02-04 19:39:53,691 INFO] Step 63850/102583; acc: 81.34; ppl: 1.83; xent: 0.61; lr: 0.00003; 2121/2477 tok/s; 6563 sec
  1386. [2019-02-04 19:39:58,819 INFO] Step 63900/102583; acc: 83.58; ppl: 1.68; xent: 0.52; lr: 0.00003; 2135/2611 tok/s; 6568 sec
  1387. [2019-02-04 19:40:03,665 INFO] Step 63950/102583; acc: 80.96; ppl: 1.83; xent: 0.60; lr: 0.00003; 1835/2810 tok/s; 6573 sec
  1388. [2019-02-04 19:40:08,686 INFO] Step 64000/102583; acc: 81.90; ppl: 1.79; xent: 0.58; lr: 0.00003; 2112/2632 tok/s; 6578 sec
  1389. [2019-02-04 19:40:13,609 INFO] Step 64050/102583; acc: 84.17; ppl: 1.70; xent: 0.53; lr: 0.00003; 2150/2513 tok/s; 6583 sec
  1390. [2019-02-04 19:40:18,558 INFO] Step 64100/102583; acc: 81.85; ppl: 1.78; xent: 0.58; lr: 0.00003; 1995/2668 tok/s; 6588 sec
  1391. [2019-02-04 19:40:23,543 INFO] Step 64150/102583; acc: 81.57; ppl: 1.80; xent: 0.59; lr: 0.00003; 2078/2656 tok/s; 6593 sec
  1392. [2019-02-04 19:40:28,213 INFO] Step 64200/102583; acc: 81.72; ppl: 1.82; xent: 0.60; lr: 0.00003; 2064/2621 tok/s; 6598 sec
  1393. [2019-02-04 19:40:33,263 INFO] Step 64250/102583; acc: 83.06; ppl: 1.74; xent: 0.55; lr: 0.00003; 2081/2643 tok/s; 6603 sec
  1394. [2019-02-04 19:40:38,404 INFO] Step 64300/102583; acc: 82.65; ppl: 1.73; xent: 0.55; lr: 0.00003; 2081/2582 tok/s; 6608 sec
  1395. [2019-02-04 19:40:43,395 INFO] Step 64350/102583; acc: 81.41; ppl: 1.85; xent: 0.61; lr: 0.00003; 1774/2809 tok/s; 6613 sec
  1396. [2019-02-04 19:40:48,252 INFO] Step 64400/102583; acc: 80.74; ppl: 1.87; xent: 0.63; lr: 0.00003; 1997/2460 tok/s; 6618 sec
  1397. [2019-02-04 19:40:53,206 INFO] Step 64450/102583; acc: 83.52; ppl: 1.69; xent: 0.53; lr: 0.00003; 2197/2743 tok/s; 6623 sec
  1398. [2019-02-04 19:40:58,158 INFO] Step 64500/102583; acc: 82.58; ppl: 1.75; xent: 0.56; lr: 0.00003; 2007/2590 tok/s; 6628 sec
  1399. [2019-02-04 19:41:03,331 INFO] Step 64550/102583; acc: 82.29; ppl: 1.77; xent: 0.57; lr: 0.00003; 2033/2612 tok/s; 6633 sec
  1400. [2019-02-04 19:41:08,205 INFO] Step 64600/102583; acc: 81.54; ppl: 1.83; xent: 0.60; lr: 0.00003; 2103/2646 tok/s; 6638 sec
  1401. [2019-02-04 19:41:13,328 INFO] Step 64650/102583; acc: 83.25; ppl: 1.71; xent: 0.54; lr: 0.00003; 2096/2493 tok/s; 6643 sec
  1402. [2019-02-04 19:41:18,268 INFO] Step 64700/102583; acc: 81.61; ppl: 1.83; xent: 0.60; lr: 0.00003; 2100/2603 tok/s; 6648 sec
  1403. [2019-02-04 19:41:23,105 INFO] Step 64750/102583; acc: 82.19; ppl: 1.78; xent: 0.58; lr: 0.00003; 1984/2659 tok/s; 6652 sec
  1404. [2019-02-04 19:41:27,796 INFO] Step 64800/102583; acc: 81.97; ppl: 1.80; xent: 0.59; lr: 0.00003; 2032/2634 tok/s; 6657 sec
  1405. [2019-02-04 19:41:33,091 INFO] Step 64850/102583; acc: 81.84; ppl: 1.79; xent: 0.58; lr: 0.00003; 1973/2607 tok/s; 6662 sec
  1406. [2019-02-04 19:41:41,598 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1407. [2019-02-04 19:41:44,136 INFO] Step 64900/102583; acc: 81.32; ppl: 1.84; xent: 0.61; lr: 0.00003; 926/1161 tok/s; 6674 sec
  1408. [2019-02-04 19:41:49,119 INFO] Step 64950/102583; acc: 83.34; ppl: 1.75; xent: 0.56; lr: 0.00003; 2175/2517 tok/s; 6679 sec
  1409. [2019-02-04 19:41:53,985 INFO] Step 65000/102583; acc: 82.47; ppl: 1.76; xent: 0.57; lr: 0.00003; 1909/2743 tok/s; 6683 sec
  1410. [2019-02-04 19:41:54,028 INFO] Saving checkpoint models/ted1_step_65000.pt
  1411. [2019-02-04 19:42:04,946 INFO] Step 65050/102583; acc: 83.91; ppl: 1.71; xent: 0.53; lr: 0.00003; 940/1236 tok/s; 6694 sec
  1412. [2019-02-04 19:42:09,852 INFO] Step 65100/102583; acc: 81.38; ppl: 1.82; xent: 0.60; lr: 0.00003; 2140/2720 tok/s; 6699 sec
  1413. [2019-02-04 19:42:15,050 INFO] Step 65150/102583; acc: 82.45; ppl: 1.73; xent: 0.55; lr: 0.00003; 2006/2705 tok/s; 6704 sec
  1414. [2019-02-04 19:42:19,904 INFO] Step 65200/102583; acc: 81.83; ppl: 1.79; xent: 0.58; lr: 0.00003; 2127/2516 tok/s; 6709 sec
  1415. [2019-02-04 19:42:24,665 INFO] Step 65250/102583; acc: 83.17; ppl: 1.77; xent: 0.57; lr: 0.00003; 2149/2710 tok/s; 6714 sec
  1416. [2019-02-04 19:42:29,773 INFO] Step 65300/102583; acc: 80.23; ppl: 1.85; xent: 0.62; lr: 0.00003; 1997/2587 tok/s; 6719 sec
  1417. [2019-02-04 19:42:34,707 INFO] Step 65350/102583; acc: 81.61; ppl: 1.83; xent: 0.61; lr: 0.00003; 1931/2647 tok/s; 6724 sec
  1418. [2019-02-04 19:42:39,674 INFO] Step 65400/102583; acc: 83.53; ppl: 1.67; xent: 0.51; lr: 0.00003; 2186/2724 tok/s; 6729 sec
  1419. [2019-02-04 19:42:44,701 INFO] Step 65450/102583; acc: 80.81; ppl: 1.88; xent: 0.63; lr: 0.00003; 1950/2531 tok/s; 6734 sec
  1420. [2019-02-04 19:42:49,425 INFO] Step 65500/102583; acc: 81.67; ppl: 1.84; xent: 0.61; lr: 0.00003; 2123/2568 tok/s; 6739 sec
  1421. [2019-02-04 19:42:54,464 INFO] Step 65550/102583; acc: 81.29; ppl: 1.78; xent: 0.58; lr: 0.00003; 2019/2762 tok/s; 6744 sec
  1422. [2019-02-04 19:42:59,501 INFO] Step 65600/102583; acc: 84.05; ppl: 1.70; xent: 0.53; lr: 0.00003; 2111/2566 tok/s; 6749 sec
  1423. [2019-02-04 19:43:04,483 INFO] Step 65650/102583; acc: 81.84; ppl: 1.79; xent: 0.58; lr: 0.00003; 2122/2658 tok/s; 6754 sec
  1424. [2019-02-04 19:43:09,176 INFO] Step 65700/102583; acc: 83.03; ppl: 1.76; xent: 0.56; lr: 0.00003; 2179/2457 tok/s; 6759 sec
  1425. [2019-02-04 19:43:14,263 INFO] Step 65750/102583; acc: 81.73; ppl: 1.77; xent: 0.57; lr: 0.00003; 1843/2744 tok/s; 6764 sec
  1426. [2019-02-04 19:43:19,487 INFO] Step 65800/102583; acc: 82.94; ppl: 1.75; xent: 0.56; lr: 0.00003; 2107/2518 tok/s; 6769 sec
  1427. [2019-02-04 19:43:24,526 INFO] Step 65850/102583; acc: 81.90; ppl: 1.81; xent: 0.59; lr: 0.00003; 2010/2544 tok/s; 6774 sec
  1428. [2019-02-04 19:43:29,627 INFO] Step 65900/102583; acc: 80.45; ppl: 1.86; xent: 0.62; lr: 0.00003; 1875/2744 tok/s; 6779 sec
  1429. [2019-02-04 19:43:34,677 INFO] Step 65950/102583; acc: 84.03; ppl: 1.70; xent: 0.53; lr: 0.00003; 1997/2595 tok/s; 6784 sec
  1430. [2019-02-04 19:43:39,715 INFO] Step 66000/102583; acc: 81.49; ppl: 1.81; xent: 0.59; lr: 0.00003; 2166/2527 tok/s; 6789 sec
  1431. [2019-02-04 19:43:44,302 INFO] Step 66050/102583; acc: 80.99; ppl: 1.85; xent: 0.61; lr: 0.00003; 2124/2622 tok/s; 6794 sec
  1432. [2019-02-04 19:43:49,315 INFO] Step 66100/102583; acc: 82.18; ppl: 1.77; xent: 0.57; lr: 0.00003; 2039/2704 tok/s; 6799 sec
  1433. [2019-02-04 19:43:54,435 INFO] Step 66150/102583; acc: 82.64; ppl: 1.76; xent: 0.56; lr: 0.00003; 2056/2662 tok/s; 6804 sec
  1434. [2019-02-04 19:43:59,393 INFO] Step 66200/102583; acc: 83.91; ppl: 1.70; xent: 0.53; lr: 0.00003; 2188/2358 tok/s; 6809 sec
  1435. [2019-02-04 19:44:04,522 INFO] Step 66250/102583; acc: 80.28; ppl: 1.86; xent: 0.62; lr: 0.00003; 1847/2782 tok/s; 6814 sec
  1436. [2019-02-04 19:44:09,374 INFO] Step 66300/102583; acc: 81.82; ppl: 1.79; xent: 0.58; lr: 0.00003; 1865/2723 tok/s; 6819 sec
  1437. [2019-02-04 19:44:14,276 INFO] Step 66350/102583; acc: 83.54; ppl: 1.71; xent: 0.54; lr: 0.00003; 2199/2531 tok/s; 6824 sec
  1438. [2019-02-04 19:44:19,113 INFO] Step 66400/102583; acc: 81.14; ppl: 1.85; xent: 0.62; lr: 0.00003; 1995/2598 tok/s; 6829 sec
  1439. [2019-02-04 19:44:24,260 INFO] Step 66450/102583; acc: 82.72; ppl: 1.74; xent: 0.55; lr: 0.00003; 2050/2535 tok/s; 6834 sec
  1440. [2019-02-04 19:44:29,297 INFO] Step 66500/102583; acc: 81.50; ppl: 1.80; xent: 0.59; lr: 0.00003; 2090/2685 tok/s; 6839 sec
  1441. [2019-02-04 19:44:34,159 INFO] Step 66550/102583; acc: 81.68; ppl: 1.81; xent: 0.59; lr: 0.00003; 1995/2672 tok/s; 6844 sec
  1442. [2019-02-04 19:44:39,143 INFO] Step 66600/102583; acc: 83.16; ppl: 1.73; xent: 0.55; lr: 0.00003; 2021/2728 tok/s; 6849 sec
  1443. [2019-02-04 19:44:44,043 INFO] Step 66650/102583; acc: 82.19; ppl: 1.80; xent: 0.59; lr: 0.00003; 2055/2604 tok/s; 6853 sec
  1444. [2019-02-04 19:44:49,070 INFO] Step 66700/102583; acc: 80.95; ppl: 1.84; xent: 0.61; lr: 0.00003; 1915/2539 tok/s; 6858 sec
  1445. [2019-02-04 19:44:53,914 INFO] Step 66750/102583; acc: 82.09; ppl: 1.78; xent: 0.58; lr: 0.00003; 2162/2692 tok/s; 6863 sec
  1446. [2019-02-04 19:44:59,054 INFO] Step 66800/102583; acc: 82.07; ppl: 1.79; xent: 0.58; lr: 0.00003; 2012/2565 tok/s; 6868 sec
  1447. [2019-02-04 19:45:04,155 INFO] Step 66850/102583; acc: 83.49; ppl: 1.73; xent: 0.55; lr: 0.00003; 2198/2456 tok/s; 6874 sec
  1448. [2019-02-04 19:45:08,735 INFO] Step 66900/102583; acc: 81.04; ppl: 1.86; xent: 0.62; lr: 0.00003; 1828/2766 tok/s; 6878 sec
  1449. [2019-02-04 19:45:13,774 INFO] Step 66950/102583; acc: 81.14; ppl: 1.83; xent: 0.61; lr: 0.00003; 2058/2606 tok/s; 6883 sec
  1450. [2019-02-04 19:45:18,708 INFO] Step 67000/102583; acc: 83.57; ppl: 1.74; xent: 0.55; lr: 0.00003; 2061/2566 tok/s; 6888 sec
  1451. [2019-02-04 19:45:23,522 INFO] Step 67050/102583; acc: 83.04; ppl: 1.72; xent: 0.54; lr: 0.00003; 2032/2697 tok/s; 6893 sec
  1452. [2019-02-04 19:45:28,554 INFO] Step 67100/102583; acc: 81.44; ppl: 1.82; xent: 0.60; lr: 0.00003; 2184/2482 tok/s; 6898 sec
  1453. [2019-02-04 19:45:33,578 INFO] Step 67150/102583; acc: 83.16; ppl: 1.71; xent: 0.54; lr: 0.00003; 2116/2594 tok/s; 6903 sec
  1454. [2019-02-04 19:45:38,644 INFO] Step 67200/102583; acc: 81.23; ppl: 1.80; xent: 0.59; lr: 0.00003; 1824/2815 tok/s; 6908 sec
  1455. [2019-02-04 19:45:43,766 INFO] Step 67250/102583; acc: 82.87; ppl: 1.75; xent: 0.56; lr: 0.00003; 2043/2581 tok/s; 6913 sec
  1456. [2019-02-04 19:45:48,376 INFO] Step 67300/102583; acc: 82.59; ppl: 1.77; xent: 0.57; lr: 0.00003; 2169/2584 tok/s; 6918 sec
  1457. [2019-02-04 19:45:53,430 INFO] Step 67350/102583; acc: 82.55; ppl: 1.75; xent: 0.56; lr: 0.00003; 2026/2636 tok/s; 6923 sec
  1458. [2019-02-04 19:45:58,362 INFO] Step 67400/102583; acc: 81.09; ppl: 1.83; xent: 0.60; lr: 0.00003; 2095/2633 tok/s; 6928 sec
  1459. [2019-02-04 19:46:03,078 INFO] Step 67450/102583; acc: 81.77; ppl: 1.83; xent: 0.60; lr: 0.00003; 2012/2665 tok/s; 6932 sec
  1460. [2019-02-04 19:46:08,106 INFO] Step 67500/102583; acc: 83.43; ppl: 1.70; xent: 0.53; lr: 0.00003; 2078/2639 tok/s; 6937 sec
  1461. [2019-02-04 19:46:13,275 INFO] Step 67550/102583; acc: 82.66; ppl: 1.73; xent: 0.55; lr: 0.00003; 2075/2567 tok/s; 6943 sec
  1462. [2019-02-04 19:46:18,321 INFO] Step 67600/102583; acc: 81.15; ppl: 1.84; xent: 0.61; lr: 0.00003; 1823/2794 tok/s; 6948 sec
  1463. [2019-02-04 19:46:23,069 INFO] Step 67650/102583; acc: 81.39; ppl: 1.85; xent: 0.61; lr: 0.00003; 2017/2509 tok/s; 6952 sec
  1464. [2019-02-04 19:46:28,012 INFO] Step 67700/102583; acc: 83.19; ppl: 1.70; xent: 0.53; lr: 0.00003; 2150/2761 tok/s; 6957 sec
  1465. [2019-02-04 19:46:33,109 INFO] Step 67750/102583; acc: 82.24; ppl: 1.78; xent: 0.58; lr: 0.00003; 1986/2472 tok/s; 6962 sec
  1466. [2019-02-04 19:46:38,121 INFO] Step 67800/102583; acc: 83.01; ppl: 1.73; xent: 0.55; lr: 0.00003; 2127/2711 tok/s; 6968 sec
  1467. [2019-02-04 19:46:43,185 INFO] Step 67850/102583; acc: 81.39; ppl: 1.82; xent: 0.60; lr: 0.00003; 2134/2571 tok/s; 6973 sec
  1468. [2019-02-04 19:46:48,289 INFO] Step 67900/102583; acc: 82.99; ppl: 1.74; xent: 0.56; lr: 0.00003; 2036/2562 tok/s; 6978 sec
  1469. [2019-02-04 19:46:52,980 INFO] Step 67950/102583; acc: 81.64; ppl: 1.83; xent: 0.60; lr: 0.00003; 2097/2579 tok/s; 6982 sec
  1470. [2019-02-04 19:46:57,863 INFO] Step 68000/102583; acc: 82.26; ppl: 1.76; xent: 0.57; lr: 0.00003; 2071/2663 tok/s; 6987 sec
  1471. [2019-02-04 19:47:02,753 INFO] Step 68050/102583; acc: 81.94; ppl: 1.80; xent: 0.59; lr: 0.00003; 1914/2673 tok/s; 6992 sec
  1472. [2019-02-04 19:47:07,917 INFO] Step 68100/102583; acc: 81.98; ppl: 1.77; xent: 0.57; lr: 0.00003; 1981/2600 tok/s; 6997 sec
  1473. [2019-02-04 19:47:16,082 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1474. [2019-02-04 19:47:19,365 INFO] Step 68150/102583; acc: 82.85; ppl: 1.75; xent: 0.56; lr: 0.00003; 905/1174 tok/s; 7009 sec
  1475. [2019-02-04 19:47:24,200 INFO] Step 68200/102583; acc: 81.83; ppl: 1.84; xent: 0.61; lr: 0.00003; 2172/2383 tok/s; 7014 sec
  1476. [2019-02-04 19:47:29,005 INFO] Step 68250/102583; acc: 82.20; ppl: 1.77; xent: 0.57; lr: 0.00003; 1949/2798 tok/s; 7018 sec
  1477. [2019-02-04 19:47:34,235 INFO] Step 68300/102583; acc: 83.35; ppl: 1.73; xent: 0.55; lr: 0.00003; 1979/2664 tok/s; 7024 sec
  1478. [2019-02-04 19:47:39,058 INFO] Step 68350/102583; acc: 82.06; ppl: 1.80; xent: 0.59; lr: 0.00003; 2159/2754 tok/s; 7028 sec
  1479. [2019-02-04 19:47:44,142 INFO] Step 68400/102583; acc: 82.67; ppl: 1.74; xent: 0.55; lr: 0.00003; 2034/2630 tok/s; 7034 sec
  1480. [2019-02-04 19:47:49,012 INFO] Step 68450/102583; acc: 82.05; ppl: 1.78; xent: 0.57; lr: 0.00003; 2199/2525 tok/s; 7038 sec
  1481. [2019-02-04 19:47:53,726 INFO] Step 68500/102583; acc: 82.29; ppl: 1.79; xent: 0.58; lr: 0.00003; 2095/2750 tok/s; 7043 sec
  1482. [2019-02-04 19:47:58,918 INFO] Step 68550/102583; acc: 80.60; ppl: 1.85; xent: 0.61; lr: 0.00003; 1934/2587 tok/s; 7048 sec
  1483. [2019-02-04 19:48:04,158 INFO] Step 68600/102583; acc: 82.24; ppl: 1.79; xent: 0.58; lr: 0.00003; 2041/2644 tok/s; 7054 sec
  1484. [2019-02-04 19:48:08,935 INFO] Step 68650/102583; acc: 83.54; ppl: 1.68; xent: 0.52; lr: 0.00003; 2160/2717 tok/s; 7058 sec
  1485. [2019-02-04 19:48:14,146 INFO] Step 68700/102583; acc: 80.55; ppl: 1.89; xent: 0.63; lr: 0.00003; 1967/2496 tok/s; 7064 sec
