API tools faq
paste
Guest User

Untitled

a guest
Jun 20th, 2026
55
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 20.41 KB | None | 0 0
raw download clone embed print report
  1. C:\Users\2ch\Desktop>llama-server --model "C:\My Files\Other\AI\gpt-oss-20b\gpt-oss-20b-F16.gguf" --temperature 1.00 --top-p 1.00 --ctx-size 65536 -lv 4
  2. 0.00.080.494 I common_params_print_info: build 9590 (d2462f8f7) with Clang 20.1.8 for Windows x86_64
  3. 0.00.080.496 I log_info: verbosity = 4 (adjust with the `-lv N` CLI arg)
  4. 0.00.080.497 I device_info:
  5. 0.00.213.310 I  - CUDA0 : NVIDIA GeForce RTX 4060 Ti (16379 MiB, 15225 MiB free)
  6. 0.00.213.317 I  - CPU : 13th Gen Intel(R) Core(TM) i5-13500 (32465 MiB, 27945 MiB free)
  7. 0.00.213.366 I system_info: n_threads = 14 (n_threads_batch = 14) / 20 | CUDA : ARCHS = 500,610,700,750,800,860,890,900 | USE_GRAPHS = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX_VNNI = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 |
  8. 0.00.213.370 I srv llama_server: n_parallel is set to auto, using n_parallel = 4 and kv_unified = true
  9. 0.00.213.394 I srv init: running without SSL
  10. 0.00.213.408 I srv init: using 19 threads for HTTP server
  11. 0.00.213.504 I srv start: binding port with default address family
  12. 0.00.220.201 I srv llama_server: loading model
  13. 0.00.220.278 I srv load_model: loading model 'C:\My Files\Other\AI\gpt-oss-20b\gpt-oss-20b-F16.gguf'
  14. 0.00.220.312 I common_init_result: fitting params to device memory ...
  15. 0.00.220.312 I common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on)
  16. 0.00.220.314 I common_params_fit_impl: getting device memory data for initial parameters:
  17. 0.01.100.266 I common_memory_breakdown_print: | memory breakdown [MiB] | total free self model context compute unaccounted |
  18. 0.01.100.268 I common_memory_breakdown_print: | - CUDA0 (RTX 4060 Ti) | 16379 = 15225 + (13748 = 12036 + 1560 + 151) + -12593 |
  19. 0.01.100.269 I common_memory_breakdown_print: | - Host | 1180 = 1104 + 0 + 76 |
  20. 0.01.140.458 I common_params_fit_impl: projected to use 13748 MiB of device memory vs. 15225 MiB of free device memory
  21. 0.01.140.461 I common_params_fit_impl: will leave 1476 >= 1024 MiB of free device memory, no changes needed
  22. 0.01.140.462 I common_fit_params: successfully fit params to free device memory
  23. 0.01.140.466 I common_fit_params: fitting params to free memory took 0.78 seconds
  24. 0.01.220.772 I llama_model_loader: loaded meta data with 37 key-value pairs and 459 tensors from C:\My Files\Other\AI\gpt-oss-20b\gpt-oss-20b-F16.gguf (version GGUF V3 (latest))
  25. 0.01.220.783 I llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
  26. 0.01.220.791 I llama_model_loader: - kv 0: general.architecture str = gpt-oss
  27. 0.01.220.792 I llama_model_loader: - kv 1: general.type str = model
  28. 0.01.220.792 I llama_model_loader: - kv 2: general.name str = Gpt-Oss-20B
  29. 0.01.220.793 I llama_model_loader: - kv 3: general.basename str = Gpt-Oss-20B