  1486. [2019-02-04 19:48:18,926 INFO] Step 68750/102583; acc: 82.38; ppl: 1.80; xent: 0.59; lr: 0.00003; 2161/2571 tok/s; 7068 sec
  1487. [2019-02-04 19:48:23,866 INFO] Step 68800/102583; acc: 81.27; ppl: 1.79; xent: 0.58; lr: 0.00003; 1968/2776 tok/s; 7073 sec
  1488. [2019-02-04 19:48:28,887 INFO] Step 68850/102583; acc: 83.63; ppl: 1.73; xent: 0.55; lr: 0.00003; 2148/2549 tok/s; 7078 sec
  1489. [2019-02-04 19:48:33,893 INFO] Step 68900/102583; acc: 81.80; ppl: 1.79; xent: 0.58; lr: 0.00003; 2058/2682 tok/s; 7083 sec
  1490. [2019-02-04 19:48:38,561 INFO] Step 68950/102583; acc: 82.34; ppl: 1.80; xent: 0.59; lr: 0.00003; 2214/2476 tok/s; 7088 sec
  1491. [2019-02-04 19:48:43,682 INFO] Step 69000/102583; acc: 82.41; ppl: 1.72; xent: 0.54; lr: 0.00003; 1828/2784 tok/s; 7093 sec
  1492. [2019-02-04 19:48:48,922 INFO] Step 69050/102583; acc: 82.13; ppl: 1.79; xent: 0.58; lr: 0.00003; 2014/2434 tok/s; 7098 sec
  1493. [2019-02-04 19:48:53,942 INFO] Step 69100/102583; acc: 82.98; ppl: 1.75; xent: 0.56; lr: 0.00003; 2157/2592 tok/s; 7103 sec
  1494. [2019-02-04 19:48:59,107 INFO] Step 69150/102583; acc: 80.03; ppl: 1.88; xent: 0.63; lr: 0.00003; 1851/2671 tok/s; 7108 sec
  1495. [2019-02-04 19:49:04,011 INFO] Step 69200/102583; acc: 84.35; ppl: 1.69; xent: 0.52; lr: 0.00003; 2017/2645 tok/s; 7113 sec
  1496. [2019-02-04 19:49:09,008 INFO] Step 69250/102583; acc: 80.57; ppl: 1.86; xent: 0.62; lr: 0.00003; 2082/2547 tok/s; 7118 sec
  1497. [2019-02-04 19:49:13,570 INFO] Step 69300/102583; acc: 81.31; ppl: 1.83; xent: 0.60; lr: 0.00003; 2206/2606 tok/s; 7123 sec
  1498. [2019-02-04 19:49:18,778 INFO] Step 69350/102583; acc: 81.86; ppl: 1.79; xent: 0.58; lr: 0.00003; 1993/2741 tok/s; 7128 sec
  1499. [2019-02-04 19:49:23,874 INFO] Step 69400/102583; acc: 82.74; ppl: 1.75; xent: 0.56; lr: 0.00003; 2053/2653 tok/s; 7133 sec
  1500. [2019-02-04 19:49:29,069 INFO] Step 69450/102583; acc: 83.66; ppl: 1.71; xent: 0.53; lr: 0.00003; 2152/2337 tok/s; 7138 sec
  1501. [2019-02-04 19:49:34,001 INFO] Step 69500/102583; acc: 80.43; ppl: 1.83; xent: 0.61; lr: 0.00003; 1840/2827 tok/s; 7143 sec
  1502. [2019-02-04 19:49:38,803 INFO] Step 69550/102583; acc: 81.40; ppl: 1.82; xent: 0.60; lr: 0.00003; 1898/2691 tok/s; 7148 sec
  1503. [2019-02-04 19:49:43,865 INFO] Step 69600/102583; acc: 84.42; ppl: 1.67; xent: 0.51; lr: 0.00003; 2253/2531 tok/s; 7153 sec
  1504. [2019-02-04 19:49:48,557 INFO] Step 69650/102583; acc: 81.02; ppl: 1.87; xent: 0.63; lr: 0.00003; 1912/2622 tok/s; 7158 sec
  1505. [2019-02-04 19:49:53,709 INFO] Step 69700/102583; acc: 82.19; ppl: 1.77; xent: 0.57; lr: 0.00003; 2072/2534 tok/s; 7163 sec
  1506. [2019-02-04 19:49:58,789 INFO] Step 69750/102583; acc: 82.43; ppl: 1.76; xent: 0.56; lr: 0.00003; 2085/2728 tok/s; 7168 sec
  1507. [2019-02-04 19:50:03,602 INFO] Step 69800/102583; acc: 81.18; ppl: 1.84; xent: 0.61; lr: 0.00003; 1963/2639 tok/s; 7173 sec
  1508. [2019-02-04 19:50:08,593 INFO] Step 69850/102583; acc: 82.49; ppl: 1.75; xent: 0.56; lr: 0.00003; 2017/2732 tok/s; 7178 sec
  1509. [2019-02-04 19:50:13,606 INFO] Step 69900/102583; acc: 83.57; ppl: 1.73; xent: 0.55; lr: 0.00003; 2150/2558 tok/s; 7183 sec
  1510. [2019-02-04 19:50:18,577 INFO] Step 69950/102583; acc: 79.97; ppl: 1.90; xent: 0.64; lr: 0.00003; 1885/2518 tok/s; 7188 sec
  1511. [2019-02-04 19:50:23,355 INFO] Step 70000/102583; acc: 82.92; ppl: 1.74; xent: 0.55; lr: 0.00001; 2169/2727 tok/s; 7193 sec
  1512. [2019-02-04 19:50:23,384 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  1513. [2019-02-04 19:50:24,685 INFO] Validation perplexity: 1.74237
  1514. [2019-02-04 19:50:24,686 INFO] Validation accuracy: 83.1895
  1515. [2019-02-04 19:50:24,687 INFO] Saving checkpoint models/ted1_step_70000.pt
  1516. [2019-02-04 19:50:36,050 INFO] Step 70050/102583; acc: 81.49; ppl: 1.83; xent: 0.60; lr: 0.00001; 784/1010 tok/s; 7205 sec
  1517. [2019-02-04 19:50:41,219 INFO] Step 70100/102583; acc: 83.74; ppl: 1.71; xent: 0.54; lr: 0.00001; 2194/2511 tok/s; 7211 sec
  1518. [2019-02-04 19:50:45,906 INFO] Step 70150/102583; acc: 80.53; ppl: 1.89; xent: 0.63; lr: 0.00001; 1861/2729 tok/s; 7215 sec
  1519. [2019-02-04 19:50:51,012 INFO] Step 70200/102583; acc: 82.59; ppl: 1.76; xent: 0.56; lr: 0.00001; 2068/2643 tok/s; 7220 sec
  1520. [2019-02-04 19:50:55,807 INFO] Step 70250/102583; acc: 82.33; ppl: 1.81; xent: 0.59; lr: 0.00001; 2021/2548 tok/s; 7225 sec
  1521. [2019-02-04 19:51:00,825 INFO] Step 70300/102583; acc: 83.84; ppl: 1.68; xent: 0.52; lr: 0.00001; 2032/2636 tok/s; 7230 sec
  1522. [2019-02-04 19:51:05,688 INFO] Step 70350/102583; acc: 81.44; ppl: 1.83; xent: 0.61; lr: 0.00001; 2196/2510 tok/s; 7235 sec
  1523. [2019-02-04 19:51:11,409 INFO] Step 70400/102583; acc: 83.18; ppl: 1.70; xent: 0.53; lr: 0.00001; 2145/2618 tok/s; 7241 sec
  1524. [2019-02-04 19:51:16,487 INFO] Step 70450/102583; acc: 80.81; ppl: 1.82; xent: 0.60; lr: 0.00001; 1840/2804 tok/s; 7246 sec
  1525. [2019-02-04 19:51:21,594 INFO] Step 70500/102583; acc: 82.81; ppl: 1.76; xent: 0.56; lr: 0.00001; 2074/2554 tok/s; 7251 sec
  1526. [2019-02-04 19:51:26,296 INFO] Step 70550/102583; acc: 82.95; ppl: 1.74; xent: 0.55; lr: 0.00001; 2103/2618 tok/s; 7256 sec
  1527. [2019-02-04 19:51:31,390 INFO] Step 70600/102583; acc: 82.56; ppl: 1.74; xent: 0.56; lr: 0.00001; 2072/2637 tok/s; 7261 sec
  1528. [2019-02-04 19:51:36,226 INFO] Step 70650/102583; acc: 80.77; ppl: 1.84; xent: 0.61; lr: 0.00001; 2054/2656 tok/s; 7266 sec
  1529. [2019-02-04 19:51:41,110 INFO] Step 70700/102583; acc: 81.97; ppl: 1.81; xent: 0.59; lr: 0.00001; 1901/2675 tok/s; 7270 sec
  1530. [2019-02-04 19:51:45,941 INFO] Step 70750/102583; acc: 83.35; ppl: 1.72; xent: 0.54; lr: 0.00001; 2178/2617 tok/s; 7275 sec
  1531. [2019-02-04 19:51:51,191 INFO] Step 70800/102583; acc: 82.80; ppl: 1.71; xent: 0.54; lr: 0.00001; 2068/2554 tok/s; 7281 sec
  1532. [2019-02-04 19:51:56,415 INFO] Step 70850/102583; acc: 80.84; ppl: 1.87; xent: 0.63; lr: 0.00001; 1814/2767 tok/s; 7286 sec
  1533. [2019-02-04 19:52:01,040 INFO] Step 70900/102583; acc: 81.97; ppl: 1.81; xent: 0.59; lr: 0.00001; 2128/2521 tok/s; 7290 sec
  1534. [2019-02-04 19:52:06,004 INFO] Step 70950/102583; acc: 82.94; ppl: 1.71; xent: 0.54; lr: 0.00001; 2033/2768 tok/s; 7295 sec
  1535. [2019-02-04 19:52:11,077 INFO] Step 71000/102583; acc: 82.92; ppl: 1.75; xent: 0.56; lr: 0.00001; 1954/2551 tok/s; 7300 sec
  1536. [2019-02-04 19:52:15,999 INFO] Step 71050/102583; acc: 82.01; ppl: 1.78; xent: 0.58; lr: 0.00001; 2181/2668 tok/s; 7305 sec
  1537. [2019-02-04 19:52:21,043 INFO] Step 71100/102583; acc: 82.18; ppl: 1.78; xent: 0.58; lr: 0.00001; 2130/2565 tok/s; 7310 sec
  1538. [2019-02-04 19:52:26,343 INFO] Step 71150/102583; acc: 83.54; ppl: 1.71; xent: 0.54; lr: 0.00001; 2056/2517 tok/s; 7316 sec
  1539. [2019-02-04 19:52:30,866 INFO] Step 71200/102583; acc: 81.24; ppl: 1.85; xent: 0.62; lr: 0.00001; 2078/2640 tok/s; 7320 sec
  1540. [2019-02-04 19:52:35,630 INFO] Step 71250/102583; acc: 82.27; ppl: 1.76; xent: 0.57; lr: 0.00001; 2069/2597 tok/s; 7325 sec
  1541. [2019-02-04 19:52:40,859 INFO] Step 71300/102583; acc: 82.81; ppl: 1.77; xent: 0.57; lr: 0.00001; 1923/2607 tok/s; 7330 sec
  1542. [2019-02-04 19:52:46,119 INFO] Step 71350/102583; acc: 81.93; ppl: 1.77; xent: 0.57; lr: 0.00001; 1975/2603 tok/s; 7336 sec
  1543. [2019-02-04 19:52:52,832 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1544. [2019-02-04 19:52:56,934 INFO] Step 71400/102583; acc: 82.79; ppl: 1.76; xent: 0.56; lr: 0.00001; 998/1213 tok/s; 7346 sec
  1545. [2019-02-04 19:53:01,458 INFO] Step 71450/102583; acc: 81.07; ppl: 1.89; xent: 0.64; lr: 0.00001; 2095/2528 tok/s; 7351 sec
  1546. [2019-02-04 19:53:06,379 INFO] Step 71500/102583; acc: 83.03; ppl: 1.73; xent: 0.55; lr: 0.00001; 1937/2790 tok/s; 7356 sec
  1547. [2019-02-04 19:53:11,557 INFO] Step 71550/102583; acc: 82.91; ppl: 1.74; xent: 0.56; lr: 0.00001; 2083/2610 tok/s; 7361 sec
  1548. [2019-02-04 19:53:16,370 INFO] Step 71600/102583; acc: 82.35; ppl: 1.78; xent: 0.57; lr: 0.00001; 2052/2839 tok/s; 7366 sec
  1549. [2019-02-04 19:53:21,527 INFO] Step 71650/102583; acc: 82.59; ppl: 1.74; xent: 0.55; lr: 0.00001; 2041/2532 tok/s; 7371 sec
  1550. [2019-02-04 19:53:26,356 INFO] Step 71700/102583; acc: 82.59; ppl: 1.76; xent: 0.56; lr: 0.00001; 2219/2574 tok/s; 7376 sec
  1551. [2019-02-04 19:53:31,142 INFO] Step 71750/102583; acc: 81.86; ppl: 1.83; xent: 0.60; lr: 0.00001; 2128/2697 tok/s; 7381 sec
  1552. [2019-02-04 19:53:36,385 INFO] Step 71800/102583; acc: 80.86; ppl: 1.84; xent: 0.61; lr: 0.00001; 1790/2629 tok/s; 7386 sec
  1553. [2019-02-04 19:53:41,264 INFO] Step 71850/102583; acc: 82.67; ppl: 1.76; xent: 0.57; lr: 0.00001; 2236/2602 tok/s; 7391 sec
  1554. [2019-02-04 19:53:46,183 INFO] Step 71900/102583; acc: 83.12; ppl: 1.70; xent: 0.53; lr: 0.00001; 1958/2764 tok/s; 7396 sec
  1555. [2019-02-04 19:53:51,342 INFO] Step 71950/102583; acc: 81.34; ppl: 1.84; xent: 0.61; lr: 0.00001; 2143/2481 tok/s; 7401 sec
  1556. [2019-02-04 19:53:56,184 INFO] Step 72000/102583; acc: 82.47; ppl: 1.78; xent: 0.58; lr: 0.00001; 2057/2629 tok/s; 7406 sec
  1557. [2019-02-04 19:54:01,020 INFO] Step 72050/102583; acc: 80.76; ppl: 1.82; xent: 0.60; lr: 0.00001; 1994/2759 tok/s; 7410 sec
  1558. [2019-02-04 19:54:06,176 INFO] Step 72100/102583; acc: 84.59; ppl: 1.68; xent: 0.52; lr: 0.00001; 2251/2464 tok/s; 7416 sec
  1559. [2019-02-04 19:54:11,249 INFO] Step 72150/102583; acc: 81.29; ppl: 1.80; xent: 0.59; lr: 0.00001; 1937/2770 tok/s; 7421 sec
  1560. [2019-02-04 19:54:15,750 INFO] Step 72200/102583; acc: 82.74; ppl: 1.78; xent: 0.58; lr: 0.00001; 2161/2506 tok/s; 7425 sec
  1561. [2019-02-04 19:54:20,847 INFO] Step 72250/102583; acc: 82.28; ppl: 1.73; xent: 0.55; lr: 0.00001; 1885/2667 tok/s; 7430 sec
  1562. [2019-02-04 19:54:26,032 INFO] Step 72300/102583; acc: 82.08; ppl: 1.79; xent: 0.58; lr: 0.00001; 2012/2496 tok/s; 7435 sec
  1563. [2019-02-04 19:54:31,106 INFO] Step 72350/102583; acc: 82.57; ppl: 1.76; xent: 0.57; lr: 0.00001; 2129/2635 tok/s; 7440 sec
  1564. [2019-02-04 19:54:36,413 INFO] Step 72400/102583; acc: 81.20; ppl: 1.81; xent: 0.59; lr: 0.00001; 1866/2702 tok/s; 7446 sec
  1565. [2019-02-04 19:54:41,215 INFO] Step 72450/102583; acc: 84.08; ppl: 1.71; xent: 0.54; lr: 0.00001; 2068/2625 tok/s; 7451 sec
  1566. [2019-02-04 19:54:46,188 INFO] Step 72500/102583; acc: 80.26; ppl: 1.86; xent: 0.62; lr: 0.00001; 2092/2539 tok/s; 7456 sec
  1567. [2019-02-04 19:54:50,866 INFO] Step 72550/102583; acc: 82.32; ppl: 1.79; xent: 0.58; lr: 0.00001; 2200/2578 tok/s; 7460 sec
  1568. [2019-02-04 19:54:55,779 INFO] Step 72600/102583; acc: 80.69; ppl: 1.85; xent: 0.61; lr: 0.00001; 1943/2738 tok/s; 7465 sec
  1569. [2019-02-04 19:55:01,095 INFO] Step 72650/102583; acc: 83.77; ppl: 1.70; xent: 0.53; lr: 0.00001; 1990/2569 tok/s; 7470 sec
  1570. [2019-02-04 19:55:06,237 INFO] Step 72700/102583; acc: 83.09; ppl: 1.73; xent: 0.55; lr: 0.00001; 2185/2405 tok/s; 7476 sec
  1571. [2019-02-04 19:55:10,986 INFO] Step 72750/102583; acc: 80.58; ppl: 1.82; xent: 0.60; lr: 0.00001; 1934/2848 tok/s; 7480 sec
  1572. [2019-02-04 19:55:15,796 INFO] Step 72800/102583; acc: 81.73; ppl: 1.81; xent: 0.59; lr: 0.00001; 1855/2729 tok/s; 7485 sec
  1573. [2019-02-04 19:55:20,871 INFO] Step 72850/102583; acc: 84.24; ppl: 1.68; xent: 0.52; lr: 0.00001; 2251/2478 tok/s; 7490 sec
  1574. [2019-02-04 19:55:25,704 INFO] Step 72900/102583; acc: 80.57; ppl: 1.88; xent: 0.63; lr: 0.00001; 1832/2683 tok/s; 7495 sec
  1575. [2019-02-04 19:55:30,777 INFO] Step 72950/102583; acc: 83.23; ppl: 1.70; xent: 0.53; lr: 0.00001; 2173/2550 tok/s; 7500 sec
  1576. [2019-02-04 19:55:35,802 INFO] Step 73000/102583; acc: 81.47; ppl: 1.80; xent: 0.59; lr: 0.00001; 2069/2711 tok/s; 7505 sec
  1577. [2019-02-04 19:55:40,642 INFO] Step 73050/102583; acc: 82.36; ppl: 1.78; xent: 0.58; lr: 0.00001; 2045/2585 tok/s; 7510 sec
  1578. [2019-02-04 19:55:45,464 INFO] Step 73100/102583; acc: 81.81; ppl: 1.80; xent: 0.59; lr: 0.00001; 1915/2720 tok/s; 7515 sec
  1579. [2019-02-04 19:55:50,561 INFO] Step 73150/102583; acc: 83.59; ppl: 1.71; xent: 0.54; lr: 0.00001; 2239/2577 tok/s; 7520 sec
  1580. [2019-02-04 19:55:55,594 INFO] Step 73200/102583; acc: 80.73; ppl: 1.87; xent: 0.63; lr: 0.00001; 1825/2579 tok/s; 7525 sec
  1581. [2019-02-04 19:56:00,451 INFO] Step 73250/102583; acc: 82.53; ppl: 1.74; xent: 0.56; lr: 0.00001; 2216/2591 tok/s; 7530 sec
  1582. [2019-02-04 19:56:05,436 INFO] Step 73300/102583; acc: 81.75; ppl: 1.82; xent: 0.60; lr: 0.00001; 1969/2626 tok/s; 7535 sec
  1583. [2019-02-04 19:56:10,598 INFO] Step 73350/102583; acc: 83.69; ppl: 1.72; xent: 0.54; lr: 0.00001; 2070/2437 tok/s; 7540 sec
  1584. [2019-02-04 19:56:15,373 INFO] Step 73400/102583; acc: 81.40; ppl: 1.84; xent: 0.61; lr: 0.00001; 2030/2743 tok/s; 7545 sec
  1585. [2019-02-04 19:56:20,449 INFO] Step 73450/102583; acc: 82.39; ppl: 1.75; xent: 0.56; lr: 0.00001; 1964/2664 tok/s; 7550 sec
  1586. [2019-02-04 19:56:25,165 INFO] Step 73500/102583; acc: 81.14; ppl: 1.88; xent: 0.63; lr: 0.00001; 2045/2609 tok/s; 7555 sec
  1587. [2019-02-04 19:56:30,370 INFO] Step 73550/102583; acc: 84.55; ppl: 1.64; xent: 0.50; lr: 0.00001; 2054/2547 tok/s; 7560 sec
  1588. [2019-02-04 19:56:35,010 INFO] Step 73600/102583; acc: 81.11; ppl: 1.87; xent: 0.63; lr: 0.00001; 2168/2582 tok/s; 7564 sec
  1589. [2019-02-04 19:56:40,151 INFO] Step 73650/102583; acc: 83.37; ppl: 1.70; xent: 0.53; lr: 0.00001; 2082/2627 tok/s; 7570 sec
  1590. [2019-02-04 19:56:45,323 INFO] Step 73700/102583; acc: 82.14; ppl: 1.75; xent: 0.56; lr: 0.00001; 1885/2762 tok/s; 7575 sec
  1591. [2019-02-04 19:56:50,176 INFO] Step 73750/102583; acc: 81.82; ppl: 1.81; xent: 0.59; lr: 0.00001; 2135/2577 tok/s; 7580 sec
  1592. [2019-02-04 19:56:54,906 INFO] Step 73800/102583; acc: 83.86; ppl: 1.70; xent: 0.53; lr: 0.00001; 2152/2565 tok/s; 7584 sec
  1593. [2019-02-04 19:56:59,986 INFO] Step 73850/102583; acc: 81.52; ppl: 1.78; xent: 0.58; lr: 0.00001; 2021/2681 tok/s; 7589 sec
  1594. [2019-02-04 19:57:04,731 INFO] Step 73900/102583; acc: 80.48; ppl: 1.85; xent: 0.62; lr: 0.00001; 2013/2653 tok/s; 7594 sec
  1595. [2019-02-04 19:57:09,746 INFO] Step 73950/102583; acc: 83.47; ppl: 1.73; xent: 0.55; lr: 0.00001; 1994/2630 tok/s; 7599 sec
  1596. [2019-02-04 19:57:14,600 INFO] Step 74000/102583; acc: 82.70; ppl: 1.75; xent: 0.56; lr: 0.00001; 2147/2655 tok/s; 7604 sec
  1597. [2019-02-04 19:57:19,814 INFO] Step 74050/102583; acc: 82.38; ppl: 1.74; xent: 0.55; lr: 0.00001; 2016/2599 tok/s; 7609 sec
  1598. [2019-02-04 19:57:25,070 INFO] Step 74100/102583; acc: 80.81; ppl: 1.86; xent: 0.62; lr: 0.00001; 1865/2705 tok/s; 7614 sec
  1599. [2019-02-04 19:57:29,802 INFO] Step 74150/102583; acc: 82.15; ppl: 1.80; xent: 0.59; lr: 0.00001; 2084/2536 tok/s; 7619 sec
  1600. [2019-02-04 19:57:34,567 INFO] Step 74200/102583; acc: 83.46; ppl: 1.71; xent: 0.54; lr: 0.00001; 2019/2793 tok/s; 7624 sec
  1601. [2019-02-04 19:57:39,880 INFO] Step 74250/102583; acc: 83.33; ppl: 1.72; xent: 0.54; lr: 0.00001; 2027/2520 tok/s; 7629 sec
  1602. [2019-02-04 19:57:44,628 INFO] Step 74300/102583; acc: 81.70; ppl: 1.80; xent: 0.59; lr: 0.00001; 2111/2732 tok/s; 7634 sec
  1603. [2019-02-04 19:57:49,730 INFO] Step 74350/102583; acc: 81.98; ppl: 1.80; xent: 0.59; lr: 0.00001; 2191/2529 tok/s; 7639 sec
  1604. [2019-02-04 19:57:55,147 INFO] Step 74400/102583; acc: 83.80; ppl: 1.70; xent: 0.53; lr: 0.00001; 2025/2520 tok/s; 7645 sec
  1605. [2019-02-04 19:57:59,718 INFO] Step 74450/102583; acc: 80.78; ppl: 1.87; xent: 0.63; lr: 0.00001; 2061/2627 tok/s; 7649 sec
  1606. [2019-02-04 19:58:04,459 INFO] Step 74500/102583; acc: 82.13; ppl: 1.77; xent: 0.57; lr: 0.00001; 2081/2677 tok/s; 7654 sec
  1607. [2019-02-04 19:58:09,780 INFO] Step 74550/102583; acc: 82.70; ppl: 1.75; xent: 0.56; lr: 0.00001; 1982/2561 tok/s; 7659 sec
  1608. [2019-02-04 19:58:14,880 INFO] Step 74600/102583; acc: 81.72; ppl: 1.79; xent: 0.58; lr: 0.00001; 1890/2616 tok/s; 7664 sec
  1609. [2019-02-04 19:58:21,484 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1610. [2019-02-04 19:58:26,183 INFO] Step 74650/102583; acc: 82.46; ppl: 1.76; xent: 0.56; lr: 0.00001; 972/1171 tok/s; 7676 sec
  1611. [2019-02-04 19:58:30,643 INFO] Step 74700/102583; acc: 81.15; ppl: 1.89; xent: 0.63; lr: 0.00001; 2110/2569 tok/s; 7680 sec
  1612. [2019-02-04 19:58:35,720 INFO] Step 74750/102583; acc: 83.41; ppl: 1.70; xent: 0.53; lr: 0.00001; 2003/2836 tok/s; 7685 sec
  1613. [2019-02-04 19:58:40,841 INFO] Step 74800/102583; acc: 81.96; ppl: 1.79; xent: 0.58; lr: 0.00001; 1983/2583 tok/s; 7690 sec
  1614. [2019-02-04 19:58:45,651 INFO] Step 74850/102583; acc: 83.00; ppl: 1.74; xent: 0.56; lr: 0.00001; 2067/2806 tok/s; 7695 sec
  1615. [2019-02-04 19:58:50,690 INFO] Step 74900/102583; acc: 82.30; ppl: 1.76; xent: 0.57; lr: 0.00001; 2137/2500 tok/s; 7700 sec
  1616. [2019-02-04 19:58:55,603 INFO] Step 74950/102583; acc: 83.27; ppl: 1.72; xent: 0.54; lr: 0.00001; 2234/2608 tok/s; 7705 sec
  1617. [2019-02-04 19:59:00,478 INFO] Step 75000/102583; acc: 81.25; ppl: 1.84; xent: 0.61; lr: 0.00001; 2039/2641 tok/s; 7710 sec
  1618. [2019-02-04 19:59:00,479 INFO] Saving checkpoint models/ted1_step_75000.pt
  1619. [2019-02-04 19:59:11,515 INFO] Step 75050/102583; acc: 80.48; ppl: 1.86; xent: 0.62; lr: 0.00001; 826/1239 tok/s; 7721 sec
  1620. [2019-02-04 19:59:16,443 INFO] Step 75100/102583; acc: 82.50; ppl: 1.77; xent: 0.57; lr: 0.00001; 2148/2651 tok/s; 7726 sec
  1621. [2019-02-04 19:59:21,547 INFO] Step 75150/102583; acc: 82.88; ppl: 1.71; xent: 0.54; lr: 0.00001; 1913/2741 tok/s; 7731 sec
  1622. [2019-02-04 19:59:26,368 INFO] Step 75200/102583; acc: 82.24; ppl: 1.80; xent: 0.59; lr: 0.00001; 2309/2463 tok/s; 7736 sec
  1623. [2019-02-04 19:59:31,197 INFO] Step 75250/102583; acc: 81.49; ppl: 1.83; xent: 0.60; lr: 0.00001; 2022/2619 tok/s; 7741 sec
  1624. [2019-02-04 19:59:36,121 INFO] Step 75300/102583; acc: 81.52; ppl: 1.81; xent: 0.59; lr: 0.00001; 1954/2673 tok/s; 7746 sec
  1625. [2019-02-04 19:59:41,253 INFO] Step 75350/102583; acc: 84.42; ppl: 1.66; xent: 0.50; lr: 0.00001; 2253/2580 tok/s; 7751 sec
  1626. [2019-02-04 19:59:46,239 INFO] Step 75400/102583; acc: 80.82; ppl: 1.84; xent: 0.61; lr: 0.00001; 1932/2726 tok/s; 7756 sec
  1627. [2019-02-04 19:59:50,903 INFO] Step 75450/102583; acc: 82.61; ppl: 1.77; xent: 0.57; lr: 0.00001; 2086/2452 tok/s; 7760 sec
  1628. [2019-02-04 19:59:55,983 INFO] Step 75500/102583; acc: 83.30; ppl: 1.69; xent: 0.53; lr: 0.00001; 1980/2781 tok/s; 7765 sec
  1629. [2019-02-04 20:00:01,028 INFO] Step 75550/102583; acc: 81.81; ppl: 1.82; xent: 0.60; lr: 0.00001; 2043/2432 tok/s; 7770 sec
  1630. [2019-02-04 20:00:06,101 INFO] Step 75600/102583; acc: 82.22; ppl: 1.77; xent: 0.57; lr: 0.00001; 2070/2627 tok/s; 7775 sec
  1631. [2019-02-04 20:00:11,492 INFO] Step 75650/102583; acc: 81.85; ppl: 1.76; xent: 0.57; lr: 0.00001; 1896/2732 tok/s; 7781 sec
  1632. [2019-02-04 20:00:16,128 INFO] Step 75700/102583; acc: 82.95; ppl: 1.76; xent: 0.57; lr: 0.00001; 2129/2528 tok/s; 7786 sec
  1633. [2019-02-04 20:00:21,182 INFO] Step 75750/102583; acc: 81.03; ppl: 1.84; xent: 0.61; lr: 0.00001; 2110/2569 tok/s; 7791 sec
  1634. [2019-02-04 20:00:26,127 INFO] Step 75800/102583; acc: 81.54; ppl: 1.80; xent: 0.59; lr: 0.00001; 1980/2705 tok/s; 7796 sec
  1635. [2019-02-04 20:00:30,827 INFO] Step 75850/102583; acc: 81.07; ppl: 1.85; xent: 0.61; lr: 0.00001; 2091/2579 tok/s; 7800 sec
  1636. [2019-02-04 20:00:36,107 INFO] Step 75900/102583; acc: 83.57; ppl: 1.70; xent: 0.53; lr: 0.00001; 1968/2620 tok/s; 7805 sec
  1637. [2019-02-04 20:00:41,247 INFO] Step 75950/102583; acc: 83.56; ppl: 1.72; xent: 0.54; lr: 0.00001; 2171/2397 tok/s; 7811 sec
  1638. [2019-02-04 20:00:46,008 INFO] Step 76000/102583; acc: 80.77; ppl: 1.82; xent: 0.60; lr: 0.00001; 1916/2871 tok/s; 7815 sec
  1639. [2019-02-04 20:00:51,081 INFO] Step 76050/102583; acc: 82.17; ppl: 1.77; xent: 0.57; lr: 0.00001; 2006/2622 tok/s; 7820 sec
  1640. [2019-02-04 20:00:55,963 INFO] Step 76100/102583; acc: 83.15; ppl: 1.75; xent: 0.56; lr: 0.00001; 2154/2515 tok/s; 7825 sec
  1641. [2019-02-04 20:01:00,641 INFO] Step 76150/102583; acc: 80.85; ppl: 1.88; xent: 0.63; lr: 0.00001; 1832/2678 tok/s; 7830 sec
  1642. [2019-02-04 20:01:05,891 INFO] Step 76200/102583; acc: 83.24; ppl: 1.68; xent: 0.52; lr: 0.00001; 2184/2621 tok/s; 7835 sec
  1643. [2019-02-04 20:01:10,833 INFO] Step 76250/102583; acc: 81.20; ppl: 1.82; xent: 0.60; lr: 0.00001; 2014/2697 tok/s; 7840 sec
  1644. [2019-02-04 20:01:15,767 INFO] Step 76300/102583; acc: 82.63; ppl: 1.77; xent: 0.57; lr: 0.00001; 2066/2513 tok/s; 7845 sec
  1645. [2019-02-04 20:01:20,821 INFO] Step 76350/102583; acc: 82.41; ppl: 1.76; xent: 0.57; lr: 0.00001; 1891/2720 tok/s; 7850 sec
  1646. [2019-02-04 20:01:25,684 INFO] Step 76400/102583; acc: 82.12; ppl: 1.78; xent: 0.58; lr: 0.00001; 2216/2593 tok/s; 7855 sec
  1647. [2019-02-04 20:01:30,715 INFO] Step 76450/102583; acc: 80.92; ppl: 1.85; xent: 0.62; lr: 0.00001; 1841/2659 tok/s; 7860 sec
  1648. [2019-02-04 20:01:35,582 INFO] Step 76500/102583; acc: 83.08; ppl: 1.72; xent: 0.54; lr: 0.00001; 2248/2544 tok/s; 7865 sec
  1649. [2019-02-04 20:01:40,542 INFO] Step 76550/102583; acc: 81.80; ppl: 1.81; xent: 0.59; lr: 0.00001; 1996/2619 tok/s; 7870 sec
  1650. [2019-02-04 20:01:45,706 INFO] Step 76600/102583; acc: 83.20; ppl: 1.75; xent: 0.56; lr: 0.00001; 2006/2508 tok/s; 7875 sec
  1651. [2019-02-04 20:01:50,500 INFO] Step 76650/102583; acc: 81.18; ppl: 1.84; xent: 0.61; lr: 0.00001; 2012/2690 tok/s; 7880 sec
  1652. [2019-02-04 20:01:55,481 INFO] Step 76700/102583; acc: 82.74; ppl: 1.74; xent: 0.55; lr: 0.00001; 2049/2647 tok/s; 7885 sec
  1653. [2019-02-04 20:02:00,332 INFO] Step 76750/102583; acc: 81.57; ppl: 1.84; xent: 0.61; lr: 0.00001; 2075/2612 tok/s; 7890 sec
  1654. [2019-02-04 20:02:05,419 INFO] Step 76800/102583; acc: 83.78; ppl: 1.68; xent: 0.52; lr: 0.00001; 2095/2546 tok/s; 7895 sec
  1655. [2019-02-04 20:02:10,061 INFO] Step 76850/102583; acc: 81.78; ppl: 1.82; xent: 0.60; lr: 0.00001; 2112/2593 tok/s; 7899 sec
  1656. [2019-02-04 20:02:15,172 INFO] Step 76900/102583; acc: 82.62; ppl: 1.73; xent: 0.55; lr: 0.00001; 2069/2603 tok/s; 7905 sec
  1657. [2019-02-04 20:02:20,506 INFO] Step 76950/102583; acc: 82.38; ppl: 1.73; xent: 0.55; lr: 0.00001; 1909/2752 tok/s; 7910 sec
  1658. [2019-02-04 20:02:25,342 INFO] Step 77000/102583; acc: 81.29; ppl: 1.83; xent: 0.60; lr: 0.00001; 1990/2661 tok/s; 7915 sec
  1659. [2019-02-04 20:02:29,995 INFO] Step 77050/102583; acc: 83.42; ppl: 1.73; xent: 0.55; lr: 0.00001; 2217/2493 tok/s; 7919 sec
  1660. [2019-02-04 20:02:35,022 INFO] Step 77100/102583; acc: 81.58; ppl: 1.79; xent: 0.58; lr: 0.00001; 2078/2715 tok/s; 7924 sec
  1661. [2019-02-04 20:02:39,830 INFO] Step 77150/102583; acc: 81.48; ppl: 1.81; xent: 0.59; lr: 0.00001; 1973/2663 tok/s; 7929 sec
  1662. [2019-02-04 20:02:44,851 INFO] Step 77200/102583; acc: 82.48; ppl: 1.76; xent: 0.57; lr: 0.00001; 1979/2653 tok/s; 7934 sec
  1663. [2019-02-04 20:02:49,577 INFO] Step 77250/102583; acc: 83.28; ppl: 1.72; xent: 0.54; lr: 0.00001; 2222/2602 tok/s; 7939 sec
  1664. [2019-02-04 20:02:54,797 INFO] Step 77300/102583; acc: 81.58; ppl: 1.79; xent: 0.58; lr: 0.00001; 1965/2594 tok/s; 7944 sec
  1665. [2019-02-04 20:03:00,034 INFO] Step 77350/102583; acc: 81.07; ppl: 1.85; xent: 0.61; lr: 0.00001; 1896/2684 tok/s; 7949 sec
  1666. [2019-02-04 20:03:04,779 INFO] Step 77400/102583; acc: 82.74; ppl: 1.77; xent: 0.57; lr: 0.00001; 2146/2539 tok/s; 7954 sec
  1667. [2019-02-04 20:03:09,670 INFO] Step 77450/102583; acc: 83.64; ppl: 1.68; xent: 0.52; lr: 0.00001; 1959/2795 tok/s; 7959 sec
  1668. [2019-02-04 20:03:14,888 INFO] Step 77500/102583; acc: 83.19; ppl: 1.73; xent: 0.55; lr: 0.00001; 2109/2533 tok/s; 7964 sec
  1669. [2019-02-04 20:03:19,784 INFO] Step 77550/102583; acc: 82.08; ppl: 1.76; xent: 0.57; lr: 0.00001; 2071/2741 tok/s; 7969 sec
  1670. [2019-02-04 20:03:24,745 INFO] Step 77600/102583; acc: 81.61; ppl: 1.84; xent: 0.61; lr: 0.00001; 2175/2379 tok/s; 7974 sec
  1671. [2019-02-04 20:03:30,029 INFO] Step 77650/102583; acc: 82.98; ppl: 1.73; xent: 0.55; lr: 0.00001; 2006/2645 tok/s; 7979 sec
  1672. [2019-02-04 20:03:34,632 INFO] Step 77700/102583; acc: 80.69; ppl: 1.87; xent: 0.62; lr: 0.00001; 1981/2696 tok/s; 7984 sec
  1673. [2019-02-04 20:03:39,396 INFO] Step 77750/102583; acc: 82.37; ppl: 1.77; xent: 0.57; lr: 0.00001; 2137/2509 tok/s; 7989 sec
  1674. [2019-02-04 20:03:44,716 INFO] Step 77800/102583; acc: 81.81; ppl: 1.78; xent: 0.58; lr: 0.00001; 1863/2718 tok/s; 7994 sec
  1675. [2019-02-04 20:03:49,870 INFO] Step 77850/102583; acc: 82.65; ppl: 1.76; xent: 0.57; lr: 0.00001; 1931/2489 tok/s; 7999 sec
  1676. [2019-02-04 20:03:55,513 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1677. [2019-02-04 20:04:00,697 INFO] Step 77900/102583; acc: 82.93; ppl: 1.75; xent: 0.56; lr: 0.00001; 1004/1181 tok/s; 8010 sec
  1678. [2019-02-04 20:04:05,413 INFO] Step 77950/102583; acc: 82.15; ppl: 1.80; xent: 0.59; lr: 0.00001; 2097/2650 tok/s; 8015 sec
  1679. [2019-02-04 20:04:10,475 INFO] Step 78000/102583; acc: 83.40; ppl: 1.71; xent: 0.54; lr: 0.00001; 2032/2737 tok/s; 8020 sec
  1680. [2019-02-04 20:04:15,368 INFO] Step 78050/102583; acc: 81.58; ppl: 1.83; xent: 0.61; lr: 0.00001; 2034/2607 tok/s; 8025 sec
  1681. [2019-02-04 20:04:20,242 INFO] Step 78100/102583; acc: 82.58; ppl: 1.75; xent: 0.56; lr: 0.00001; 1967/2779 tok/s; 8030 sec
  1682. [2019-02-04 20:04:25,196 INFO] Step 78150/102583; acc: 82.60; ppl: 1.75; xent: 0.56; lr: 0.00001; 2203/2528 tok/s; 8035 sec
  1683. [2019-02-04 20:04:30,128 INFO] Step 78200/102583; acc: 82.86; ppl: 1.74; xent: 0.56; lr: 0.00001; 2173/2634 tok/s; 8040 sec
  1684. [2019-02-04 20:04:34,981 INFO] Step 78250/102583; acc: 81.56; ppl: 1.83; xent: 0.60; lr: 0.00001; 2048/2633 tok/s; 8044 sec
  1685. [2019-02-04 20:04:40,266 INFO] Step 78300/102583; acc: 80.88; ppl: 1.83; xent: 0.60; lr: 0.00001; 1840/2698 tok/s; 8050 sec
  1686. [2019-02-04 20:04:45,008 INFO] Step 78350/102583; acc: 82.63; ppl: 1.76; xent: 0.57; lr: 0.00001; 2146/2626 tok/s; 8054 sec
  1687. [2019-02-04 20:04:50,209 INFO] Step 78400/102583; acc: 82.64; ppl: 1.73; xent: 0.55; lr: 0.00001; 1932/2725 tok/s; 8060 sec