  30. 0.01.220.794 I llama_model_loader: - kv 4: general.quantized_by str = Unsloth
  31. 0.01.220.794 I llama_model_loader: - kv 5: general.size_label str = 20B
  32. 0.01.220.795 I llama_model_loader: - kv 6: general.license str = apache-2.0
  33. 0.01.220.796 I llama_model_loader: - kv 7: general.repo_url str = https://huggingface.co/unsloth
  34. 0.01.220.804 I llama_model_loader: - kv 8: general.tags arr[str,2] = ["vllm", "text-generation"]
  35. 0.01.220.805 I llama_model_loader: - kv 9: gpt-oss.block_count u32 = 24
  36. 0.01.220.805 I llama_model_loader: - kv 10: gpt-oss.context_length u32 = 131072
  37. 0.01.220.806 I llama_model_loader: - kv 11: gpt-oss.embedding_length u32 = 2880
  38. 0.01.220.806 I llama_model_loader: - kv 12: gpt-oss.feed_forward_length u32 = 2880
  39. 0.01.220.807 I llama_model_loader: - kv 13: gpt-oss.attention.head_count u32 = 64
  40. 0.01.220.807 I llama_model_loader: - kv 14: gpt-oss.attention.head_count_kv u32 = 8
  41. 0.01.220.811 I llama_model_loader: - kv 15: gpt-oss.rope.freq_base f32 = 150000.000000
  42. 0.01.220.813 I llama_model_loader: - kv 16: gpt-oss.attention.layer_norm_rms_epsilon f32 = 0.000010
  43. 0.01.220.814 I llama_model_loader: - kv 17: gpt-oss.expert_count u32 = 32
  44. 0.01.220.814 I llama_model_loader: - kv 18: gpt-oss.expert_used_count u32 = 4
  45. 0.01.220.815 I llama_model_loader: - kv 19: gpt-oss.attention.key_length u32 = 64
  46. 0.01.220.815 I llama_model_loader: - kv 20: gpt-oss.attention.value_length u32 = 64
  47. 0.01.220.815 I llama_model_loader: - kv 21: general.file_type u32 = 1
  48. 0.01.220.816 I llama_model_loader: - kv 22: gpt-oss.attention.sliding_window u32 = 128
  49. 0.01.220.816 I llama_model_loader: - kv 23: gpt-oss.expert_feed_forward_length u32 = 2880
  50. 0.01.220.817 I llama_model_loader: - kv 24: gpt-oss.rope.scaling.type str = yarn
  51. 0.01.220.818 I llama_model_loader: - kv 25: gpt-oss.rope.scaling.factor f32 = 32.000000
  52. 0.01.220.818 I llama_model_loader: - kv 26: gpt-oss.rope.scaling.original_context_length u32 = 4096
  53. 0.01.220.818 I llama_model_loader: - kv 27: general.quantization_version u32 = 2
  54. 0.01.220.819 I llama_model_loader: - kv 28: tokenizer.ggml.model str = gpt2
  55. 0.01.220.819 I llama_model_loader: - kv 29: tokenizer.ggml.pre str = gpt-4o
  56. 0.01.252.954 I llama_model_loader: - kv 30: tokenizer.ggml.tokens arr[str,201088] = ["!", "\"", "#", "$", "%", "&", "'", ...
  57. 0.01.263.658 I llama_model_loader: - kv 31: tokenizer.ggml.token_type arr[i32,201088] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
  58. 0.01.329.782 I llama_model_loader: - kv 32: tokenizer.ggml.merges arr[str,446189] = ["Ġ Ġ", "Ġ ĠĠĠ", "ĠĠ ĠĠ", "...
  59. 0.01.329.786 I llama_model_loader: - kv 33: tokenizer.ggml.bos_token_id u32 = 199998
  60. 0.01.329.786 I llama_model_loader: - kv 34: tokenizer.ggml.eos_token_id u32 = 200002
  61. 0.01.329.787 I llama_model_loader: - kv 35: tokenizer.ggml.padding_token_id u32 = 200017
  62. 0.01.329.792 I llama_model_loader: - kv 36: tokenizer.chat_template str = {# Chat template fixes by Unsloth #}\n...
  63. 0.01.329.793 I llama_model_loader: - type f32: 289 tensors
  64. 0.01.329.794 I llama_model_loader: - type f16: 98 tensors
  65. 0.01.329.794 I llama_model_loader: - type mxfp4: 72 tensors
  66. 0.01.329.795 I print_info: file format = GGUF V3 (latest)
  67. 0.01.329.796 I print_info: file type = F16
  68. 0.01.329.799 I print_info: file size = 12.83 GiB (5.27 BPW)
  69. 0.01.415.794 I llama_prepare_model_devices: using device CUDA0 (NVIDIA GeForce RTX 4060 Ti) (0000:01:00.0) - 15225 MiB free
  70. 0.01.551.882 I load: 0 unused tokens
  71. 0.01.568.762 W load: setting token '<|constrain|>' (200003) attribute to USER_DEFINED (16), old attributes: 8
  72. 0.01.568.879 W load: setting token '<|message|>' (200008) attribute to USER_DEFINED (16), old attributes: 8
  73. 0.01.571.803 W load: setting token '<|channel|>' (200005) attribute to USER_DEFINED (16), old attributes: 8
  74. 0.01.571.805 W load: setting token '<|start|>' (200006) attribute to USER_DEFINED (16), old attributes: 8
  75. 0.01.571.809 I load: printing all EOG tokens:
  76. 0.01.571.810 I load: - 199999 ('<|endoftext|>')
  77. 0.01.571.810 I load: - 200002 ('<|return|>')
  78. 0.01.571.810 I load: - 200007 ('<|end|>')
  79. 0.01.571.810 I load: - 200012 ('<|call|>')
  80. 0.01.571.813 W load: special_eog_ids contains both '<|return|>' and '<|call|>', or '<|calls|>' and '<|flush|>' tokens, removing '<|end|>' token from EOG list
  81. 0.01.572.099 I load: special tokens cache size = 21
  82. 0.01.599.124 I load: token to piece cache size = 1.3332 MB
  83. 0.01.599.134 I print_info: arch = gpt-oss
  84. 0.01.599.134 I print_info: vocab_only = 0
  85. 0.01.599.134 I print_info: no_alloc = 0
  86. 0.01.599.135 I print_info: n_ctx_train = 131072
  87. 0.01.599.135 I print_info: n_embd_inp = 2880
  88. 0.01.599.135 I print_info: n_embd = 2880
  89. 0.01.599.135 I print_info: n_embd_out = 2880
  90. 0.01.599.136 I print_info: n_layer = 24
  91. 0.01.599.136 I print_info: n_layer_all = 24
  92. 0.01.599.168 I print_info: n_head = 64
  93. 0.01.599.170 I print_info: n_head_kv = 8
  94. 0.01.599.170 I print_info: n_rot = 64
  95. 0.01.599.171 I print_info: n_swa = 128
  96. 0.01.599.171 I print_info: is_swa_any = 1
  97. 0.01.599.171 I print_info: n_embd_head_k = 64
  98. 0.01.599.171 I print_info: n_embd_head_v = 64
  99. 0.01.599.172 I print_info: n_gqa = 8
  100. 0.01.599.174 I print_info: n_embd_k_gqa = 512
  101. 0.01.599.175 I print_info: n_embd_v_gqa = 512
  102. 0.01.599.176 I print_info: f_norm_eps = 0.0e+00
  103. 0.01.599.177 I print_info: f_norm_rms_eps = 1.0e-05
  104. 0.01.599.177 I print_info: f_clamp_kqv = 0.0e+00
  105. 0.01.599.177 I print_info: f_max_alibi_bias = 0.0e+00
  106. 0.01.599.178 I print_info: f_logit_scale = 0.0e+00
  107. 0.01.599.178 I print_info: f_attn_scale = 0.0e+00
  108. 0.01.599.178 I print_info: f_attn_value_scale = 0.0000
  109. 0.01.599.180 I print_info: n_ff = 2880
  110. 0.01.599.180 I print_info: n_expert = 32
  111. 0.01.599.181 I print_info: n_expert_used = 4
  112. 0.01.599.181 I print_info: n_expert_groups = 0
  113. 0.01.599.181 I print_info: n_group_used = 0
  114. 0.01.599.181 I print_info: causal attn = 1
  115. 0.01.599.182 I print_info: pooling type = -1
  116. 0.01.599.182 I print_info: rope type = 2
  117. 0.01.599.182 I print_info: rope scaling = yarn
  118. 0.01.599.183 I print_info: freq_base_train = 150000.0
  119. 0.01.599.184 I print_info: freq_scale_train = 0.03125
  120. 0.01.599.184 I print_info: freq_base_swa = 150000.0
  121. 0.01.599.185 I print_info: freq_scale_swa = 0.03125
  122. 0.01.599.185 I print_info: n_embd_head_k_swa = 64
  123. 0.01.599.185 I print_info: n_embd_head_v_swa = 64
  124. 0.01.599.185 I print_info: n_rot_swa = 64
  125. 0.01.599.185 I print_info: n_ctx_orig_yarn = 4096