  1688. [2019-02-04 20:04:55,056 INFO] Step 78450/102583; acc: 82.03; ppl: 1.82; xent: 0.60; lr: 0.00001; 2237/2561 tok/s; 8064 sec
  1689. [2019-02-04 20:05:00,036 INFO] Step 78500/102583; acc: 81.46; ppl: 1.80; xent: 0.59; lr: 0.00001; 2040/2588 tok/s; 8069 sec
  1690. [2019-02-04 20:05:04,722 INFO] Step 78550/102583; acc: 81.60; ppl: 1.84; xent: 0.61; lr: 0.00001; 1957/2658 tok/s; 8074 sec
  1691. [2019-02-04 20:05:09,877 INFO] Step 78600/102583; acc: 83.70; ppl: 1.68; xent: 0.52; lr: 0.00001; 2305/2572 tok/s; 8079 sec
  1692. [2019-02-04 20:05:14,850 INFO] Step 78650/102583; acc: 82.31; ppl: 1.78; xent: 0.57; lr: 0.00001; 1924/2737 tok/s; 8084 sec
  1693. [2019-02-04 20:05:19,858 INFO] Step 78700/102583; acc: 82.57; ppl: 1.77; xent: 0.57; lr: 0.00001; 2046/2432 tok/s; 8089 sec
  1694. [2019-02-04 20:05:24,717 INFO] Step 78750/102583; acc: 82.47; ppl: 1.75; xent: 0.56; lr: 0.00001; 1989/2714 tok/s; 8094 sec
  1695. [2019-02-04 20:05:29,841 INFO] Step 78800/102583; acc: 82.29; ppl: 1.76; xent: 0.57; lr: 0.00001; 2100/2469 tok/s; 8099 sec
  1696. [2019-02-04 20:05:34,813 INFO] Step 78850/102583; acc: 82.04; ppl: 1.79; xent: 0.58; lr: 0.00001; 1989/2692 tok/s; 8104 sec
  1697. [2019-02-04 20:05:40,101 INFO] Step 78900/102583; acc: 82.00; ppl: 1.77; xent: 0.57; lr: 0.00001; 1927/2728 tok/s; 8109 sec
  1698. [2019-02-04 20:05:44,799 INFO] Step 78950/102583; acc: 82.53; ppl: 1.78; xent: 0.58; lr: 0.00001; 2101/2543 tok/s; 8114 sec
  1699. [2019-02-04 20:05:49,777 INFO] Step 79000/102583; acc: 81.44; ppl: 1.82; xent: 0.60; lr: 0.00001; 2129/2563 tok/s; 8119 sec
  1700. [2019-02-04 20:05:54,740 INFO] Step 79050/102583; acc: 81.17; ppl: 1.82; xent: 0.60; lr: 0.00001; 1974/2695 tok/s; 8124 sec
  1701. [2019-02-04 20:05:59,402 INFO] Step 79100/102583; acc: 81.35; ppl: 1.84; xent: 0.61; lr: 0.00001; 2136/2570 tok/s; 8129 sec
  1702. [2019-02-04 20:06:04,739 INFO] Step 79150/102583; acc: 83.42; ppl: 1.71; xent: 0.54; lr: 0.00001; 1923/2642 tok/s; 8134 sec
  1703. [2019-02-04 20:06:09,962 INFO] Step 79200/102583; acc: 83.86; ppl: 1.69; xent: 0.52; lr: 0.00001; 2192/2391 tok/s; 8139 sec
  1704. [2019-02-04 20:06:14,686 INFO] Step 79250/102583; acc: 80.10; ppl: 1.87; xent: 0.62; lr: 0.00001; 1894/2822 tok/s; 8144 sec
  1705. [2019-02-04 20:06:19,866 INFO] Step 79300/102583; acc: 82.76; ppl: 1.72; xent: 0.55; lr: 0.00001; 2053/2614 tok/s; 8149 sec
  1706. [2019-02-04 20:06:24,599 INFO] Step 79350/102583; acc: 82.33; ppl: 1.78; xent: 0.58; lr: 0.00001; 2090/2614 tok/s; 8154 sec
  1707. [2019-02-04 20:06:29,210 INFO] Step 79400/102583; acc: 80.95; ppl: 1.88; xent: 0.63; lr: 0.00001; 1901/2619 tok/s; 8159 sec
  1708. [2019-02-04 20:06:34,529 INFO] Step 79450/102583; acc: 83.03; ppl: 1.68; xent: 0.52; lr: 0.00001; 2130/2589 tok/s; 8164 sec
  1709. [2019-02-04 20:06:39,410 INFO] Step 79500/102583; acc: 80.84; ppl: 1.84; xent: 0.61; lr: 0.00001; 2068/2740 tok/s; 8169 sec
  1710. [2019-02-04 20:06:44,454 INFO] Step 79550/102583; acc: 83.08; ppl: 1.74; xent: 0.56; lr: 0.00001; 1970/2540 tok/s; 8174 sec
  1711. [2019-02-04 20:06:49,511 INFO] Step 79600/102583; acc: 82.83; ppl: 1.76; xent: 0.57; lr: 0.00001; 1953/2740 tok/s; 8179 sec
  1712. [2019-02-04 20:06:54,364 INFO] Step 79650/102583; acc: 82.00; ppl: 1.78; xent: 0.58; lr: 0.00001; 2185/2532 tok/s; 8184 sec
  1713. [2019-02-04 20:06:59,353 INFO] Step 79700/102583; acc: 81.33; ppl: 1.82; xent: 0.60; lr: 0.00001; 1904/2685 tok/s; 8189 sec
  1714. [2019-02-04 20:07:04,239 INFO] Step 79750/102583; acc: 82.54; ppl: 1.76; xent: 0.56; lr: 0.00001; 2204/2578 tok/s; 8194 sec
  1715. [2019-02-04 20:07:09,188 INFO] Step 79800/102583; acc: 82.64; ppl: 1.76; xent: 0.57; lr: 0.00001; 2053/2586 tok/s; 8199 sec
  1716. [2019-02-04 20:07:14,352 INFO] Step 79850/102583; acc: 82.55; ppl: 1.78; xent: 0.57; lr: 0.00001; 1932/2525 tok/s; 8204 sec
  1717. [2019-02-04 20:07:19,115 INFO] Step 79900/102583; acc: 81.67; ppl: 1.82; xent: 0.60; lr: 0.00001; 2045/2676 tok/s; 8209 sec
  1718. [2019-02-04 20:07:24,231 INFO] Step 79950/102583; acc: 83.14; ppl: 1.74; xent: 0.55; lr: 0.00001; 2102/2550 tok/s; 8214 sec
  1719. [2019-02-04 20:07:29,017 INFO] Step 80000/102583; acc: 81.47; ppl: 1.82; xent: 0.60; lr: 0.00001; 1994/2682 tok/s; 8218 sec
  1720. [2019-02-04 20:07:29,043 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  1721. [2019-02-04 20:07:30,344 INFO] Validation perplexity: 1.74359
  1722. [2019-02-04 20:07:30,345 INFO] Validation accuracy: 83.2174
  1723. [2019-02-04 20:07:30,346 INFO] Saving checkpoint models/ted1_step_80000.pt
  1724. [2019-02-04 20:07:41,050 INFO] Step 80050/102583; acc: 83.58; ppl: 1.69; xent: 0.53; lr: 0.00001; 888/1086 tok/s; 8230 sec
  1725. [2019-02-04 20:07:45,747 INFO] Step 80100/102583; acc: 81.81; ppl: 1.80; xent: 0.59; lr: 0.00001; 2160/2587 tok/s; 8235 sec
  1726. [2019-02-04 20:07:50,919 INFO] Step 80150/102583; acc: 82.66; ppl: 1.74; xent: 0.55; lr: 0.00001; 1983/2634 tok/s; 8240 sec
  1727. [2019-02-04 20:07:56,085 INFO] Step 80200/102583; acc: 82.12; ppl: 1.76; xent: 0.57; lr: 0.00001; 1940/2705 tok/s; 8245 sec
  1728. [2019-02-04 20:08:00,888 INFO] Step 80250/102583; acc: 81.51; ppl: 1.82; xent: 0.60; lr: 0.00001; 2044/2627 tok/s; 8250 sec
  1729. [2019-02-04 20:08:05,850 INFO] Step 80300/102583; acc: 83.93; ppl: 1.69; xent: 0.52; lr: 0.00001; 2083/2561 tok/s; 8255 sec
  1730. [2019-02-04 20:08:10,733 INFO] Step 80350/102583; acc: 81.50; ppl: 1.81; xent: 0.59; lr: 0.00001; 2146/2680 tok/s; 8260 sec
  1731. [2019-02-04 20:08:15,394 INFO] Step 80400/102583; acc: 81.69; ppl: 1.80; xent: 0.59; lr: 0.00001; 2017/2653 tok/s; 8265 sec
  1732. [2019-02-04 20:08:20,503 INFO] Step 80450/102583; acc: 82.06; ppl: 1.78; xent: 0.58; lr: 0.00001; 1945/2682 tok/s; 8270 sec
  1733. [2019-02-04 20:08:25,300 INFO] Step 80500/102583; acc: 83.05; ppl: 1.72; xent: 0.54; lr: 0.00001; 2218/2559 tok/s; 8275 sec
  1734. [2019-02-04 20:08:30,446 INFO] Step 80550/102583; acc: 82.18; ppl: 1.79; xent: 0.58; lr: 0.00001; 1922/2631 tok/s; 8280 sec
  1735. [2019-02-04 20:08:35,595 INFO] Step 80600/102583; acc: 81.20; ppl: 1.82; xent: 0.60; lr: 0.00001; 1979/2654 tok/s; 8285 sec
  1736. [2019-02-04 20:08:40,569 INFO] Step 80650/102583; acc: 83.11; ppl: 1.73; xent: 0.55; lr: 0.00001; 2114/2580 tok/s; 8290 sec
  1737. [2019-02-04 20:08:45,401 INFO] Step 80700/102583; acc: 83.58; ppl: 1.70; xent: 0.53; lr: 0.00001; 1895/2783 tok/s; 8295 sec
  1738. [2019-02-04 20:08:50,605 INFO] Step 80750/102583; acc: 82.81; ppl: 1.74; xent: 0.55; lr: 0.00001; 2166/2487 tok/s; 8300 sec
  1739. [2019-02-04 20:08:55,411 INFO] Step 80800/102583; acc: 81.67; ppl: 1.79; xent: 0.58; lr: 0.00001; 2008/2792 tok/s; 8305 sec
  1740. [2019-02-04 20:09:00,324 INFO] Step 80850/102583; acc: 81.44; ppl: 1.84; xent: 0.61; lr: 0.00001; 2181/2374 tok/s; 8310 sec
  1741. [2019-02-04 20:09:05,627 INFO] Step 80900/102583; acc: 82.93; ppl: 1.72; xent: 0.54; lr: 0.00001; 2059/2662 tok/s; 8315 sec
  1742. [2019-02-04 20:09:10,254 INFO] Step 80950/102583; acc: 80.81; ppl: 1.86; xent: 0.62; lr: 0.00001; 1916/2713 tok/s; 8320 sec
  1743. [2019-02-04 20:09:15,075 INFO] Step 81000/102583; acc: 82.52; ppl: 1.76; xent: 0.56; lr: 0.00001; 2129/2552 tok/s; 8324 sec
  1744. [2019-02-04 20:09:20,287 INFO] Step 81050/102583; acc: 82.43; ppl: 1.76; xent: 0.57; lr: 0.00001; 1926/2662 tok/s; 8330 sec
  1745. [2019-02-04 20:09:25,627 INFO] Step 81100/102583; acc: 82.52; ppl: 1.76; xent: 0.57; lr: 0.00001; 1961/2403 tok/s; 8335 sec
  1746. [2019-02-04 20:09:31,154 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1747. [2019-02-04 20:09:36,933 INFO] Step 81150/102583; acc: 82.34; ppl: 1.78; xent: 0.58; lr: 0.00001; 932/1163 tok/s; 8346 sec
  1748. [2019-02-04 20:09:41,674 INFO] Step 81200/102583; acc: 82.86; ppl: 1.76; xent: 0.56; lr: 0.00001; 2059/2693 tok/s; 8351 sec
  1749. [2019-02-04 20:09:46,777 INFO] Step 81250/102583; acc: 83.60; ppl: 1.70; xent: 0.53; lr: 0.00001; 2048/2683 tok/s; 8356 sec
  1750. [2019-02-04 20:09:51,553 INFO] Step 81300/102583; acc: 81.20; ppl: 1.86; xent: 0.62; lr: 0.00001; 2048/2610 tok/s; 8361 sec
  1751. [2019-02-04 20:09:56,471 INFO] Step 81350/102583; acc: 82.97; ppl: 1.73; xent: 0.55; lr: 0.00001; 2049/2729 tok/s; 8366 sec
  1752. [2019-02-04 20:10:01,461 INFO] Step 81400/102583; acc: 81.44; ppl: 1.80; xent: 0.59; lr: 0.00001; 2072/2575 tok/s; 8371 sec
  1753. [2019-02-04 20:10:06,337 INFO] Step 81450/102583; acc: 82.64; ppl: 1.76; xent: 0.56; lr: 0.00001; 2128/2682 tok/s; 8376 sec
  1754. [2019-02-04 20:10:11,270 INFO] Step 81500/102583; acc: 82.16; ppl: 1.78; xent: 0.58; lr: 0.00001; 2043/2668 tok/s; 8381 sec
  1755. [2019-02-04 20:10:16,391 INFO] Step 81550/102583; acc: 81.00; ppl: 1.84; xent: 0.61; lr: 0.00001; 1996/2575 tok/s; 8386 sec
  1756. [2019-02-04 20:10:21,157 INFO] Step 81600/102583; acc: 83.05; ppl: 1.74; xent: 0.55; lr: 0.00001; 2122/2663 tok/s; 8391 sec
  1757. [2019-02-04 20:10:26,273 INFO] Step 81650/102583; acc: 82.47; ppl: 1.75; xent: 0.56; lr: 0.00001; 1928/2686 tok/s; 8396 sec
  1758. [2019-02-04 20:10:31,129 INFO] Step 81700/102583; acc: 82.07; ppl: 1.80; xent: 0.59; lr: 0.00001; 2194/2577 tok/s; 8401 sec
  1759. [2019-02-04 20:10:36,140 INFO] Step 81750/102583; acc: 81.23; ppl: 1.82; xent: 0.60; lr: 0.00001; 2065/2578 tok/s; 8406 sec
  1760. [2019-02-04 20:10:41,005 INFO] Step 81800/102583; acc: 83.32; ppl: 1.74; xent: 0.55; lr: 0.00001; 2038/2660 tok/s; 8410 sec
  1761. [2019-02-04 20:10:46,091 INFO] Step 81850/102583; acc: 82.73; ppl: 1.73; xent: 0.55; lr: 0.00001; 2165/2654 tok/s; 8415 sec
  1762. [2019-02-04 20:10:50,928 INFO] Step 81900/102583; acc: 82.00; ppl: 1.79; xent: 0.58; lr: 0.00001; 1992/2647 tok/s; 8420 sec
  1763. [2019-02-04 20:10:56,054 INFO] Step 81950/102583; acc: 82.83; ppl: 1.73; xent: 0.55; lr: 0.00001; 1986/2503 tok/s; 8425 sec
  1764. [2019-02-04 20:11:01,092 INFO] Step 82000/102583; acc: 82.44; ppl: 1.76; xent: 0.56; lr: 0.00001; 2003/2642 tok/s; 8430 sec
  1765. [2019-02-04 20:11:06,040 INFO] Step 82050/102583; acc: 82.32; ppl: 1.79; xent: 0.58; lr: 0.00001; 2098/2449 tok/s; 8435 sec
  1766. [2019-02-04 20:11:10,997 INFO] Step 82100/102583; acc: 82.42; ppl: 1.76; xent: 0.57; lr: 0.00001; 1998/2777 tok/s; 8440 sec
  1767. [2019-02-04 20:11:16,253 INFO] Step 82150/102583; acc: 81.92; ppl: 1.77; xent: 0.57; lr: 0.00001; 1914/2613 tok/s; 8446 sec
  1768. [2019-02-04 20:11:21,005 INFO] Step 82200/102583; acc: 82.38; ppl: 1.78; xent: 0.58; lr: 0.00001; 2141/2597 tok/s; 8450 sec
  1769. [2019-02-04 20:11:26,152 INFO] Step 82250/102583; acc: 82.14; ppl: 1.77; xent: 0.57; lr: 0.00001; 2104/2532 tok/s; 8456 sec
  1770. [2019-02-04 20:11:30,938 INFO] Step 82300/102583; acc: 81.42; ppl: 1.81; xent: 0.59; lr: 0.00001; 1996/2687 tok/s; 8460 sec
  1771. [2019-02-04 20:11:35,564 INFO] Step 82350/102583; acc: 81.43; ppl: 1.84; xent: 0.61; lr: 0.00001; 2085/2681 tok/s; 8465 sec
  1772. [2019-02-04 20:11:40,863 INFO] Step 82400/102583; acc: 83.05; ppl: 1.72; xent: 0.54; lr: 0.00001; 2012/2556 tok/s; 8470 sec
  1773. [2019-02-04 20:11:46,188 INFO] Step 82450/102583; acc: 83.44; ppl: 1.71; xent: 0.53; lr: 0.00001; 2099/2513 tok/s; 8476 sec
  1774. [2019-02-04 20:11:50,910 INFO] Step 82500/102583; acc: 80.99; ppl: 1.83; xent: 0.61; lr: 0.00001; 1919/2707 tok/s; 8480 sec
  1775. [2019-02-04 20:11:56,036 INFO] Step 82550/102583; acc: 82.55; ppl: 1.75; xent: 0.56; lr: 0.00001; 2089/2670 tok/s; 8485 sec
  1776. [2019-02-04 20:12:00,790 INFO] Step 82600/102583; acc: 82.26; ppl: 1.79; xent: 0.58; lr: 0.00001; 2010/2545 tok/s; 8490 sec
  1777. [2019-02-04 20:12:05,578 INFO] Step 82650/102583; acc: 81.51; ppl: 1.84; xent: 0.61; lr: 0.00001; 1944/2668 tok/s; 8495 sec
  1778. [2019-02-04 20:12:10,719 INFO] Step 82700/102583; acc: 83.05; ppl: 1.70; xent: 0.53; lr: 0.00001; 2097/2540 tok/s; 8500 sec
  1779. [2019-02-04 20:12:15,608 INFO] Step 82750/102583; acc: 81.02; ppl: 1.82; xent: 0.60; lr: 0.00001; 2082/2805 tok/s; 8505 sec
  1780. [2019-02-04 20:12:20,578 INFO] Step 82800/102583; acc: 83.60; ppl: 1.72; xent: 0.54; lr: 0.00001; 2061/2510 tok/s; 8510 sec
  1781. [2019-02-04 20:12:25,622 INFO] Step 82850/102583; acc: 82.56; ppl: 1.78; xent: 0.58; lr: 0.00001; 1895/2739 tok/s; 8515 sec
  1782. [2019-02-04 20:12:30,459 INFO] Step 82900/102583; acc: 81.60; ppl: 1.81; xent: 0.59; lr: 0.00001; 2205/2516 tok/s; 8520 sec
  1783. [2019-02-04 20:12:35,537 INFO] Step 82950/102583; acc: 81.72; ppl: 1.79; xent: 0.58; lr: 0.00001; 1890/2681 tok/s; 8525 sec
  1784. [2019-02-04 20:12:40,418 INFO] Step 83000/102583; acc: 82.80; ppl: 1.75; xent: 0.56; lr: 0.00001; 2169/2535 tok/s; 8530 sec
  1785. [2019-02-04 20:12:45,548 INFO] Step 83050/102583; acc: 83.10; ppl: 1.75; xent: 0.56; lr: 0.00001; 2149/2637 tok/s; 8535 sec
  1786. [2019-02-04 20:12:50,475 INFO] Step 83100/102583; acc: 82.29; ppl: 1.79; xent: 0.58; lr: 0.00001; 1901/2561 tok/s; 8540 sec
  1787. [2019-02-04 20:12:55,329 INFO] Step 83150/102583; acc: 81.33; ppl: 1.84; xent: 0.61; lr: 0.00001; 1956/2652 tok/s; 8545 sec
  1788. [2019-02-04 20:13:00,365 INFO] Step 83200/102583; acc: 82.92; ppl: 1.73; xent: 0.55; lr: 0.00001; 2140/2563 tok/s; 8550 sec
  1789. [2019-02-04 20:13:05,138 INFO] Step 83250/102583; acc: 82.08; ppl: 1.80; xent: 0.59; lr: 0.00001; 2000/2699 tok/s; 8555 sec
  1790. [2019-02-04 20:13:10,160 INFO] Step 83300/102583; acc: 82.77; ppl: 1.74; xent: 0.56; lr: 0.00001; 2116/2584 tok/s; 8560 sec
  1791. [2019-02-04 20:13:14,892 INFO] Step 83350/102583; acc: 82.17; ppl: 1.77; xent: 0.57; lr: 0.00001; 2139/2605 tok/s; 8564 sec
  1792. [2019-02-04 20:13:20,140 INFO] Step 83400/102583; acc: 82.54; ppl: 1.74; xent: 0.55; lr: 0.00001; 1919/2624 tok/s; 8570 sec
  1793. [2019-02-04 20:13:25,162 INFO] Step 83450/102583; acc: 82.39; ppl: 1.77; xent: 0.57; lr: 0.00001; 2058/2668 tok/s; 8575 sec
  1794. [2019-02-04 20:13:29,986 INFO] Step 83500/102583; acc: 82.06; ppl: 1.78; xent: 0.58; lr: 0.00001; 2048/2635 tok/s; 8579 sec
  1795. [2019-02-04 20:13:35,033 INFO] Step 83550/102583; acc: 83.41; ppl: 1.71; xent: 0.53; lr: 0.00001; 2074/2596 tok/s; 8584 sec
  1796. [2019-02-04 20:13:39,872 INFO] Step 83600/102583; acc: 81.43; ppl: 1.81; xent: 0.59; lr: 0.00001; 2039/2685 tok/s; 8589 sec
  1797. [2019-02-04 20:13:44,497 INFO] Step 83650/102583; acc: 81.89; ppl: 1.81; xent: 0.59; lr: 0.00001; 2102/2597 tok/s; 8594 sec
  1798. [2019-02-04 20:13:49,844 INFO] Step 83700/102583; acc: 82.86; ppl: 1.73; xent: 0.55; lr: 0.00001; 1992/2658 tok/s; 8599 sec
  1799. [2019-02-04 20:13:54,742 INFO] Step 83750/102583; acc: 82.22; ppl: 1.76; xent: 0.56; lr: 0.00001; 2209/2523 tok/s; 8604 sec
  1800. [2019-02-04 20:13:59,761 INFO] Step 83800/102583; acc: 82.52; ppl: 1.77; xent: 0.57; lr: 0.00001; 1864/2727 tok/s; 8609 sec
  1801. [2019-02-04 20:14:04,867 INFO] Step 83850/102583; acc: 81.02; ppl: 1.85; xent: 0.61; lr: 0.00001; 1926/2561 tok/s; 8614 sec
  1802. [2019-02-04 20:14:09,720 INFO] Step 83900/102583; acc: 83.45; ppl: 1.71; xent: 0.53; lr: 0.00001; 2164/2687 tok/s; 8619 sec
  1803. [2019-02-04 20:14:14,501 INFO] Step 83950/102583; acc: 82.43; ppl: 1.75; xent: 0.56; lr: 0.00001; 1902/2735 tok/s; 8624 sec
  1804. [2019-02-04 20:14:19,735 INFO] Step 84000/102583; acc: 82.67; ppl: 1.76; xent: 0.56; lr: 0.00001; 2141/2503 tok/s; 8629 sec
  1805. [2019-02-04 20:14:24,574 INFO] Step 84050/102583; acc: 82.68; ppl: 1.75; xent: 0.56; lr: 0.00001; 2030/2705 tok/s; 8634 sec
  1806. [2019-02-04 20:14:29,667 INFO] Step 84100/102583; acc: 81.62; ppl: 1.80; xent: 0.59; lr: 0.00001; 2084/2501 tok/s; 8639 sec
  1807. [2019-02-04 20:14:34,775 INFO] Step 84150/102583; acc: 82.20; ppl: 1.76; xent: 0.57; lr: 0.00001; 2135/2609 tok/s; 8644 sec
  1808. [2019-02-04 20:14:39,553 INFO] Step 84200/102583; acc: 82.13; ppl: 1.78; xent: 0.58; lr: 0.00001; 1981/2680 tok/s; 8649 sec
  1809. [2019-02-04 20:14:44,202 INFO] Step 84250/102583; acc: 82.24; ppl: 1.79; xent: 0.58; lr: 0.00001; 2116/2572 tok/s; 8654 sec
  1810. [2019-02-04 20:14:49,572 INFO] Step 84300/102583; acc: 82.17; ppl: 1.76; xent: 0.57; lr: 0.00001; 1936/2639 tok/s; 8659 sec
  1811. [2019-02-04 20:14:59,656 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1812. [2019-02-04 20:15:01,209 INFO] Step 84350/102583; acc: 81.80; ppl: 1.82; xent: 0.60; lr: 0.00001; 847/1081 tok/s; 8671 sec
  1813. [2019-02-04 20:15:06,292 INFO] Step 84400/102583; acc: 83.11; ppl: 1.73; xent: 0.55; lr: 0.00001; 2153/2562 tok/s; 8676 sec
  1814. [2019-02-04 20:15:11,022 INFO] Step 84450/102583; acc: 82.14; ppl: 1.78; xent: 0.58; lr: 0.00001; 1969/2704 tok/s; 8680 sec
  1815. [2019-02-04 20:15:16,266 INFO] Step 84500/102583; acc: 83.77; ppl: 1.69; xent: 0.52; lr: 0.00001; 2053/2663 tok/s; 8686 sec
  1816. [2019-02-04 20:15:21,064 INFO] Step 84550/102583; acc: 81.72; ppl: 1.82; xent: 0.60; lr: 0.00001; 2095/2627 tok/s; 8690 sec
  1817. [2019-02-04 20:15:26,061 INFO] Step 84600/102583; acc: 82.57; ppl: 1.75; xent: 0.56; lr: 0.00001; 2094/2664 tok/s; 8695 sec
  1818. [2019-02-04 20:15:31,067 INFO] Step 84650/102583; acc: 81.64; ppl: 1.78; xent: 0.58; lr: 0.00001; 2023/2617 tok/s; 8700 sec
  1819. [2019-02-04 20:15:35,807 INFO] Step 84700/102583; acc: 83.18; ppl: 1.74; xent: 0.55; lr: 0.00001; 2136/2775 tok/s; 8705 sec
  1820. [2019-02-04 20:15:40,788 INFO] Step 84750/102583; acc: 81.10; ppl: 1.84; xent: 0.61; lr: 0.00001; 1920/2692 tok/s; 8710 sec
  1821. [2019-02-04 20:15:45,914 INFO] Step 84800/102583; acc: 81.87; ppl: 1.81; xent: 0.59; lr: 0.00001; 2090/2483 tok/s; 8715 sec
  1822. [2019-02-04 20:15:50,586 INFO] Step 84850/102583; acc: 82.64; ppl: 1.75; xent: 0.56; lr: 0.00001; 2087/2688 tok/s; 8720 sec
  1823. [2019-02-04 20:15:55,799 INFO] Step 84900/102583; acc: 82.28; ppl: 1.76; xent: 0.57; lr: 0.00001; 1987/2662 tok/s; 8725 sec
  1824. [2019-02-04 20:16:00,465 INFO] Step 84950/102583; acc: 82.24; ppl: 1.79; xent: 0.58; lr: 0.00001; 2190/2546 tok/s; 8730 sec
  1825. [2019-02-04 20:16:05,489 INFO] Step 85000/102583; acc: 80.99; ppl: 1.81; xent: 0.59; lr: 0.00001; 2048/2612 tok/s; 8735 sec
  1826. [2019-02-04 20:16:05,490 INFO] Saving checkpoint models/ted1_step_85000.pt
  1827. [2019-02-04 20:16:16,245 INFO] Step 85050/102583; acc: 83.45; ppl: 1.73; xent: 0.55; lr: 0.00001; 940/1216 tok/s; 8746 sec
  1828. [2019-02-04 20:16:21,463 INFO] Step 85100/102583; acc: 82.33; ppl: 1.74; xent: 0.56; lr: 0.00001; 2151/2603 tok/s; 8751 sec
  1829. [2019-02-04 20:16:26,217 INFO] Step 85150/102583; acc: 81.55; ppl: 1.80; xent: 0.59; lr: 0.00001; 2007/2646 tok/s; 8756 sec
  1830. [2019-02-04 20:16:31,263 INFO] Step 85200/102583; acc: 82.40; ppl: 1.74; xent: 0.55; lr: 0.00001; 2006/2578 tok/s; 8761 sec
  1831. [2019-02-04 20:16:36,372 INFO] Step 85250/102583; acc: 83.27; ppl: 1.72; xent: 0.54; lr: 0.00001; 1987/2644 tok/s; 8766 sec
  1832. [2019-02-04 20:16:41,309 INFO] Step 85300/102583; acc: 81.10; ppl: 1.84; xent: 0.61; lr: 0.00001; 2064/2475 tok/s; 8771 sec
  1833. [2019-02-04 20:16:46,349 INFO] Step 85350/102583; acc: 81.93; ppl: 1.77; xent: 0.57; lr: 0.00001; 2000/2732 tok/s; 8776 sec
  1834. [2019-02-04 20:16:51,613 INFO] Step 85400/102583; acc: 82.88; ppl: 1.74; xent: 0.55; lr: 0.00001; 1928/2544 tok/s; 8781 sec
  1835. [2019-02-04 20:16:56,618 INFO] Step 85450/102583; acc: 83.66; ppl: 1.71; xent: 0.54; lr: 0.00001; 2183/2626 tok/s; 8786 sec
  1836. [2019-02-04 20:17:01,395 INFO] Step 85500/102583; acc: 80.68; ppl: 1.85; xent: 0.61; lr: 0.00001; 2092/2595 tok/s; 8791 sec
  1837. [2019-02-04 20:17:06,221 INFO] Step 85550/102583; acc: 80.48; ppl: 1.86; xent: 0.62; lr: 0.00001; 1946/2712 tok/s; 8796 sec
  1838. [2019-02-04 20:17:10,958 INFO] Step 85600/102583; acc: 82.88; ppl: 1.75; xent: 0.56; lr: 0.00001; 2111/2634 tok/s; 8800 sec
  1839. [2019-02-04 20:17:16,293 INFO] Step 85650/102583; acc: 83.32; ppl: 1.71; xent: 0.54; lr: 0.00001; 2063/2479 tok/s; 8806 sec
  1840. [2019-02-04 20:17:21,394 INFO] Step 85700/102583; acc: 82.42; ppl: 1.75; xent: 0.56; lr: 0.00001; 2068/2598 tok/s; 8811 sec
  1841. [2019-02-04 20:17:26,312 INFO] Step 85750/102583; acc: 81.21; ppl: 1.80; xent: 0.59; lr: 0.00001; 1827/2729 tok/s; 8816 sec
  1842. [2019-02-04 20:17:31,277 INFO] Step 85800/102583; acc: 82.86; ppl: 1.75; xent: 0.56; lr: 0.00001; 2172/2675 tok/s; 8821 sec
  1843. [2019-02-04 20:17:36,126 INFO] Step 85850/102583; acc: 82.17; ppl: 1.79; xent: 0.58; lr: 0.00001; 1976/2512 tok/s; 8826 sec
  1844. [2019-02-04 20:17:40,815 INFO] Step 85900/102583; acc: 81.72; ppl: 1.83; xent: 0.61; lr: 0.00001; 1974/2624 tok/s; 8830 sec
  1845. [2019-02-04 20:17:46,074 INFO] Step 85950/102583; acc: 82.54; ppl: 1.71; xent: 0.54; lr: 0.00001; 2111/2564 tok/s; 8835 sec
  1846. [2019-02-04 20:17:50,965 INFO] Step 86000/102583; acc: 80.29; ppl: 1.87; xent: 0.62; lr: 0.00001; 1990/2775 tok/s; 8840 sec
  1847. [2019-02-04 20:17:55,983 INFO] Step 86050/102583; acc: 83.66; ppl: 1.71; xent: 0.54; lr: 0.00001; 2098/2588 tok/s; 8845 sec
  1848. [2019-02-04 20:18:01,075 INFO] Step 86100/102583; acc: 83.27; ppl: 1.74; xent: 0.55; lr: 0.00001; 2009/2668 tok/s; 8850 sec
  1849. [2019-02-04 20:18:05,700 INFO] Step 86150/102583; acc: 80.84; ppl: 1.85; xent: 0.61; lr: 0.00001; 2036/2599 tok/s; 8855 sec
  1850. [2019-02-04 20:18:10,783 INFO] Step 86200/102583; acc: 82.24; ppl: 1.77; xent: 0.57; lr: 0.00001; 1989/2631 tok/s; 8860 sec
  1851. [2019-02-04 20:18:15,682 INFO] Step 86250/102583; acc: 82.85; ppl: 1.75; xent: 0.56; lr: 0.00001; 2135/2560 tok/s; 8865 sec
  1852. [2019-02-04 20:18:20,971 INFO] Step 86300/102583; acc: 83.36; ppl: 1.72; xent: 0.54; lr: 0.00001; 2154/2538 tok/s; 8870 sec
  1853. [2019-02-04 20:18:25,681 INFO] Step 86350/102583; acc: 81.84; ppl: 1.80; xent: 0.59; lr: 0.00001; 1856/2635 tok/s; 8875 sec
  1854. [2019-02-04 20:18:30,609 INFO] Step 86400/102583; acc: 81.03; ppl: 1.86; xent: 0.62; lr: 0.00001; 1955/2664 tok/s; 8880 sec
  1855. [2019-02-04 20:18:35,705 INFO] Step 86450/102583; acc: 82.56; ppl: 1.75; xent: 0.56; lr: 0.00001; 2088/2548 tok/s; 8885 sec
  1856. [2019-02-04 20:18:40,494 INFO] Step 86500/102583; acc: 83.12; ppl: 1.73; xent: 0.55; lr: 0.00001; 2092/2724 tok/s; 8890 sec
  1857. [2019-02-04 20:18:45,537 INFO] Step 86550/102583; acc: 82.22; ppl: 1.78; xent: 0.58; lr: 0.00001; 2107/2469 tok/s; 8895 sec
  1858. [2019-02-04 20:18:50,255 INFO] Step 86600/102583; acc: 82.08; ppl: 1.76; xent: 0.57; lr: 0.00001; 2118/2729 tok/s; 8900 sec
  1859. [2019-02-04 20:18:55,440 INFO] Step 86650/102583; acc: 82.63; ppl: 1.74; xent: 0.55; lr: 0.00001; 1949/2638 tok/s; 8905 sec
  1860. [2019-02-04 20:19:00,402 INFO] Step 86700/102583; acc: 82.10; ppl: 1.77; xent: 0.57; lr: 0.00001; 2045/2657 tok/s; 8910 sec
  1861. [2019-02-04 20:19:05,321 INFO] Step 86750/102583; acc: 82.57; ppl: 1.76; xent: 0.56; lr: 0.00001; 2089/2516 tok/s; 8915 sec
  1862. [2019-02-04 20:19:10,263 INFO] Step 86800/102583; acc: 82.98; ppl: 1.73; xent: 0.55; lr: 0.00001; 2026/2717 tok/s; 8920 sec
  1863. [2019-02-04 20:19:15,142 INFO] Step 86850/102583; acc: 80.92; ppl: 1.83; xent: 0.60; lr: 0.00001; 2048/2696 tok/s; 8925 sec
  1864. [2019-02-04 20:19:19,968 INFO] Step 86900/102583; acc: 82.45; ppl: 1.77; xent: 0.57; lr: 0.00001; 2044/2538 tok/s; 8929 sec
  1865. [2019-02-04 20:19:25,038 INFO] Step 86950/102583; acc: 82.84; ppl: 1.74; xent: 0.56; lr: 0.00001; 2047/2678 tok/s; 8934 sec
  1866. [2019-02-04 20:19:30,055 INFO] Step 87000/102583; acc: 82.62; ppl: 1.73; xent: 0.55; lr: 0.00001; 2184/2551 tok/s; 8939 sec
  1867. [2019-02-04 20:19:35,209 INFO] Step 87050/102583; acc: 81.59; ppl: 1.80; xent: 0.59; lr: 0.00001; 1767/2786 tok/s; 8945 sec
  1868. [2019-02-04 20:19:40,056 INFO] Step 87100/102583; acc: 81.15; ppl: 1.86; xent: 0.62; lr: 0.00001; 2080/2477 tok/s; 8949 sec
  1869. [2019-02-04 20:19:44,942 INFO] Step 87150/102583; acc: 83.73; ppl: 1.69; xent: 0.53; lr: 0.00001; 2110/2708 tok/s; 8954 sec
  1870. [2019-02-04 20:19:49,753 INFO] Step 87200/102583; acc: 82.15; ppl: 1.76; xent: 0.57; lr: 0.00001; 1930/2764 tok/s; 8959 sec
  1871. [2019-02-04 20:19:55,061 INFO] Step 87250/102583; acc: 81.97; ppl: 1.79; xent: 0.58; lr: 0.00001; 2129/2463 tok/s; 8964 sec
  1872. [2019-02-04 20:19:59,867 INFO] Step 87300/102583; acc: 83.35; ppl: 1.72; xent: 0.54; lr: 0.00001; 2095/2693 tok/s; 8969 sec
  1873. [2019-02-04 20:20:04,997 INFO] Step 87350/102583; acc: 82.28; ppl: 1.77; xent: 0.57; lr: 0.00001; 2124/2489 tok/s; 8974 sec
  1874. [2019-02-04 20:20:10,110 INFO] Step 87400/102583; acc: 81.91; ppl: 1.79; xent: 0.58; lr: 0.00001; 1996/2644 tok/s; 8979 sec
  1875. [2019-02-04 20:20:14,769 INFO] Step 87450/102583; acc: 82.57; ppl: 1.76; xent: 0.57; lr: 0.00001; 2023/2706 tok/s; 8984 sec
  1876. [2019-02-04 20:20:19,409 INFO] Step 87500/102583; acc: 81.82; ppl: 1.79; xent: 0.58; lr: 0.00001; 2134/2580 tok/s; 8989 sec
  1877. [2019-02-04 20:20:24,724 INFO] Step 87550/102583; acc: 81.95; ppl: 1.78; xent: 0.58; lr: 0.00001; 1957/2607 tok/s; 8994 sec
  1878. [2019-02-04 20:20:33,457 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1879. [2019-02-04 20:20:35,233 INFO] Step 87600/102583; acc: 81.56; ppl: 1.84; xent: 0.61; lr: 0.00001; 937/1223 tok/s; 9005 sec
  1880. [2019-02-04 20:20:40,341 INFO] Step 87650/102583; acc: 83.39; ppl: 1.72; xent: 0.54; lr: 0.00001; 2155/2539 tok/s; 9010 sec
  1881. [2019-02-04 20:20:45,062 INFO] Step 87700/102583; acc: 82.46; ppl: 1.77; xent: 0.57; lr: 0.00001; 1921/2725 tok/s; 9014 sec
  1882. [2019-02-04 20:20:50,196 INFO] Step 87750/102583; acc: 83.82; ppl: 1.70; xent: 0.53; lr: 0.00001; 2080/2655 tok/s; 9020 sec
  1883. [2019-02-04 20:20:55,101 INFO] Step 87800/102583; acc: 82.03; ppl: 1.79; xent: 0.58; lr: 0.00001; 2154/2667 tok/s; 9024 sec
  1884. [2019-02-04 20:21:00,134 INFO] Step 87850/102583; acc: 82.87; ppl: 1.74; xent: 0.55; lr: 0.00001; 2092/2673 tok/s; 9030 sec
  1885. [2019-02-04 20:21:04,990 INFO] Step 87900/102583; acc: 81.60; ppl: 1.81; xent: 0.59; lr: 0.00001; 2007/2543 tok/s; 9034 sec
  1886. [2019-02-04 20:21:09,888 INFO] Step 87950/102583; acc: 82.81; ppl: 1.75; xent: 0.56; lr: 0.00001; 2088/2794 tok/s; 9039 sec
  1887. [2019-02-04 20:21:14,734 INFO] Step 88000/102583; acc: 81.30; ppl: 1.82; xent: 0.60; lr: 0.00001; 1992/2698 tok/s; 9044 sec
  1888. [2019-02-04 20:21:19,932 INFO] Step 88050/102583; acc: 81.28; ppl: 1.83; xent: 0.61; lr: 0.00001; 2000/2523 tok/s; 9049 sec
  1889. [2019-02-04 20:21:24,616 INFO] Step 88100/102583; acc: 83.80; ppl: 1.68; xent: 0.52; lr: 0.00001; 2219/2653 tok/s; 9054 sec