  126. 0.01.599.186 I print_info: rope_yarn_log_mul = 0.0000
  127. 0.01.599.186 I print_info: rope_finetuned = unknown
  128. 0.01.599.187 I print_info: model type = 20B
  129. 0.01.599.187 I print_info: model params = 20.91 B
  130. 0.01.599.188 I print_info: general.name = Gpt-Oss-20B
  131. 0.01.599.188 I print_info: n_ff_exp = 2880
  132. 0.01.599.188 I print_info: vocab type = BPE
  133. 0.01.599.189 I print_info: n_vocab = 201088
  134. 0.01.599.189 I print_info: n_merges = 446189
  135. 0.01.599.189 I print_info: BOS token = 199998 '<|startoftext|>'
  136. 0.01.599.190 I print_info: EOS token = 200002 '<|return|>'
  137. 0.01.599.190 I print_info: EOT token = 199999 '<|endoftext|>'
  138. 0.01.599.190 I print_info: PAD token = 200017 '<|reserved_200017|>'
  139. 0.01.599.191 I print_info: LF token = 198 'Ċ'
  140. 0.01.599.191 I print_info: EOG token = 199999 '<|endoftext|>'
  141. 0.01.599.192 I print_info: EOG token = 200002 '<|return|>'
  142. 0.01.599.192 I print_info: EOG token = 200012 '<|call|>'
  143. 0.01.599.192 I print_info: max token length = 256
  144. 0.01.599.193 I load_tensors: loading model tensors, this can take a while... (mmap = true, direct_io = false)
  145. 0.02.571.941 I load_tensors: offloading output layer to GPU
  146. 0.02.571.944 I load_tensors: offloading 23 repeating layers to GPU
  147. 0.02.571.945 I load_tensors: offloaded 25/25 layers to GPU
  148. 0.02.571.949 I load_tensors: CPU_Mapped model buffer size = 1104.61 MiB
  149. 0.02.571.951 I load_tensors: CUDA0 model buffer size = 12036.68 MiB
  150. ....................................................................................
  151. 0.06.012.452 I common_init_result: added <|endoftext|> logit bias = -inf
  152. 0.06.012.454 I common_init_result: added <|return|> logit bias = -inf
  153. 0.06.012.454 I common_init_result: added <|call|> logit bias = -inf
  154. 0.06.012.743 I llama_context: constructing llama_context
  155. 0.06.012.745 I llama_context: n_seq_max = 4
  156. 0.06.012.746 I llama_context: n_ctx = 65536
  157. 0.06.012.746 I llama_context: n_ctx_seq = 65536
  158. 0.06.012.746 I llama_context: n_batch = 2048
  159. 0.06.012.746 I llama_context: n_ubatch = 512
  160. 0.06.012.747 I llama_context: causal_attn = 1
  161. 0.06.012.747 I llama_context: flash_attn = auto
  162. 0.06.012.747 I llama_context: kv_unified = true
  163. 0.06.012.750 I llama_context: freq_base = 150000.0
  164. 0.06.012.751 I llama_context: freq_scale = 0.03125
  165. 0.06.012.751 I llama_context: n_rs_seq = 0
  166. 0.06.012.751 I llama_context: n_outputs_max = 4
  167. 0.06.012.752 W llama_context: n_ctx_seq (65536) < n_ctx_train (131072) -- the full capacity of the model will not be utilized
  168. 0.06.013.993 I llama_context: CUDA_Host output buffer size = 3.07 MiB
  169. 0.06.013.994 I llama_kv_cache_iswa: creating non-SWA KV cache, size = 65536 cells
  170. 0.06.016.414 I llama_kv_cache: CUDA0 KV buffer size = 1536.00 MiB
  171. 0.06.053.353 I llama_kv_cache: size = 1536.00 MiB ( 65536 cells, 12 layers, 4/1 seqs), K (f16): 768.00 MiB, V (f16): 768.00 MiB
  172. 0.06.053.357 I llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 64
  173. 0.06.053.357 I llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 64
  174. 0.06.053.358 I llama_kv_cache_iswa: creating SWA KV cache, size = 1024 cells
  175. 0.06.053.448 I llama_kv_cache: CUDA0 KV buffer size = 24.00 MiB
  176. 0.06.054.344 I llama_kv_cache: size = 24.00 MiB ( 1024 cells, 12 layers, 4/1 seqs), K (f16): 12.00 MiB, V (f16): 12.00 MiB
  177. 0.06.054.346 I llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 64
  178. 0.06.054.346 I llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 64
  179. 0.06.054.352 I sched_reserve: reserving ...
  180. 0.06.056.990 I sched_reserve: Flash Attention was auto, set to enabled
  181. 0.06.056.992 I sched_reserve: resolving fused Gated Delta Net support:
  182. 0.06.057.269 I sched_reserve: fused Gated Delta Net (autoregressive) enabled
  183. 0.06.057.538 I sched_reserve: fused Gated Delta Net (chunked) enabled
  184. 0.06.074.452 I sched_reserve: CUDA0 compute buffer size = 151.77 MiB
  185. 0.06.074.455 I sched_reserve: CUDA_Host compute buffer size = 76.27 MiB
  186. 0.06.074.455 I sched_reserve: graph nodes = 1350
  187. 0.06.074.455 I sched_reserve: graph splits = 2
  188. 0.06.074.456 I sched_reserve: reserve took 20.10 ms, sched copies = 1
  189. 0.06.074.738 I common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
  190. 0.06.130.606 I srv load_model: initializing slots, n_slots = 4
  191. 0.06.169.138 W common_speculative_init: no implementations specified for speculative decoding
  192. 0.06.169.141 I slot load_model: id 0 | task -1 | new slot, n_ctx = 65536
  193. 0.06.169.145 I slot load_model: id 1 | task -1 | new slot, n_ctx = 65536
  194. 0.06.169.145 I slot load_model: id 2 | task -1 | new slot, n_ctx = 65536
  195. 0.06.169.146 I slot load_model: id 3 | task -1 | new slot, n_ctx = 65536
  196. 0.06.169.290 I srv load_model: prompt cache is enabled, size limit: 8192 MiB
  197. 0.06.169.291 I srv load_model: use `--cache-ram 0` to disable the prompt cache
  198. 0.06.169.291 I srv load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391
  199. 0.06.169.292 I srv load_model: context checkpoints enabled, max = 32, min spacing = 256
  200. 0.06.169.310 I srv init: idle slots will be saved to prompt cache and cleared upon starting a new task
  201. 0.06.173.252 I init: chat template, example_format: '<|start|>system<|message|>You are ChatGPT, a large language model trained by OpenAI.
  202. Knowledge cutoff: 2024-06
  203. Current date: 2026-06-20
  204.  
  205. Reasoning: medium
  206.  
  207. # Valid channels: analysis, commentary, final. Channel must be included for every message.<|end|><|start|>developer<|message|># Instructions
  208.  
  209. You are a helpful assistant<|end|><|start|>user<|message|>Hello<|end|><|start|>assistant<|channel|>final<|message|>Hi there<|end|><|start|>user<|message|>How are you?<|end|><|start|>assistant'
  210. 0.06.173.849 I srv init: init: chat template, thinking = 1
  211. 0.06.173.863 I srv llama_server: model loaded
  212. 0.06.173.864 I srv llama_server: server is listening on http://127.0.0.1:8080
  213. 0.06.173.867 I srv update_slots: all slots are idle
Add Comment
Please, Sign In to add comment
Public Pastes
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!