  1890. [2019-02-04 20:21:29,933 INFO] Step 88150/102583; acc: 82.24; ppl: 1.78; xent: 0.58; lr: 0.00001; 1968/2607 tok/s; 9059 sec
  1891. [2019-02-04 20:21:34,412 INFO] Step 88200/102583; acc: 81.55; ppl: 1.86; xent: 0.62; lr: 0.00001; 2114/2610 tok/s; 9064 sec
  1892. [2019-02-04 20:21:39,617 INFO] Step 88250/102583; acc: 81.13; ppl: 1.79; xent: 0.58; lr: 0.00001; 1988/2677 tok/s; 9069 sec
  1893. [2019-02-04 20:21:44,414 INFO] Step 88300/102583; acc: 83.80; ppl: 1.73; xent: 0.55; lr: 0.00001; 2134/2595 tok/s; 9074 sec
  1894. [2019-02-04 20:21:49,611 INFO] Step 88350/102583; acc: 82.89; ppl: 1.72; xent: 0.54; lr: 0.00001; 2124/2620 tok/s; 9079 sec
  1895. [2019-02-04 20:21:54,311 INFO] Step 88400/102583; acc: 82.27; ppl: 1.79; xent: 0.58; lr: 0.00001; 2058/2582 tok/s; 9084 sec
  1896. [2019-02-04 20:21:59,406 INFO] Step 88450/102583; acc: 82.30; ppl: 1.74; xent: 0.55; lr: 0.00001; 1974/2602 tok/s; 9089 sec
  1897. [2019-02-04 20:22:04,535 INFO] Step 88500/102583; acc: 82.92; ppl: 1.74; xent: 0.55; lr: 0.00001; 2017/2646 tok/s; 9094 sec
  1898. [2019-02-04 20:22:09,537 INFO] Step 88550/102583; acc: 82.33; ppl: 1.79; xent: 0.58; lr: 0.00001; 2077/2431 tok/s; 9099 sec
  1899. [2019-02-04 20:22:14,520 INFO] Step 88600/102583; acc: 80.52; ppl: 1.83; xent: 0.60; lr: 0.00001; 1945/2788 tok/s; 9104 sec
  1900. [2019-02-04 20:22:19,745 INFO] Step 88650/102583; acc: 83.58; ppl: 1.71; xent: 0.54; lr: 0.00001; 1967/2513 tok/s; 9109 sec
  1901. [2019-02-04 20:22:24,771 INFO] Step 88700/102583; acc: 82.76; ppl: 1.74; xent: 0.55; lr: 0.00001; 2174/2645 tok/s; 9114 sec
  1902. [2019-02-04 20:22:29,537 INFO] Step 88750/102583; acc: 81.18; ppl: 1.83; xent: 0.60; lr: 0.00001; 2110/2577 tok/s; 9119 sec
  1903. [2019-02-04 20:22:34,473 INFO] Step 88800/102583; acc: 80.97; ppl: 1.84; xent: 0.61; lr: 0.00001; 1889/2692 tok/s; 9124 sec
  1904. [2019-02-04 20:22:39,379 INFO] Step 88850/102583; acc: 82.59; ppl: 1.75; xent: 0.56; lr: 0.00001; 2164/2700 tok/s; 9129 sec
  1905. [2019-02-04 20:22:44,425 INFO] Step 88900/102583; acc: 83.44; ppl: 1.72; xent: 0.54; lr: 0.00001; 2097/2405 tok/s; 9134 sec
  1906. [2019-02-04 20:22:49,658 INFO] Step 88950/102583; acc: 82.33; ppl: 1.76; xent: 0.56; lr: 0.00001; 1967/2661 tok/s; 9139 sec
  1907. [2019-02-04 20:22:54,433 INFO] Step 89000/102583; acc: 81.52; ppl: 1.80; xent: 0.59; lr: 0.00001; 1841/2675 tok/s; 9144 sec
  1908. [2019-02-04 20:22:59,538 INFO] Step 89050/102583; acc: 82.17; ppl: 1.77; xent: 0.57; lr: 0.00001; 2136/2657 tok/s; 9149 sec
  1909. [2019-02-04 20:23:04,246 INFO] Step 89100/102583; acc: 82.41; ppl: 1.80; xent: 0.59; lr: 0.00001; 2036/2513 tok/s; 9154 sec
  1910. [2019-02-04 20:23:09,006 INFO] Step 89150/102583; acc: 81.68; ppl: 1.80; xent: 0.59; lr: 0.00001; 1985/2662 tok/s; 9158 sec
  1911. [2019-02-04 20:23:14,334 INFO] Step 89200/102583; acc: 82.79; ppl: 1.71; xent: 0.54; lr: 0.00001; 2083/2607 tok/s; 9164 sec
  1912. [2019-02-04 20:23:19,250 INFO] Step 89250/102583; acc: 80.84; ppl: 1.85; xent: 0.62; lr: 0.00001; 2062/2657 tok/s; 9169 sec
  1913. [2019-02-04 20:23:24,179 INFO] Step 89300/102583; acc: 83.02; ppl: 1.73; xent: 0.55; lr: 0.00001; 2016/2670 tok/s; 9174 sec
  1914. [2019-02-04 20:23:29,192 INFO] Step 89350/102583; acc: 83.21; ppl: 1.75; xent: 0.56; lr: 0.00001; 2066/2617 tok/s; 9179 sec
  1915. [2019-02-04 20:23:33,957 INFO] Step 89400/102583; acc: 80.84; ppl: 1.87; xent: 0.63; lr: 0.00001; 1947/2551 tok/s; 9183 sec
  1916. [2019-02-04 20:23:39,085 INFO] Step 89450/102583; acc: 82.96; ppl: 1.71; xent: 0.54; lr: 0.00001; 2052/2685 tok/s; 9188 sec
  1917. [2019-02-04 20:23:44,009 INFO] Step 89500/102583; acc: 82.34; ppl: 1.78; xent: 0.57; lr: 0.00001; 2101/2568 tok/s; 9193 sec
  1918. [2019-02-04 20:23:49,314 INFO] Step 89550/102583; acc: 83.22; ppl: 1.73; xent: 0.55; lr: 0.00001; 2136/2468 tok/s; 9199 sec
  1919. [2019-02-04 20:23:53,876 INFO] Step 89600/102583; acc: 82.31; ppl: 1.78; xent: 0.58; lr: 0.00001; 1934/2738 tok/s; 9203 sec
  1920. [2019-02-04 20:23:58,824 INFO] Step 89650/102583; acc: 80.02; ppl: 1.91; xent: 0.65; lr: 0.00001; 1947/2638 tok/s; 9208 sec
  1921. [2019-02-04 20:24:03,906 INFO] Step 89700/102583; acc: 83.41; ppl: 1.73; xent: 0.55; lr: 0.00001; 2006/2586 tok/s; 9213 sec
  1922. [2019-02-04 20:24:08,638 INFO] Step 89750/102583; acc: 83.04; ppl: 1.73; xent: 0.55; lr: 0.00001; 2117/2693 tok/s; 9218 sec
  1923. [2019-02-04 20:24:13,661 INFO] Step 89800/102583; acc: 81.93; ppl: 1.81; xent: 0.59; lr: 0.00001; 2128/2473 tok/s; 9223 sec
  1924. [2019-02-04 20:24:18,767 INFO] Step 89850/102583; acc: 83.61; ppl: 1.68; xent: 0.52; lr: 0.00001; 2144/2597 tok/s; 9228 sec
  1925. [2019-02-04 20:24:23,552 INFO] Step 89900/102583; acc: 80.93; ppl: 1.83; xent: 0.60; lr: 0.00001; 1858/2788 tok/s; 9233 sec
  1926. [2019-02-04 20:24:28,642 INFO] Step 89950/102583; acc: 82.14; ppl: 1.77; xent: 0.57; lr: 0.00001; 2058/2664 tok/s; 9238 sec
  1927. [2019-02-04 20:24:33,603 INFO] Step 90000/102583; acc: 84.02; ppl: 1.69; xent: 0.52; lr: 0.00000; 2147/2532 tok/s; 9243 sec
  1928. [2019-02-04 20:24:33,663 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  1929. [2019-02-04 20:24:34,966 INFO] Validation perplexity: 1.74294
  1930. [2019-02-04 20:24:34,967 INFO] Validation accuracy: 83.2452
  1931. [2019-02-04 20:24:34,968 INFO] Saving checkpoint models/ted1_step_90000.pt
  1932. [2019-02-04 20:24:45,524 INFO] Step 90050/102583; acc: 82.18; ppl: 1.77; xent: 0.57; lr: 0.00000; 828/1081 tok/s; 9255 sec
  1933. [2019-02-04 20:24:50,533 INFO] Step 90100/102583; acc: 81.90; ppl: 1.78; xent: 0.57; lr: 0.00000; 2080/2668 tok/s; 9260 sec
  1934. [2019-02-04 20:24:55,201 INFO] Step 90150/102583; acc: 81.86; ppl: 1.80; xent: 0.59; lr: 0.00000; 2023/2635 tok/s; 9265 sec
  1935. [2019-02-04 20:25:00,233 INFO] Step 90200/102583; acc: 82.79; ppl: 1.74; xent: 0.56; lr: 0.00000; 2101/2628 tok/s; 9270 sec
  1936. [2019-02-04 20:25:05,349 INFO] Step 90250/102583; acc: 82.75; ppl: 1.73; xent: 0.55; lr: 0.00000; 2091/2552 tok/s; 9275 sec
  1937. [2019-02-04 20:25:10,397 INFO] Step 90300/102583; acc: 81.39; ppl: 1.82; xent: 0.60; lr: 0.00000; 1754/2847 tok/s; 9280 sec
  1938. [2019-02-04 20:25:15,261 INFO] Step 90350/102583; acc: 80.79; ppl: 1.87; xent: 0.62; lr: 0.00000; 2021/2458 tok/s; 9285 sec
  1939. [2019-02-04 20:25:20,189 INFO] Step 90400/102583; acc: 83.96; ppl: 1.68; xent: 0.52; lr: 0.00000; 2182/2736 tok/s; 9290 sec
  1940. [2019-02-04 20:25:24,958 INFO] Step 90450/102583; acc: 82.40; ppl: 1.75; xent: 0.56; lr: 0.00000; 2014/2676 tok/s; 9294 sec
  1941. [2019-02-04 20:25:30,321 INFO] Step 90500/102583; acc: 82.64; ppl: 1.76; xent: 0.57; lr: 0.00000; 2047/2527 tok/s; 9300 sec
  1942. [2019-02-04 20:25:35,105 INFO] Step 90550/102583; acc: 81.71; ppl: 1.81; xent: 0.59; lr: 0.00000; 2064/2688 tok/s; 9304 sec
  1943. [2019-02-04 20:25:40,243 INFO] Step 90600/102583; acc: 83.50; ppl: 1.70; xent: 0.53; lr: 0.00000; 2158/2473 tok/s; 9310 sec
  1944. [2019-02-04 20:25:45,293 INFO] Step 90650/102583; acc: 81.53; ppl: 1.81; xent: 0.60; lr: 0.00000; 2046/2606 tok/s; 9315 sec
  1945. [2019-02-04 20:25:49,940 INFO] Step 90700/102583; acc: 81.87; ppl: 1.79; xent: 0.58; lr: 0.00000; 1945/2704 tok/s; 9319 sec
  1946. [2019-02-04 20:25:54,768 INFO] Step 90750/102583; acc: 82.12; ppl: 1.78; xent: 0.57; lr: 0.00000; 2103/2568 tok/s; 9324 sec
  1947. [2019-02-04 20:26:00,037 INFO] Step 90800/102583; acc: 82.65; ppl: 1.74; xent: 0.56; lr: 0.00000; 2007/2617 tok/s; 9329 sec
  1948. [2019-02-04 20:26:08,943 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  1949. [2019-02-04 20:26:11,211 INFO] Step 90850/102583; acc: 80.66; ppl: 1.88; xent: 0.63; lr: 0.00000; 866/1138 tok/s; 9341 sec
  1950. [2019-02-04 20:26:16,286 INFO] Step 90900/102583; acc: 83.46; ppl: 1.73; xent: 0.55; lr: 0.00000; 2169/2512 tok/s; 9346 sec
  1951. [2019-02-04 20:26:21,052 INFO] Step 90950/102583; acc: 82.41; ppl: 1.77; xent: 0.57; lr: 0.00000; 1889/2763 tok/s; 9350 sec
  1952. [2019-02-04 20:26:26,252 INFO] Step 91000/102583; acc: 84.30; ppl: 1.68; xent: 0.52; lr: 0.00000; 2078/2637 tok/s; 9356 sec
  1953. [2019-02-04 20:26:31,114 INFO] Step 91050/102583; acc: 81.67; ppl: 1.80; xent: 0.59; lr: 0.00000; 2107/2705 tok/s; 9361 sec
  1954. [2019-02-04 20:26:36,307 INFO] Step 91100/102583; acc: 82.97; ppl: 1.71; xent: 0.54; lr: 0.00000; 2057/2690 tok/s; 9366 sec
  1955. [2019-02-04 20:26:41,187 INFO] Step 91150/102583; acc: 81.43; ppl: 1.82; xent: 0.60; lr: 0.00000; 2050/2536 tok/s; 9371 sec
  1956. [2019-02-04 20:26:45,891 INFO] Step 91200/102583; acc: 83.31; ppl: 1.75; xent: 0.56; lr: 0.00000; 2195/2721 tok/s; 9375 sec
  1957. [2019-02-04 20:26:50,993 INFO] Step 91250/102583; acc: 80.76; ppl: 1.84; xent: 0.61; lr: 0.00000; 1974/2589 tok/s; 9380 sec
  1958. [2019-02-04 20:26:55,905 INFO] Step 91300/102583; acc: 81.42; ppl: 1.84; xent: 0.61; lr: 0.00000; 1940/2638 tok/s; 9385 sec
  1959. [2019-02-04 20:27:00,823 INFO] Step 91350/102583; acc: 83.75; ppl: 1.66; xent: 0.51; lr: 0.00000; 2221/2690 tok/s; 9390 sec
  1960. [2019-02-04 20:27:05,898 INFO] Step 91400/102583; acc: 81.23; ppl: 1.85; xent: 0.61; lr: 0.00000; 1932/2562 tok/s; 9395 sec
  1961. [2019-02-04 20:27:10,621 INFO] Step 91450/102583; acc: 81.78; ppl: 1.82; xent: 0.60; lr: 0.00000; 2123/2569 tok/s; 9400 sec
  1962. [2019-02-04 20:27:15,624 INFO] Step 91500/102583; acc: 81.21; ppl: 1.79; xent: 0.58; lr: 0.00000; 2033/2748 tok/s; 9405 sec
  1963. [2019-02-04 20:27:20,601 INFO] Step 91550/102583; acc: 84.50; ppl: 1.68; xent: 0.52; lr: 0.00000; 2091/2627 tok/s; 9410 sec
  1964. [2019-02-04 20:27:25,669 INFO] Step 91600/102583; acc: 82.00; ppl: 1.78; xent: 0.58; lr: 0.00000; 2131/2612 tok/s; 9415 sec
  1965. [2019-02-04 20:27:30,368 INFO] Step 91650/102583; acc: 82.91; ppl: 1.76; xent: 0.56; lr: 0.00000; 2163/2469 tok/s; 9420 sec
  1966. [2019-02-04 20:27:35,468 INFO] Step 91700/102583; acc: 81.84; ppl: 1.75; xent: 0.56; lr: 0.00000; 1864/2728 tok/s; 9425 sec
  1967. [2019-02-04 20:27:40,502 INFO] Step 91750/102583; acc: 82.73; ppl: 1.76; xent: 0.56; lr: 0.00000; 2055/2597 tok/s; 9430 sec
  1968. [2019-02-04 20:27:45,743 INFO] Step 91800/102583; acc: 82.01; ppl: 1.79; xent: 0.58; lr: 0.00000; 2019/2483 tok/s; 9435 sec
  1969. [2019-02-04 20:27:50,777 INFO] Step 91850/102583; acc: 80.80; ppl: 1.83; xent: 0.60; lr: 0.00000; 1899/2690 tok/s; 9440 sec
  1970. [2019-02-04 20:27:55,865 INFO] Step 91900/102583; acc: 83.80; ppl: 1.70; xent: 0.53; lr: 0.00000; 1970/2629 tok/s; 9445 sec
  1971. [2019-02-04 20:28:00,884 INFO] Step 91950/102583; acc: 82.36; ppl: 1.77; xent: 0.57; lr: 0.00000; 2226/2515 tok/s; 9450 sec
  1972. [2019-02-04 20:28:05,494 INFO] Step 92000/102583; acc: 81.15; ppl: 1.84; xent: 0.61; lr: 0.00000; 2100/2630 tok/s; 9455 sec
  1973. [2019-02-04 20:28:10,498 INFO] Step 92050/102583; acc: 81.91; ppl: 1.79; xent: 0.58; lr: 0.00000; 2004/2701 tok/s; 9460 sec
  1974. [2019-02-04 20:28:15,590 INFO] Step 92100/102583; acc: 82.58; ppl: 1.74; xent: 0.55; lr: 0.00000; 2092/2647 tok/s; 9465 sec
  1975. [2019-02-04 20:28:20,571 INFO] Step 92150/102583; acc: 83.54; ppl: 1.71; xent: 0.54; lr: 0.00000; 2165/2377 tok/s; 9470 sec
  1976. [2019-02-04 20:28:25,621 INFO] Step 92200/102583; acc: 80.97; ppl: 1.81; xent: 0.59; lr: 0.00000; 1901/2733 tok/s; 9475 sec
  1977. [2019-02-04 20:28:30,491 INFO] Step 92250/102583; acc: 81.84; ppl: 1.80; xent: 0.59; lr: 0.00000; 1832/2738 tok/s; 9480 sec
  1978. [2019-02-04 20:28:35,478 INFO] Step 92300/102583; acc: 83.41; ppl: 1.71; xent: 0.53; lr: 0.00000; 2174/2581 tok/s; 9485 sec
  1979. [2019-02-04 20:28:40,287 INFO] Step 92350/102583; acc: 81.75; ppl: 1.81; xent: 0.60; lr: 0.00000; 1980/2592 tok/s; 9490 sec
  1980. [2019-02-04 20:28:45,355 INFO] Step 92400/102583; acc: 82.17; ppl: 1.77; xent: 0.57; lr: 0.00000; 2051/2545 tok/s; 9495 sec
  1981. [2019-02-04 20:28:50,446 INFO] Step 92450/102583; acc: 81.76; ppl: 1.78; xent: 0.58; lr: 0.00000; 2035/2673 tok/s; 9500 sec
  1982. [2019-02-04 20:28:55,248 INFO] Step 92500/102583; acc: 82.50; ppl: 1.77; xent: 0.57; lr: 0.00000; 2135/2666 tok/s; 9505 sec
  1983. [2019-02-04 20:29:00,292 INFO] Step 92550/102583; acc: 82.70; ppl: 1.75; xent: 0.56; lr: 0.00000; 1976/2689 tok/s; 9510 sec
  1984. [2019-02-04 20:29:05,190 INFO] Step 92600/102583; acc: 82.47; ppl: 1.79; xent: 0.58; lr: 0.00000; 2030/2630 tok/s; 9515 sec
  1985. [2019-02-04 20:29:10,143 INFO] Step 92650/102583; acc: 81.73; ppl: 1.81; xent: 0.59; lr: 0.00000; 1972/2528 tok/s; 9520 sec
  1986. [2019-02-04 20:29:15,089 INFO] Step 92700/102583; acc: 82.20; ppl: 1.75; xent: 0.56; lr: 0.00000; 2068/2701 tok/s; 9524 sec
  1987. [2019-02-04 20:29:20,182 INFO] Step 92750/102583; acc: 82.50; ppl: 1.76; xent: 0.57; lr: 0.00000; 2080/2545 tok/s; 9530 sec
  1988. [2019-02-04 20:29:25,336 INFO] Step 92800/102583; acc: 83.42; ppl: 1.73; xent: 0.55; lr: 0.00000; 2163/2486 tok/s; 9535 sec
  1989. [2019-02-04 20:29:29,907 INFO] Step 92850/102583; acc: 81.21; ppl: 1.84; xent: 0.61; lr: 0.00000; 1832/2770 tok/s; 9539 sec
  1990. [2019-02-04 20:29:34,838 INFO] Step 92900/102583; acc: 80.98; ppl: 1.83; xent: 0.60; lr: 0.00000; 2084/2605 tok/s; 9544 sec
  1991. [2019-02-04 20:29:39,887 INFO] Step 92950/102583; acc: 83.22; ppl: 1.76; xent: 0.56; lr: 0.00000; 2019/2552 tok/s; 9549 sec
  1992. [2019-02-04 20:29:44,684 INFO] Step 93000/102583; acc: 83.33; ppl: 1.71; xent: 0.54; lr: 0.00000; 2026/2685 tok/s; 9554 sec
  1993. [2019-02-04 20:29:49,739 INFO] Step 93050/102583; acc: 82.23; ppl: 1.78; xent: 0.58; lr: 0.00000; 2199/2491 tok/s; 9559 sec
  1994. [2019-02-04 20:29:54,719 INFO] Step 93100/102583; acc: 83.04; ppl: 1.71; xent: 0.53; lr: 0.00000; 2147/2585 tok/s; 9564 sec
  1995. [2019-02-04 20:29:59,810 INFO] Step 93150/102583; acc: 81.99; ppl: 1.77; xent: 0.57; lr: 0.00000; 1802/2806 tok/s; 9569 sec
  1996. [2019-02-04 20:30:04,864 INFO] Step 93200/102583; acc: 82.26; ppl: 1.77; xent: 0.57; lr: 0.00000; 2032/2613 tok/s; 9574 sec
  1997. [2019-02-04 20:30:09,562 INFO] Step 93250/102583; acc: 82.89; ppl: 1.75; xent: 0.56; lr: 0.00000; 2183/2565 tok/s; 9579 sec
  1998. [2019-02-04 20:30:14,456 INFO] Step 93300/102583; acc: 82.55; ppl: 1.75; xent: 0.56; lr: 0.00000; 2017/2642 tok/s; 9584 sec
  1999. [2019-02-04 20:30:19,548 INFO] Step 93350/102583; acc: 81.91; ppl: 1.78; xent: 0.58; lr: 0.00000; 2089/2641 tok/s; 9589 sec
  2000. [2019-02-04 20:30:24,240 INFO] Step 93400/102583; acc: 81.05; ppl: 1.84; xent: 0.61; lr: 0.00000; 1995/2664 tok/s; 9594 sec
  2001. [2019-02-04 20:30:29,299 INFO] Step 93450/102583; acc: 83.83; ppl: 1.69; xent: 0.52; lr: 0.00000; 2091/2637 tok/s; 9599 sec
  2002. [2019-02-04 20:30:34,358 INFO] Step 93500/102583; acc: 82.93; ppl: 1.72; xent: 0.54; lr: 0.00000; 2114/2565 tok/s; 9604 sec
  2003. [2019-02-04 20:30:39,407 INFO] Step 93550/102583; acc: 81.38; ppl: 1.83; xent: 0.61; lr: 0.00000; 1779/2848 tok/s; 9609 sec
  2004. [2019-02-04 20:30:44,276 INFO] Step 93600/102583; acc: 81.50; ppl: 1.84; xent: 0.61; lr: 0.00000; 2005/2462 tok/s; 9614 sec
  2005. [2019-02-04 20:30:49,135 INFO] Step 93650/102583; acc: 83.51; ppl: 1.69; xent: 0.53; lr: 0.00000; 2187/2715 tok/s; 9619 sec
  2006. [2019-02-04 20:30:54,144 INFO] Step 93700/102583; acc: 82.48; ppl: 1.76; xent: 0.56; lr: 0.00000; 1983/2520 tok/s; 9624 sec
  2007. [2019-02-04 20:30:59,319 INFO] Step 93750/102583; acc: 83.03; ppl: 1.72; xent: 0.54; lr: 0.00000; 2107/2706 tok/s; 9629 sec
  2008. [2019-02-04 20:31:04,359 INFO] Step 93800/102583; acc: 81.65; ppl: 1.81; xent: 0.60; lr: 0.00000; 2119/2558 tok/s; 9634 sec
  2009. [2019-02-04 20:31:09,336 INFO] Step 93850/102583; acc: 82.89; ppl: 1.73; xent: 0.55; lr: 0.00000; 2032/2587 tok/s; 9639 sec
  2010. [2019-02-04 20:31:14,177 INFO] Step 93900/102583; acc: 81.79; ppl: 1.82; xent: 0.60; lr: 0.00000; 2116/2553 tok/s; 9644 sec
  2011. [2019-02-04 20:31:19,005 INFO] Step 93950/102583; acc: 82.30; ppl: 1.77; xent: 0.57; lr: 0.00000; 2002/2648 tok/s; 9648 sec
  2012. [2019-02-04 20:31:23,863 INFO] Step 94000/102583; acc: 82.14; ppl: 1.79; xent: 0.58; lr: 0.00000; 1962/2724 tok/s; 9653 sec
  2013. [2019-02-04 20:31:29,133 INFO] Step 94050/102583; acc: 82.41; ppl: 1.74; xent: 0.55; lr: 0.00000; 2006/2576 tok/s; 9659 sec
  2014. [2019-02-04 20:31:37,584 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  2015. [2019-02-04 20:31:40,615 INFO] Step 94100/102583; acc: 82.37; ppl: 1.78; xent: 0.58; lr: 0.00000; 886/1126 tok/s; 9670 sec
  2016. [2019-02-04 20:31:45,530 INFO] Step 94150/102583; acc: 82.67; ppl: 1.79; xent: 0.58; lr: 0.00000; 2176/2431 tok/s; 9675 sec
  2017. [2019-02-04 20:31:50,327 INFO] Step 94200/102583; acc: 82.58; ppl: 1.75; xent: 0.56; lr: 0.00000; 1940/2803 tok/s; 9680 sec
  2018. [2019-02-04 20:31:55,525 INFO] Step 94250/102583; acc: 83.22; ppl: 1.71; xent: 0.54; lr: 0.00000; 1991/2638 tok/s; 9685 sec
  2019. [2019-02-04 20:32:00,352 INFO] Step 94300/102583; acc: 81.65; ppl: 1.82; xent: 0.60; lr: 0.00000; 2157/2745 tok/s; 9690 sec
  2020. [2019-02-04 20:32:05,450 INFO] Step 94350/102583; acc: 82.92; ppl: 1.71; xent: 0.54; lr: 0.00000; 2020/2658 tok/s; 9695 sec
  2021. [2019-02-04 20:32:10,304 INFO] Step 94400/102583; acc: 81.96; ppl: 1.78; xent: 0.58; lr: 0.00000; 2180/2507 tok/s; 9700 sec
  2022. [2019-02-04 20:32:15,005 INFO] Step 94450/102583; acc: 82.70; ppl: 1.77; xent: 0.57; lr: 0.00000; 2101/2736 tok/s; 9704 sec
  2023. [2019-02-04 20:32:20,228 INFO] Step 94500/102583; acc: 80.77; ppl: 1.84; xent: 0.61; lr: 0.00000; 1947/2591 tok/s; 9710 sec
  2024. [2019-02-04 20:32:25,176 INFO] Step 94550/102583; acc: 82.28; ppl: 1.79; xent: 0.58; lr: 0.00000; 2002/2628 tok/s; 9715 sec
  2025. [2019-02-04 20:32:30,011 INFO] Step 94600/102583; acc: 83.37; ppl: 1.67; xent: 0.51; lr: 0.00000; 2161/2745 tok/s; 9719 sec
  2026. [2019-02-04 20:32:35,220 INFO] Step 94650/102583; acc: 80.79; ppl: 1.87; xent: 0.62; lr: 0.00000; 1955/2497 tok/s; 9725 sec
  2027. [2019-02-04 20:32:39,908 INFO] Step 94700/102583; acc: 82.39; ppl: 1.80; xent: 0.59; lr: 0.00000; 2188/2572 tok/s; 9729 sec
  2028. [2019-02-04 20:32:44,882 INFO] Step 94750/102583; acc: 81.14; ppl: 1.79; xent: 0.58; lr: 0.00000; 1943/2753 tok/s; 9734 sec
  2029. [2019-02-04 20:32:49,958 INFO] Step 94800/102583; acc: 84.27; ppl: 1.69; xent: 0.53; lr: 0.00000; 2175/2559 tok/s; 9739 sec
  2030. [2019-02-04 20:32:54,934 INFO] Step 94850/102583; acc: 81.62; ppl: 1.79; xent: 0.58; lr: 0.00000; 2046/2673 tok/s; 9744 sec
  2031. [2019-02-04 20:32:59,625 INFO] Step 94900/102583; acc: 82.89; ppl: 1.76; xent: 0.56; lr: 0.00000; 2214/2488 tok/s; 9749 sec
  2032. [2019-02-04 20:33:04,730 INFO] Step 94950/102583; acc: 82.54; ppl: 1.72; xent: 0.54; lr: 0.00000; 1858/2767 tok/s; 9754 sec
  2033. [2019-02-04 20:33:09,898 INFO] Step 95000/102583; acc: 81.96; ppl: 1.79; xent: 0.58; lr: 0.00000; 2034/2467 tok/s; 9759 sec
  2034. [2019-02-04 20:33:09,900 INFO] Saving checkpoint models/ted1_step_95000.pt
  2035. [2019-02-04 20:33:20,514 INFO] Step 95050/102583; acc: 83.00; ppl: 1.74; xent: 0.55; lr: 0.00000; 999/1227 tok/s; 9770 sec
  2036. [2019-02-04 20:33:25,528 INFO] Step 95100/102583; acc: 79.95; ppl: 1.88; xent: 0.63; lr: 0.00000; 1830/2718 tok/s; 9775 sec
  2037. [2019-02-04 20:33:30,658 INFO] Step 95150/102583; acc: 84.58; ppl: 1.66; xent: 0.50; lr: 0.00000; 2016/2607 tok/s; 9780 sec
  2038. [2019-02-04 20:33:35,570 INFO] Step 95200/102583; acc: 81.26; ppl: 1.83; xent: 0.60; lr: 0.00000; 2131/2504 tok/s; 9785 sec
  2039. [2019-02-04 20:33:40,163 INFO] Step 95250/102583; acc: 81.11; ppl: 1.84; xent: 0.61; lr: 0.00000; 2177/2619 tok/s; 9790 sec
  2040. [2019-02-04 20:33:45,231 INFO] Step 95300/102583; acc: 81.96; ppl: 1.77; xent: 0.57; lr: 0.00000; 1992/2742 tok/s; 9795 sec
  2041. [2019-02-04 20:33:50,362 INFO] Step 95350/102583; acc: 83.31; ppl: 1.73; xent: 0.55; lr: 0.00000; 2064/2663 tok/s; 9800 sec
  2042. [2019-02-04 20:33:55,536 INFO] Step 95400/102583; acc: 83.95; ppl: 1.69; xent: 0.53; lr: 0.00000; 2161/2325 tok/s; 9805 sec
  2043. [2019-02-04 20:34:00,440 INFO] Step 95450/102583; acc: 80.60; ppl: 1.84; xent: 0.61; lr: 0.00000; 1851/2826 tok/s; 9810 sec
  2044. [2019-02-04 20:34:05,277 INFO] Step 95500/102583; acc: 81.69; ppl: 1.81; xent: 0.59; lr: 0.00000; 1858/2718 tok/s; 9815 sec
  2045. [2019-02-04 20:34:10,313 INFO] Step 95550/102583; acc: 84.63; ppl: 1.66; xent: 0.51; lr: 0.00000; 2251/2502 tok/s; 9820 sec
  2046. [2019-02-04 20:34:15,053 INFO] Step 95600/102583; acc: 81.19; ppl: 1.85; xent: 0.61; lr: 0.00000; 1945/2640 tok/s; 9824 sec
  2047. [2019-02-04 20:34:20,162 INFO] Step 95650/102583; acc: 82.19; ppl: 1.77; xent: 0.57; lr: 0.00000; 2077/2507 tok/s; 9830 sec
  2048. [2019-02-04 20:34:25,296 INFO] Step 95700/102583; acc: 82.50; ppl: 1.74; xent: 0.55; lr: 0.00000; 2050/2752 tok/s; 9835 sec
  2049. [2019-02-04 20:34:30,039 INFO] Step 95750/102583; acc: 81.35; ppl: 1.83; xent: 0.61; lr: 0.00000; 2005/2591 tok/s; 9839 sec
  2050. [2019-02-04 20:34:35,052 INFO] Step 95800/102583; acc: 82.69; ppl: 1.75; xent: 0.56; lr: 0.00000; 1996/2745 tok/s; 9844 sec
  2051. [2019-02-04 20:34:40,106 INFO] Step 95850/102583; acc: 83.42; ppl: 1.72; xent: 0.54; lr: 0.00000; 2107/2585 tok/s; 9849 sec
  2052. [2019-02-04 20:34:45,083 INFO] Step 95900/102583; acc: 80.57; ppl: 1.86; xent: 0.62; lr: 0.00000; 1896/2512 tok/s; 9854 sec
  2053. [2019-02-04 20:34:49,897 INFO] Step 95950/102583; acc: 82.66; ppl: 1.74; xent: 0.55; lr: 0.00000; 2172/2751 tok/s; 9859 sec
  2054. [2019-02-04 20:34:54,949 INFO] Step 96000/102583; acc: 81.84; ppl: 1.80; xent: 0.59; lr: 0.00000; 1991/2529 tok/s; 9864 sec
  2055. [2019-02-04 20:35:00,078 INFO] Step 96050/102583; acc: 83.87; ppl: 1.70; xent: 0.53; lr: 0.00000; 2211/2490 tok/s; 9869 sec
  2056. [2019-02-04 20:35:04,795 INFO] Step 96100/102583; acc: 80.83; ppl: 1.87; xent: 0.62; lr: 0.00000; 1890/2737 tok/s; 9874 sec
  2057. [2019-02-04 20:35:09,871 INFO] Step 96150/102583; acc: 81.67; ppl: 1.80; xent: 0.59; lr: 0.00000; 2017/2627 tok/s; 9879 sec
  2058. [2019-02-04 20:35:14,730 INFO] Step 96200/102583; acc: 82.83; ppl: 1.77; xent: 0.57; lr: 0.00000; 2034/2578 tok/s; 9884 sec
  2059. [2019-02-04 20:35:19,607 INFO] Step 96250/102583; acc: 83.72; ppl: 1.69; xent: 0.52; lr: 0.00000; 2062/2628 tok/s; 9889 sec
  2060. [2019-02-04 20:35:24,597 INFO] Step 96300/102583; acc: 81.78; ppl: 1.81; xent: 0.60; lr: 0.00000; 2155/2503 tok/s; 9894 sec
  2061. [2019-02-04 20:35:29,538 INFO] Step 96350/102583; acc: 83.52; ppl: 1.69; xent: 0.52; lr: 0.00000; 2178/2597 tok/s; 9899 sec
  2062. [2019-02-04 20:35:34,608 INFO] Step 96400/102583; acc: 81.39; ppl: 1.80; xent: 0.59; lr: 0.00000; 1797/2836 tok/s; 9904 sec
  2063. [2019-02-04 20:35:39,728 INFO] Step 96450/102583; acc: 82.96; ppl: 1.75; xent: 0.56; lr: 0.00000; 2068/2564 tok/s; 9909 sec
  2064. [2019-02-04 20:35:44,416 INFO] Step 96500/102583; acc: 83.29; ppl: 1.72; xent: 0.54; lr: 0.00000; 2122/2606 tok/s; 9914 sec
  2065. [2019-02-04 20:35:49,520 INFO] Step 96550/102583; acc: 82.64; ppl: 1.74; xent: 0.56; lr: 0.00000; 2043/2644 tok/s; 9919 sec
  2066. [2019-02-04 20:35:54,339 INFO] Step 96600/102583; acc: 80.64; ppl: 1.85; xent: 0.62; lr: 0.00000; 2075/2640 tok/s; 9924 sec
  2067. [2019-02-04 20:35:59,224 INFO] Step 96650/102583; acc: 81.88; ppl: 1.80; xent: 0.59; lr: 0.00000; 1914/2675 tok/s; 9929 sec
  2068. [2019-02-04 20:36:04,048 INFO] Step 96700/102583; acc: 84.15; ppl: 1.68; xent: 0.52; lr: 0.00000; 2195/2621 tok/s; 9933 sec
  2069. [2019-02-04 20:36:09,337 INFO] Step 96750/102583; acc: 82.19; ppl: 1.73; xent: 0.55; lr: 0.00000; 2040/2571 tok/s; 9939 sec
  2070. [2019-02-04 20:36:14,541 INFO] Step 96800/102583; acc: 81.39; ppl: 1.85; xent: 0.61; lr: 0.00000; 1833/2754 tok/s; 9944 sec
  2071. [2019-02-04 20:36:19,053 INFO] Step 96850/102583; acc: 81.31; ppl: 1.85; xent: 0.61; lr: 0.00000; 2075/2533 tok/s; 9948 sec
  2072. [2019-02-04 20:36:24,141 INFO] Step 96900/102583; acc: 83.48; ppl: 1.69; xent: 0.52; lr: 0.00000; 2078/2758 tok/s; 9954 sec
  2073. [2019-02-04 20:36:29,147 INFO] Step 96950/102583; acc: 82.95; ppl: 1.74; xent: 0.55; lr: 0.00000; 1993/2496 tok/s; 9959 sec
  2074. [2019-02-04 20:36:34,132 INFO] Step 97000/102583; acc: 82.10; ppl: 1.77; xent: 0.57; lr: 0.00000; 2151/2708 tok/s; 9964 sec
  2075. [2019-02-04 20:36:39,167 INFO] Step 97050/102583; acc: 82.32; ppl: 1.77; xent: 0.57; lr: 0.00000; 2134/2561 tok/s; 9969 sec
  2076. [2019-02-04 20:36:44,464 INFO] Step 97100/102583; acc: 83.25; ppl: 1.72; xent: 0.54; lr: 0.00000; 2057/2515 tok/s; 9974 sec
  2077. [2019-02-04 20:36:48,976 INFO] Step 97150/102583; acc: 81.32; ppl: 1.86; xent: 0.62; lr: 0.00000; 2040/2647 tok/s; 9978 sec
  2078. [2019-02-04 20:36:53,771 INFO] Step 97200/102583; acc: 82.58; ppl: 1.75; xent: 0.56; lr: 0.00000; 2095/2607 tok/s; 9983 sec
  2079. [2019-02-04 20:36:58,795 INFO] Step 97250/102583; acc: 82.66; ppl: 1.76; xent: 0.57; lr: 0.00000; 1969/2670 tok/s; 9988 sec
  2080. [2019-02-04 20:37:04,084 INFO] Step 97300/102583; acc: 81.54; ppl: 1.77; xent: 0.57; lr: 0.00000; 1910/2573 tok/s; 9993 sec
  2081. [2019-02-04 20:37:11,403 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  2082. [2019-02-04 20:37:15,251 INFO] Step 97350/102583; acc: 83.03; ppl: 1.75; xent: 0.56; lr: 0.00000; 967/1195 tok/s; 10005 sec
  2083. [2019-02-04 20:37:19,849 INFO] Step 97400/102583; acc: 81.40; ppl: 1.87; xent: 0.62; lr: 0.00000; 2145/2468 tok/s; 10009 sec
  2084. [2019-02-04 20:37:24,671 INFO] Step 97450/102583; acc: 82.91; ppl: 1.73; xent: 0.55; lr: 0.00000; 1969/2804 tok/s; 10014 sec
  2085. [2019-02-04 20:37:29,971 INFO] Step 97500/102583; acc: 83.15; ppl: 1.73; xent: 0.55; lr: 0.00000; 2018/2614 tok/s; 10019 sec
  2086. [2019-02-04 20:37:34,755 INFO] Step 97550/102583; acc: 82.34; ppl: 1.77; xent: 0.57; lr: 0.00000; 2104/2823 tok/s; 10024 sec
  2087. [2019-02-04 20:37:39,736 INFO] Step 97600/102583; acc: 81.93; ppl: 1.77; xent: 0.57; lr: 0.00000; 2004/2601 tok/s; 10029 sec
  2088. [2019-02-04 20:37:44,777 INFO] Step 97650/102583; acc: 82.92; ppl: 1.73; xent: 0.55; lr: 0.00000; 2222/2525 tok/s; 10034 sec
  2089. [2019-02-04 20:37:49,367 INFO] Step 97700/102583; acc: 81.67; ppl: 1.83; xent: 0.61; lr: 0.00000; 2073/2767 tok/s; 10039 sec
  2090. [2019-02-04 20:37:54,608 INFO] Step 97750/102583; acc: 81.51; ppl: 1.80; xent: 0.59; lr: 0.00000; 1904/2605 tok/s; 10044 sec
  2091. [2019-02-04 20:37:59,672 INFO] Step 97800/102583; acc: 82.70; ppl: 1.77; xent: 0.57; lr: 0.00000; 2156/2561 tok/s; 10049 sec
  2092. [2019-02-04 20:38:04,514 INFO] Step 97850/102583; acc: 83.45; ppl: 1.68; xent: 0.52; lr: 0.00000; 2031/2805 tok/s; 10054 sec
  2093. [2019-02-04 20:38:09,540 INFO] Step 97900/102583; acc: 80.37; ppl: 1.89; xent: 0.64; lr: 0.00000; 2048/2440 tok/s; 10059 sec
  2094. [2019-02-04 20:38:14,475 INFO] Step 97950/102583; acc: 83.31; ppl: 1.74; xent: 0.55; lr: 0.00000; 2070/2633 tok/s; 10064 sec
  2095. [2019-02-04 20:38:19,312 INFO] Step 98000/102583; acc: 81.04; ppl: 1.81; xent: 0.60; lr: 0.00000; 1994/2759 tok/s; 10069 sec
  2096. [2019-02-04 20:38:24,272 INFO] Step 98050/102583; acc: 84.52; ppl: 1.69; xent: 0.53; lr: 0.00000; 2212/2514 tok/s; 10074 sec
  2097. [2019-02-04 20:38:29,500 INFO] Step 98100/102583; acc: 81.58; ppl: 1.78; xent: 0.58; lr: 0.00000; 1963/2699 tok/s; 10079 sec
  2098. [2019-02-04 20:38:34,008 INFO] Step 98150/102583; acc: 82.38; ppl: 1.79; xent: 0.58; lr: 0.00000; 2214/2484 tok/s; 10083 sec
  2099. [2019-02-04 20:38:39,013 INFO] Step 98200/102583; acc: 82.69; ppl: 1.71; xent: 0.54; lr: 0.00000; 1883/2730 tok/s; 10088 sec
  2100. [2019-02-04 20:38:44,339 INFO] Step 98250/102583; acc: 82.44; ppl: 1.77; xent: 0.57; lr: 0.00000; 1994/2478 tok/s; 10094 sec
  2101. [2019-02-04 20:38:49,347 INFO] Step 98300/102583; acc: 82.70; ppl: 1.75; xent: 0.56; lr: 0.00000; 2161/2588 tok/s; 10099 sec
  2102. [2019-02-04 20:38:54,676 INFO] Step 98350/102583; acc: 81.17; ppl: 1.80; xent: 0.59; lr: 0.00000; 1858/2696 tok/s; 10104 sec
  2103. [2019-02-04 20:38:59,464 INFO] Step 98400/102583; acc: 83.59; ppl: 1.73; xent: 0.55; lr: 0.00000; 2048/2632 tok/s; 10109 sec
  2104. [2019-02-04 20:39:04,468 INFO] Step 98450/102583; acc: 80.40; ppl: 1.86; xent: 0.62; lr: 0.00000; 2041/2576 tok/s; 10114 sec
  2105. [2019-02-04 20:39:09,106 INFO] Step 98500/102583; acc: 82.54; ppl: 1.77; xent: 0.57; lr: 0.00000; 2247/2536 tok/s; 10118 sec
  2106. [2019-02-04 20:39:14,055 INFO] Step 98550/102583; acc: 80.76; ppl: 1.83; xent: 0.60; lr: 0.00000; 1967/2764 tok/s; 10123 sec
  2107. [2019-02-04 20:39:19,165 INFO] Step 98600/102583; acc: 83.68; ppl: 1.70; xent: 0.53; lr: 0.00000; 2060/2660 tok/s; 10129 sec
  2108. [2019-02-04 20:39:24,468 INFO] Step 98650/102583; acc: 83.39; ppl: 1.72; xent: 0.54; lr: 0.00000; 2076/2307 tok/s; 10134 sec
  2109. [2019-02-04 20:39:29,282 INFO] Step 98700/102583; acc: 80.95; ppl: 1.80; xent: 0.59; lr: 0.00000; 1948/2861 tok/s; 10139 sec
  2110. [2019-02-04 20:39:34,011 INFO] Step 98750/102583; acc: 81.54; ppl: 1.83; xent: 0.60; lr: 0.00000; 1873/2684 tok/s; 10143 sec
  2111. [2019-02-04 20:39:39,072 INFO] Step 98800/102583; acc: 84.35; ppl: 1.67; xent: 0.51; lr: 0.00000; 2241/2519 tok/s; 10148 sec
  2112. [2019-02-04 20:39:43,880 INFO] Step 98850/102583; acc: 80.09; ppl: 1.90; xent: 0.64; lr: 0.00000; 1837/2682 tok/s; 10153 sec
  2113. [2019-02-04 20:39:49,086 INFO] Step 98900/102583; acc: 83.74; ppl: 1.68; xent: 0.52; lr: 0.00000; 2161/2535 tok/s; 10158 sec
  2114. [2019-02-04 20:39:54,082 INFO] Step 98950/102583; acc: 81.68; ppl: 1.79; xent: 0.58; lr: 0.00000; 2031/2712 tok/s; 10163 sec
  2115. [2019-02-04 20:39:58,945 INFO] Step 99000/102583; acc: 82.10; ppl: 1.78; xent: 0.58; lr: 0.00000; 2049/2595 tok/s; 10168 sec
  2116. [2019-02-04 20:40:03,789 INFO] Step 99050/102583; acc: 82.16; ppl: 1.78; xent: 0.58; lr: 0.00000; 1972/2714 tok/s; 10173 sec
  2117. [2019-02-04 20:40:08,846 INFO] Step 99100/102583; acc: 83.58; ppl: 1.70; xent: 0.53; lr: 0.00000; 2207/2573 tok/s; 10178 sec
  2118. [2019-02-04 20:40:13,820 INFO] Step 99150/102583; acc: 79.50; ppl: 1.93; xent: 0.66; lr: 0.00000; 1820/2531 tok/s; 10183 sec
  2119. [2019-02-04 20:40:18,741 INFO] Step 99200/102583; acc: 83.52; ppl: 1.69; xent: 0.53; lr: 0.00000; 2239/2649 tok/s; 10188 sec
  2120. [2019-02-04 20:40:23,647 INFO] Step 99250/102583; acc: 81.17; ppl: 1.85; xent: 0.61; lr: 0.00000; 1896/2613 tok/s; 10193 sec
  2121. [2019-02-04 20:40:28,778 INFO] Step 99300/102583; acc: 84.35; ppl: 1.68; xent: 0.52; lr: 0.00000; 2173/2493 tok/s; 10198 sec
  2122. [2019-02-04 20:40:33,625 INFO] Step 99350/102583; acc: 81.02; ppl: 1.85; xent: 0.62; lr: 0.00000; 1930/2697 tok/s; 10203 sec
  2123. [2019-02-04 20:40:38,768 INFO] Step 99400/102583; acc: 82.81; ppl: 1.73; xent: 0.55; lr: 0.00000; 2001/2658 tok/s; 10208 sec
  2124. [2019-02-04 20:40:43,490 INFO] Step 99450/102583; acc: 81.23; ppl: 1.87; xent: 0.62; lr: 0.00000; 2056/2606 tok/s; 10213 sec
  2125. [2019-02-04 20:40:48,637 INFO] Step 99500/102583; acc: 84.22; ppl: 1.66; xent: 0.51; lr: 0.00000; 2002/2548 tok/s; 10218 sec
  2126. [2019-02-04 20:40:53,330 INFO] Step 99550/102583; acc: 81.18; ppl: 1.85; xent: 0.61; lr: 0.00000; 2212/2581 tok/s; 10223 sec
  2127. [2019-02-04 20:40:58,400 INFO] Step 99600/102583; acc: 82.97; ppl: 1.70; xent: 0.53; lr: 0.00000; 2099/2577 tok/s; 10228 sec
  2128. [2019-02-04 20:41:03,548 INFO] Step 99650/102583; acc: 81.99; ppl: 1.77; xent: 0.57; lr: 0.00000; 1878/2790 tok/s; 10233 sec
  2129. [2019-02-04 20:41:08,485 INFO] Step 99700/102583; acc: 82.33; ppl: 1.77; xent: 0.57; lr: 0.00000; 2087/2592 tok/s; 10238 sec
  2130. [2019-02-04 20:41:13,229 INFO] Step 99750/102583; acc: 83.91; ppl: 1.70; xent: 0.53; lr: 0.00000; 2159/2569 tok/s; 10243 sec
  2131. [2019-02-04 20:41:18,274 INFO] Step 99800/102583; acc: 81.70; ppl: 1.78; xent: 0.58; lr: 0.00000; 1984/2685 tok/s; 10248 sec
  2132. [2019-02-04 20:41:23,031 INFO] Step 99850/102583; acc: 80.96; ppl: 1.83; xent: 0.61; lr: 0.00000; 2088/2634 tok/s; 10252 sec
  2133. [2019-02-04 20:41:28,079 INFO] Step 99900/102583; acc: 82.86; ppl: 1.74; xent: 0.55; lr: 0.00000; 1968/2651 tok/s; 10257 sec
  2134. [2019-02-04 20:41:32,761 INFO] Step 99950/102583; acc: 82.80; ppl: 1.75; xent: 0.56; lr: 0.00000; 2136/2647 tok/s; 10262 sec
  2135. [2019-02-04 20:41:38,064 INFO] Step 100000/102583; acc: 82.47; ppl: 1.73; xent: 0.55; lr: 0.00000; 2034/2582 tok/s; 10267 sec
  2136. [2019-02-04 20:41:38,083 INFO] Loading dataset from data/ted.valid.0.pt, number of examples: 873
  2137. [2019-02-04 20:41:39,380 INFO] Validation perplexity: 1.7422
  2138. [2019-02-04 20:41:39,381 INFO] Validation accuracy: 83.3844
  2139. [2019-02-04 20:41:39,382 INFO] Saving checkpoint models/ted1_step_100000.pt
  2140. [2019-02-04 20:41:50,081 INFO] Step 100050/102583; acc: 81.32; ppl: 1.85; xent: 0.62; lr: 0.00000; 805/1159 tok/s; 10279 sec
  2141. [2019-02-04 20:41:54,883 INFO] Step 100100/102583; acc: 82.26; ppl: 1.79; xent: 0.58; lr: 0.00000; 2108/2524 tok/s; 10284 sec
  2142. [2019-02-04 20:41:59,662 INFO] Step 100150/102583; acc: 83.14; ppl: 1.71; xent: 0.53; lr: 0.00000; 1987/2789 tok/s; 10289 sec
  2143. [2019-02-04 20:42:04,956 INFO] Step 100200/102583; acc: 83.43; ppl: 1.72; xent: 0.54; lr: 0.00000; 2046/2525 tok/s; 10294 sec
  2144. [2019-02-04 20:42:09,741 INFO] Step 100250/102583; acc: 81.47; ppl: 1.80; xent: 0.59; lr: 0.00000; 2076/2736 tok/s; 10299 sec
  2145. [2019-02-04 20:42:14,731 INFO] Step 100300/102583; acc: 82.13; ppl: 1.79; xent: 0.58; lr: 0.00000; 2166/2528 tok/s; 10304 sec
  2146. [2019-02-04 20:42:20,179 INFO] Step 100350/102583; acc: 83.97; ppl: 1.68; xent: 0.52; lr: 0.00000; 2061/2518 tok/s; 10310 sec
  2147. [2019-02-04 20:42:24,590 INFO] Step 100400/102583; acc: 80.68; ppl: 1.87; xent: 0.63; lr: 0.00000; 2070/2660 tok/s; 10314 sec
  2148. [2019-02-04 20:42:29,363 INFO] Step 100450/102583; acc: 81.87; ppl: 1.77; xent: 0.57; lr: 0.00000; 2011/2626 tok/s; 10319 sec
  2149. [2019-02-04 20:42:34,710 INFO] Step 100500/102583; acc: 82.78; ppl: 1.74; xent: 0.56; lr: 0.00000; 1997/2570 tok/s; 10324 sec
  2150. [2019-02-04 20:42:39,844 INFO] Step 100550/102583; acc: 81.99; ppl: 1.77; xent: 0.57; lr: 0.00000; 1928/2619 tok/s; 10329 sec
  2151. [2019-02-04 20:42:46,710 INFO] Loading dataset from data/ted.train.0.pt, number of examples: 207592
  2152. [2019-02-04 20:42:51,143 INFO] Step 100600/102583; acc: 83.11; ppl: 1.75; xent: 0.56; lr: 0.00000; 955/1154 tok/s; 10341 sec
  2153. [2019-02-04 20:42:55,713 INFO] Step 100650/102583; acc: 81.20; ppl: 1.87; xent: 0.63; lr: 0.00000; 2088/2550 tok/s; 10345 sec
  2154. [2019-02-04 20:43:00,717 INFO] Step 100700/102583; acc: 83.73; ppl: 1.68; xent: 0.52; lr: 0.00000; 2033/2799 tok/s; 10350 sec
  2155. [2019-02-04 20:43:05,890 INFO] Step 100750/102583; acc: 82.19; ppl: 1.78; xent: 0.58; lr: 0.00000; 1963/2620 tok/s; 10355 sec
  2156. [2019-02-04 20:43:10,601 INFO] Step 100800/102583; acc: 82.68; ppl: 1.77; xent: 0.57; lr: 0.00000; 2095/2819 tok/s; 10360 sec
  2157. [2019-02-04 20:43:15,733 INFO] Step 100850/102583; acc: 82.82; ppl: 1.74; xent: 0.55; lr: 0.00000; 2113/2497 tok/s; 10365 sec
  2158. [2019-02-04 20:43:20,639 INFO] Step 100900/102583; acc: 83.20; ppl: 1.72; xent: 0.54; lr: 0.00000; 2185/2635 tok/s; 10370 sec
  2159. [2019-02-04 20:43:25,428 INFO] Step 100950/102583; acc: 81.34; ppl: 1.83; xent: 0.60; lr: 0.00000; 2111/2638 tok/s; 10375 sec
  2160. [2019-02-04 20:43:30,573 INFO] Step 101000/102583; acc: 80.54; ppl: 1.85; xent: 0.62; lr: 0.00000; 1777/2658 tok/s; 10380 sec
  2161. [2019-02-04 20:43:35,508 INFO] Step 101050/102583; acc: 83.26; ppl: 1.74; xent: 0.55; lr: 0.00000; 2198/2640 tok/s; 10385 sec
  2162. [2019-02-04 20:43:40,633 INFO] Step 101100/102583; acc: 82.77; ppl: 1.71; xent: 0.54; lr: 0.00000; 1892/2756 tok/s; 10390 sec
  2163. [2019-02-04 20:43:45,404 INFO] Step 101150/102583; acc: 81.47; ppl: 1.85; xent: 0.61; lr: 0.00000; 2280/2441 tok/s; 10395 sec
  2164. [2019-02-04 20:43:50,270 INFO] Step 101200/102583; acc: 82.61; ppl: 1.77; xent: 0.57; lr: 0.00000; 2046/2635 tok/s; 10400 sec
  2165. [2019-02-04 20:43:55,200 INFO] Step 101250/102583; acc: 80.88; ppl: 1.83; xent: 0.61; lr: 0.00000; 1913/2689 tok/s; 10405 sec
  2166. [2019-02-04 20:44:00,311 INFO] Step 101300/102583; acc: 84.89; ppl: 1.65; xent: 0.50; lr: 0.00000; 2275/2564 tok/s; 10410 sec
  2167. [2019-02-04 20:44:05,340 INFO] Step 101350/102583; acc: 80.91; ppl: 1.81; xent: 0.59; lr: 0.00000; 1928/2730 tok/s; 10415 sec
  2168. [2019-02-04 20:44:09,894 INFO] Step 101400/102583; acc: 82.34; ppl: 1.79; xent: 0.58; lr: 0.00000; 2136/2528 tok/s; 10419 sec
  2169. [2019-02-04 20:44:15,037 INFO] Step 101450/102583; acc: 83.38; ppl: 1.69; xent: 0.53; lr: 0.00000; 1942/2682 tok/s; 10424 sec
  2170. [2019-02-04 20:44:20,139 INFO] Step 101500/102583; acc: 82.27; ppl: 1.78; xent: 0.58; lr: 0.00000; 2007/2468 tok/s; 10430 sec
  2171. [2019-02-04 20:44:25,188 INFO] Step 101550/102583; acc: 82.52; ppl: 1.76; xent: 0.56; lr: 0.00000; 2105/2614 tok/s; 10435 sec
  2172. [2019-02-04 20:44:30,523 INFO] Step 101600/102583; acc: 81.42; ppl: 1.78; xent: 0.58; lr: 0.00000; 1892/2703 tok/s; 10440 sec
  2173. [2019-02-04 20:44:35,232 INFO] Step 101650/102583; acc: 83.59; ppl: 1.72; xent: 0.54; lr: 0.00000; 2082/2582 tok/s; 10445 sec
  2174. [2019-02-04 20:44:40,243 INFO] Step 101700/102583; acc: 80.70; ppl: 1.84; xent: 0.61; lr: 0.00000; 2140/2536 tok/s; 10450 sec
  2175. [2019-02-04 20:44:45,090 INFO] Step 101750/102583; acc: 82.06; ppl: 1.78; xent: 0.58; lr: 0.00000; 2031/2719 tok/s; 10454 sec
  2176. [2019-02-04 20:44:49,846 INFO] Step 101800/102583; acc: 81.09; ppl: 1.84; xent: 0.61; lr: 0.00000; 2061/2616 tok/s; 10459 sec
  2177. [2019-02-04 20:44:55,163 INFO] Step 101850/102583; acc: 83.10; ppl: 1.73; xent: 0.55; lr: 0.00000; 1927/2585 tok/s; 10465 sec
  2178. [2019-02-04 20:45:00,297 INFO] Step 101900/102583; acc: 83.94; ppl: 1.69; xent: 0.52; lr: 0.00000; 2211/2383 tok/s; 10470 sec
  2179. [2019-02-04 20:45:05,074 INFO] Step 101950/102583; acc: 80.87; ppl: 1.82; xent: 0.60; lr: 0.00000; 1896/2875 tok/s; 10474 sec
  2180. [2019-02-04 20:45:09,966 INFO] Step 102000/102583; acc: 81.75; ppl: 1.78; xent: 0.58; lr: 0.00000; 1981/2705 tok/s; 10479 sec
  2181. [2019-02-04 20:45:15,042 INFO] Step 102050/102583; acc: 84.30; ppl: 1.69; xent: 0.53; lr: 0.00000; 2181/2444 tok/s; 10484 sec
  2182. [2019-02-04 20:45:19,752 INFO] Step 102100/102583; acc: 80.81; ppl: 1.87; xent: 0.63; lr: 0.00000; 1832/2685 tok/s; 10489 sec
  2183. [2019-02-04 20:45:24,960 INFO] Step 102150/102583; acc: 83.24; ppl: 1.69; xent: 0.52; lr: 0.00000; 2153/2604 tok/s; 10494 sec
  2184. [2019-02-04 20:45:29,900 INFO] Step 102200/102583; acc: 81.51; ppl: 1.81; xent: 0.59; lr: 0.00000; 2053/2670 tok/s; 10499 sec
  2185. [2019-02-04 20:45:34,773 INFO] Step 102250/102583; acc: 82.02; ppl: 1.79; xent: 0.58; lr: 0.00000; 2018/2592 tok/s; 10504 sec
  2186. [2019-02-04 20:45:39,833 INFO] Step 102300/102583; acc: 83.06; ppl: 1.74; xent: 0.55; lr: 0.00000; 1878/2721 tok/s; 10509 sec
  2187. [2019-02-04 20:45:44,738 INFO] Step 102350/102583; acc: 83.05; ppl: 1.75; xent: 0.56; lr: 0.00000; 2309/2506 tok/s; 10514 sec
  2188. [2019-02-04 20:45:49,821 INFO] Step 102400/102583; acc: 80.38; ppl: 1.86; xent: 0.62; lr: 0.00000; 1809/2674 tok/s; 10519 sec
  2189. [2019-02-04 20:45:54,650 INFO] Step 102450/102583; acc: 83.30; ppl: 1.73; xent: 0.55; lr: 0.00000; 2239/2526 tok/s; 10524 sec
  2190. [2019-02-04 20:45:59,655 INFO] Step 102500/102583; acc: 81.88; ppl: 1.79; xent: 0.58; lr: 0.00000; 1978/2644 tok/s; 10529 sec
  2191. [2019-02-04 20:46:04,736 INFO] Step 102550/102583; acc: 83.49; ppl: 1.73; xent: 0.55; lr: 0.00000; 2052/2456 tok/s; 10534 sec
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement