[GH-ISSUE #10923] ollama 0.90 not use GPU #69246

New Issue

GiteaMirror · 2026-05-04T17:34:48-05:00

GiteaMirror commented

2026-05-04 17:34:48 -05:00

Originally created by @wlsoft2006 on GitHub (May 31, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/10923

What is the issue?

Relevant log output

5月 30 22:16:51 ai systemd[1]: Started Ollama Service.
5月 30 22:17:14 ai ollama[1180]: time=2025-05-30T22:17:13.277+08:00 level=INFO source=routes.go:1234 msg="server config" env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VI>
5月 30 22:17:15 ai ollama[1180]: time=2025-05-30T22:17:15.272+08:00 level=INFO source=images.go:479 msg="total blobs: 62"
5月 30 22:17:15 ai ollama[1180]: time=2025-05-30T22:17:15.275+08:00 level=INFO source=images.go:486 msg="total unused blobs removed: 0"
5月 30 22:17:15 ai ollama[1180]: time=2025-05-30T22:17:15.276+08:00 level=INFO source=routes.go:1287 msg="Listening on [::]:11434 (version 0.9.0)"
5月 30 22:17:15 ai ollama[1180]: time=2025-05-30T22:17:15.384+08:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs"
5月 30 22:17:28 ai ollama[1180]: time=2025-05-30T22:17:28.371+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-9ef6fcbc-ad1a-6752-6985-09e1e601e662 library=cud>
5月 30 22:17:28 ai ollama[1180]: time=2025-05-30T22:17:28.372+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-083bdbe0-7f5c-38d0-aa01-2635abd8764e library=cud>
5月 31 08:44:56 ai ollama[1180]: [GIN] 2025/05/31 - 08:44:56 | 200 |   26.883628ms |       127.0.0.1 | HEAD     "/"
5月 31 08:44:56 ai ollama[1180]: [GIN] 2025/05/31 - 08:44:56 | 200 |   22.248672ms |       127.0.0.1 | GET      "/api/ps"
5月 31 08:45:00 ai ollama[1180]: [GIN] 2025/05/31 - 08:45:00 | 200 |      68.544µs |       127.0.0.1 | HEAD     "/"
5月 31 08:45:01 ai ollama[1180]: [GIN] 2025/05/31 - 08:45:01 | 200 |  620.481972ms |       127.0.0.1 | POST     "/api/show"
5月 31 08:45:02 ai ollama[1180]: time=2025-05-31T08:45:02.009+08:00 level=INFO source=sched.go:788 msg="new model will fit in available VRAM in single GPU, loading" model=/mnt/di>
5月 31 08:45:02 ai ollama[1180]: time=2025-05-31T08:45:02.640+08:00 level=INFO source=server.go:135 msg="system memory" total="54.9 GiB" free="50.8 GiB" free_swap="4.0 GiB"
5月 31 08:45:02 ai ollama[1180]: time=2025-05-31T08:45:02.641+08:00 level=INFO source=server.go:168 msg=offload library=cuda layers.requested=-1 layers.model=37 layers.offload=37>
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: loaded meta data with 32 key-value pairs and 399 tensors from /mnt/diskb/ollama_models/models/blobs/sha256-e6a7edc1a4d7d9b2de>
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   0:                       general.architecture str              = qwen3
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   1:                               general.type str              = model
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   2:                               general.name str              = DeepSeek R1 0528 Qwen3 8B
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   3:                           general.basename str              = DeepSeek-R1-0528-Qwen3
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   4:                         general.size_label str              = 8B
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   5:                            general.license str              = mit
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   6:                          qwen3.block_count u32              = 36
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   7:                       qwen3.context_length u32              = 131072
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   8:                     qwen3.embedding_length u32              = 4096
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv   9:                  qwen3.feed_forward_length u32              = 12288
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  10:                 qwen3.attention.head_count u32              = 32
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  11:              qwen3.attention.head_count_kv u32              = 8
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  12:                       qwen3.rope.freq_base f32              = 1000000.000000
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  13:     qwen3.attention.layer_norm_rms_epsilon f32              = 0.000001
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  14:                 qwen3.attention.key_length u32              = 128
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  15:               qwen3.attention.value_length u32              = 128
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  16:                    qwen3.rope.scaling.type str              = yarn
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  17:                  qwen3.rope.scaling.factor f32              = 4.000000
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  18: qwen3.rope.scaling.original_context_length u32              = 32768
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  19:                       tokenizer.ggml.model str              = gpt2
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  20:                         tokenizer.ggml.pre str              = qwen2
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  21:                      tokenizer.ggml.tokens arr[str,151936]  = ["!", "\"", "#", "$", "%", "&", "'", ...
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  22:                  tokenizer.ggml.token_type arr[i32,151936]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  23:                      tokenizer.ggml.merges arr[str,151387]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  24:                tokenizer.ggml.bos_token_id u32              = 151643
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  25:                tokenizer.ggml.eos_token_id u32              = 151645
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  26:            tokenizer.ggml.padding_token_id u32              = 151645
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  27:               tokenizer.ggml.add_bos_token bool             = false
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  28:               tokenizer.ggml.add_eos_token bool             = false
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  29:                    tokenizer.chat_template str              = {% if not add_generation_prompt is de...
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  30:               general.quantization_version u32              = 2
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv  31:                          general.file_type u32              = 15
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - type  f32:  145 tensors
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - type  f16:   36 tensors
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - type q4_K:  199 tensors
5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - type q6_K:   19 tensors
5月 31 08:45:02 ai ollama[1180]: print_info: file format = GGUF V3 (latest)
5月 31 08:45:02 ai ollama[1180]: print_info: file type   = Q4_K - Medium
5月 31 08:45:02 ai ollama[1180]: print_info: file size   = 4.86 GiB (5.10 BPW)
5月 31 08:45:03 ai ollama[1180]: load: special_eos_id is not in special_eog_ids - the tokenizer config may be incorrect
5月 31 08:45:03 ai ollama[1180]: load: special tokens cache size = 28
5月 31 08:45:03 ai ollama[1180]: load: token to piece cache size = 0.9311 MB
5月 31 08:45:03 ai ollama[1180]: print_info: arch             = qwen3
5月 31 08:45:03 ai ollama[1180]: print_info: vocab_only       = 1
5月 31 08:45:03 ai ollama[1180]: print_info: model type       = ?B
5月 31 08:45:03 ai ollama[1180]: print_info: model params     = 8.19 B
5月 31 08:45:03 ai ollama[1180]: print_info: general.name     = DeepSeek R1 0528 Qwen3 8B
5月 31 08:45:03 ai ollama[1180]: print_info: vocab type       = BPE
5月 31 08:45:03 ai ollama[1180]: print_info: n_vocab          = 151936
5月 31 08:45:03 ai ollama[1180]: print_info: n_merges         = 151387
5月 31 08:45:03 ai ollama[1180]: print_info: BOS token        = 151643 '<｜begin▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOS token        = 151645 '<｜end▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOT token        = 151645 '<｜end▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: PAD token        = 151645 '<｜end▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: LF token         = 198 'Ċ'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM PRE token    = 151659 '<|fim_prefix|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM SUF token    = 151661 '<|fim_suffix|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM MID token    = 151660 '<|fim_middle|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM PAD token    = 151662 '<|fim_pad|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM REP token    = 151663 '<|repo_name|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM SEP token    = 151664 '<|file_sep|>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOG token        = 151645 '<｜end▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOG token        = 151662 '<|fim_pad|>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOG token        = 151663 '<|repo_name|>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOG token        = 151664 '<|file_sep|>'
5月 31 08:45:03 ai ollama[1180]: print_info: max token length = 256
5月 31 08:45:03 ai ollama[1180]: llama_model_load: vocab only - skipping tensors
5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.101+08:00 level=INFO source=server.go:431 msg="starting llama server" cmd="/usr/local/bin/ollama runner --model /mnt/dis>
5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.102+08:00 level=INFO source=sched.go:483 msg="loaded runners" count=1
5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.102+08:00 level=INFO source=server.go:591 msg="waiting for llama runner to start responding"
5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.103+08:00 level=INFO source=server.go:625 msg="waiting for server to become available" status="llm server not responding"
5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.126+08:00 level=INFO source=runner.go:815 msg="starting go runner"
5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.126+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc)
5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.132+08:00 level=INFO source=runner.go:874 msg="Server listening on 127.0.0.1:40057"
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: loaded meta data with 32 key-value pairs and 399 tensors from /mnt/diskb/ollama_models/models/blobs/sha256-e6a7edc1a4d7d9b2de>
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   0:                       general.architecture str              = qwen3
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   1:                               general.type str              = model
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   2:                               general.name str              = DeepSeek R1 0528 Qwen3 8B
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   3:                           general.basename str              = DeepSeek-R1-0528-Qwen3
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   4:                         general.size_label str              = 8B
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   5:                            general.license str              = mit
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   6:                          qwen3.block_count u32              = 36
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   7:                       qwen3.context_length u32              = 131072
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   8:                     qwen3.embedding_length u32              = 4096
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv   9:                  qwen3.feed_forward_length u32              = 12288
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  10:                 qwen3.attention.head_count u32              = 32
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  11:              qwen3.attention.head_count_kv u32              = 8
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  12:                       qwen3.rope.freq_base f32              = 1000000.000000
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  13:     qwen3.attention.layer_norm_rms_epsilon f32              = 0.000001
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  14:                 qwen3.attention.key_length u32              = 128
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  15:               qwen3.attention.value_length u32              = 128
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  16:                    qwen3.rope.scaling.type str              = yarn
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  17:                  qwen3.rope.scaling.factor f32              = 4.000000
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  18: qwen3.rope.scaling.original_context_length u32              = 32768
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  19:                       tokenizer.ggml.model str              = gpt2
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  20:                         tokenizer.ggml.pre str              = qwen2
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  21:                      tokenizer.ggml.tokens arr[str,151936]  = ["!", "\"", "#", "$", "%", "&", "'", ...
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  22:                  tokenizer.ggml.token_type arr[i32,151936]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  23:                      tokenizer.ggml.merges arr[str,151387]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  24:                tokenizer.ggml.bos_token_id u32              = 151643
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  25:                tokenizer.ggml.eos_token_id u32              = 151645
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  26:            tokenizer.ggml.padding_token_id u32              = 151645
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  27:               tokenizer.ggml.add_bos_token bool             = false
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  28:               tokenizer.ggml.add_eos_token bool             = false
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  29:                    tokenizer.chat_template str              = {% if not add_generation_prompt is de...
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  30:               general.quantization_version u32              = 2
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv  31:                          general.file_type u32              = 15
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - type  f32:  145 tensors
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - type  f16:   36 tensors
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - type q4_K:  199 tensors
5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - type q6_K:   19 tensors
5月 31 08:45:03 ai ollama[1180]: print_info: file format = GGUF V3 (latest)
5月 31 08:45:03 ai ollama[1180]: print_info: file type   = Q4_K - Medium
5月 31 08:45:03 ai ollama[1180]: print_info: file size   = 4.86 GiB (5.10 BPW)
5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.356+08:00 level=INFO source=server.go:625 msg="waiting for server to become available" status="llm server loading model"
5月 31 08:45:03 ai ollama[1180]: load: special_eos_id is not in special_eog_ids - the tokenizer config may be incorrect
5月 31 08:45:03 ai ollama[1180]: load: special tokens cache size = 28
5月 31 08:45:03 ai ollama[1180]: load: token to piece cache size = 0.9311 MB
5月 31 08:45:03 ai ollama[1180]: print_info: arch             = qwen3
5月 31 08:45:03 ai ollama[1180]: print_info: vocab_only       = 0
5月 31 08:45:03 ai ollama[1180]: print_info: n_ctx_train      = 131072
5月 31 08:45:03 ai ollama[1180]: print_info: n_embd           = 4096
5月 31 08:45:03 ai ollama[1180]: print_info: n_layer          = 36
5月 31 08:45:03 ai ollama[1180]: print_info: n_head           = 32
5月 31 08:45:03 ai ollama[1180]: print_info: n_head_kv        = 8
5月 31 08:45:03 ai ollama[1180]: print_info: n_rot            = 128
5月 31 08:45:03 ai ollama[1180]: print_info: n_swa            = 0
5月 31 08:45:03 ai ollama[1180]: print_info: n_swa_pattern    = 1
5月 31 08:45:03 ai ollama[1180]: print_info: n_embd_head_k    = 128
5月 31 08:45:03 ai ollama[1180]: print_info: n_embd_head_v    = 128
5月 31 08:45:03 ai ollama[1180]: print_info: n_gqa            = 4
5月 31 08:45:03 ai ollama[1180]: print_info: n_embd_k_gqa     = 1024
5月 31 08:45:03 ai ollama[1180]: print_info: n_embd_v_gqa     = 1024
5月 31 08:45:03 ai ollama[1180]: print_info: f_norm_eps       = 0.0e+00
5月 31 08:45:03 ai ollama[1180]: print_info: f_norm_rms_eps   = 1.0e-06
5月 31 08:45:03 ai ollama[1180]: print_info: f_clamp_kqv      = 0.0e+00
5月 31 08:45:03 ai ollama[1180]: print_info: f_max_alibi_bias = 0.0e+00
5月 31 08:45:03 ai ollama[1180]: print_info: f_logit_scale    = 0.0e+00
5月 31 08:45:03 ai ollama[1180]: print_info: f_attn_scale     = 0.0e+00
5月 31 08:45:03 ai ollama[1180]: print_info: n_ff             = 12288
5月 31 08:45:03 ai ollama[1180]: print_info: n_expert         = 0
5月 31 08:45:03 ai ollama[1180]: print_info: n_expert_used    = 0
5月 31 08:45:03 ai ollama[1180]: print_info: causal attn      = 1
5月 31 08:45:03 ai ollama[1180]: print_info: pooling type     = 0
5月 31 08:45:03 ai ollama[1180]: print_info: rope type        = 2
5月 31 08:45:03 ai ollama[1180]: print_info: rope scaling     = yarn
5月 31 08:45:03 ai ollama[1180]: print_info: freq_base_train  = 1000000.0
5月 31 08:45:03 ai ollama[1180]: print_info: freq_scale_train = 0.25
5月 31 08:45:03 ai ollama[1180]: print_info: n_ctx_orig_yarn  = 32768
5月 31 08:45:03 ai ollama[1180]: print_info: rope_finetuned   = unknown
5月 31 08:45:03 ai ollama[1180]: print_info: ssm_d_conv       = 0
5月 31 08:45:03 ai ollama[1180]: print_info: ssm_d_inner      = 0
5月 31 08:45:03 ai ollama[1180]: print_info: ssm_d_state      = 0
5月 31 08:45:03 ai ollama[1180]: print_info: ssm_dt_rank      = 0
5月 31 08:45:03 ai ollama[1180]: print_info: ssm_dt_b_c_rms   = 0
5月 31 08:45:03 ai ollama[1180]: print_info: model type       = 8B
5月 31 08:45:03 ai ollama[1180]: print_info: model params     = 8.19 B
5月 31 08:45:03 ai ollama[1180]: print_info: general.name     = DeepSeek R1 0528 Qwen3 8B
5月 31 08:45:03 ai ollama[1180]: print_info: vocab type       = BPE
5月 31 08:45:03 ai ollama[1180]: print_info: n_vocab          = 151936
5月 31 08:45:03 ai ollama[1180]: print_info: n_merges         = 151387
5月 31 08:45:03 ai ollama[1180]: print_info: BOS token        = 151643 '<｜begin▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOS token        = 151645 '<｜end▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOT token        = 151645 '<｜end▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: PAD token        = 151645 '<｜end▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: LF token         = 198 'Ċ'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM PRE token    = 151659 '<|fim_prefix|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM SUF token    = 151661 '<|fim_suffix|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM MID token    = 151660 '<|fim_middle|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM PAD token    = 151662 '<|fim_pad|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM REP token    = 151663 '<|repo_name|>'
5月 31 08:45:03 ai ollama[1180]: print_info: FIM SEP token    = 151664 '<|file_sep|>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOG token        = 151645 '<｜end▁of▁sentence｜>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOG token        = 151662 '<|fim_pad|>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOG token        = 151663 '<|repo_name|>'
5月 31 08:45:03 ai ollama[1180]: print_info: EOG token        = 151664 '<|file_sep|>'
5月 31 08:45:03 ai ollama[1180]: print_info: max token length = 256
5月 31 08:45:03 ai ollama[1180]: load_tensors: loading model tensors, this can take a while... (mmap = true)
5月 31 08:46:30 ai ollama[1180]: load_tensors:   CPU_Mapped model buffer size =  4977.62 MiB
5月 31 08:46:30 ai ollama[1180]: llama_context: constructing llama_context
5月 31 08:46:30 ai ollama[1180]: llama_context: n_seq_max     = 2
5月 31 08:46:30 ai ollama[1180]: llama_context: n_ctx         = 8192
5月 31 08:46:30 ai ollama[1180]: llama_context: n_ctx_per_seq = 4096
5月 31 08:46:30 ai ollama[1180]: llama_context: n_batch       = 1024
5月 31 08:46:30 ai ollama[1180]: llama_context: n_ubatch      = 512
5月 31 08:46:30 ai ollama[1180]: llama_context: causal_attn   = 1
5月 31 08:46:30 ai ollama[1180]: llama_context: flash_attn    = 0
5月 31 08:46:30 ai ollama[1180]: llama_context: freq_base     = 1000000.0
5月 31 08:46:30 ai ollama[1180]: llama_context: freq_scale    = 0.25
5月 31 08:46:30 ai ollama[1180]: llama_context: n_ctx_per_seq (4096) < n_ctx_train (131072) -- the full capacity of the model will not be utilized
5月 31 08:46:30 ai ollama[1180]: llama_context:        CPU  output buffer size =     1.19 MiB
5月 31 08:46:30 ai ollama[1180]: llama_kv_cache_unified: kv_size = 8192, type_k = 'f16', type_v = 'f16', n_layer = 36, can_shift = 1, padding = 32
5月 31 08:46:32 ai ollama[1180]: llama_kv_cache_unified:        CPU KV buffer size =  1152.00 MiB
5月 31 08:46:32 ai ollama[1180]: llama_kv_cache_unified: KV self size  = 1152.00 MiB, K (f16):  576.00 MiB, V (f16):  576.00 MiB
5月 31 08:46:32 ai ollama[1180]: llama_context:        CPU compute buffer size =   560.01 MiB
5月 31 08:46:32 ai ollama[1180]: llama_context: graph nodes  = 1374
5月 31 08:46:32 ai ollama[1180]: llama_context: graph splits = 1
5月 31 08:46:33 ai ollama[1180]: time=2025-05-31T08:46:33.014+08:00 level=INFO source=server.go:630 msg="llama runner started in 89.91 seconds"
5月 31 08:46:33 ai ollama[1180]: [GIN] 2025/05/31 - 08:46:33 | 200 |         1m31s |       127.0.0.1 | POST     "/api/generate"
5月 31 08:48:41 ai ollama[1180]: [GIN] 2025/05/31 - 08:48:41 | 200 |         1m55s |       127.0.0.1 | POST     "/api/chat"
5月 31 09:30:29 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:29 | 200 |     405.308µs |       127.0.0.1 | HEAD     "/"
5月 31 09:30:29 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:29 | 200 |   90.556909ms |       127.0.0.1 | POST     "/api/show"
5月 31 09:30:29 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:29 | 200 |   49.681837ms |       127.0.0.1 | POST     "/api/generate"
5月 31 09:30:41 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:41 | 200 |      83.707µs |       127.0.0.1 | HEAD     "/"
5月 31 09:30:42 ai ollama[1180]: time=2025-05-31T09:30:42.206+08:00 level=WARN source=routes.go:920 msg="bad manifest filepath" name=registry.ollama.ai/library/gemma3:latest erro>
5月 31 09:30:42 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:42 | 200 |  326.388416ms |       127.0.0.1 | GET      "/api/tags"
5月 31 09:30:55 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:55 | 200 |       39.07µs |       127.0.0.1 | HEAD     "/"
5月 31 09:30:55 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:55 | 200 |  437.917824ms |       127.0.0.1 | POST     "/api/show"
5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.436+08:00 level=INFO source=sched.go:548 msg="updated VRAM based on existing loaded models" gpu=GPU-9ef6fcbc-ad1a-6752-6>
5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.436+08:00 level=INFO source=sched.go:548 msg="updated VRAM based on existing loaded models" gpu=GPU-083bdbe0-7f5c-38d0-a>
5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.437+08:00 level=INFO source=sched.go:788 msg="new model will fit in available VRAM in single GPU, loading" model=/mnt/di>
5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.990+08:00 level=INFO source=server.go:135 msg="system memory" total="54.9 GiB" free="49.5 GiB" free_swap="4.0 GiB"
5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.991+08:00 level=INFO source=server.go:168 msg=offload library=cuda layers.requested=-1 layers.model=37 layers.offload=37>
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: loaded meta data with 28 key-value pairs and 399 tensors from /mnt/diskb/ollama_models/models/blobs/sha256-a3de86cd1c132c8224>
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   0:                       general.architecture str              = qwen3
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   1:                               general.type str              = model
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   2:                               general.name str              = Qwen3 8B
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   3:                           general.basename str              = Qwen3
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   4:                         general.size_label str              = 8B
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   5:                            general.license str              = apache-2.0
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   6:                          qwen3.block_count u32              = 36
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   7:                       qwen3.context_length u32              = 40960
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   8:                     qwen3.embedding_length u32              = 4096
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   9:                  qwen3.feed_forward_length u32              = 12288
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  10:                 qwen3.attention.head_count u32              = 32
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  11:              qwen3.attention.head_count_kv u32              = 8
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  12:                       qwen3.rope.freq_base f32              = 1000000.000000
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  13:     qwen3.attention.layer_norm_rms_epsilon f32              = 0.000001
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  14:                 qwen3.attention.key_length u32              = 128
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  15:               qwen3.attention.value_length u32              = 128
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  16:                       tokenizer.ggml.model str              = gpt2
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  17:                         tokenizer.ggml.pre str              = qwen2
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  18:                      tokenizer.ggml.tokens arr[str,151936]  = ["!", "\"", "#", "$", "%", "&", "'", ...
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  19:                  tokenizer.ggml.token_type arr[i32,151936]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  20:                      tokenizer.ggml.merges arr[str,151387]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  21:                tokenizer.ggml.eos_token_id u32              = 151645
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  22:            tokenizer.ggml.padding_token_id u32              = 151643
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  23:                tokenizer.ggml.bos_token_id u32              = 151643
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  24:               tokenizer.ggml.add_bos_token bool             = false
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  25:                    tokenizer.chat_template str              = {%- if tools %}\n    {{- '<|im_start|>...
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  26:               general.quantization_version u32              = 2
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  27:                          general.file_type u32              = 15
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type  f32:  145 tensors
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type  f16:   36 tensors
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type q4_K:  199 tensors
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type q6_K:   19 tensors
5月 31 09:30:57 ai ollama[1180]: print_info: file format = GGUF V3 (latest)
5月 31 09:30:57 ai ollama[1180]: print_info: file type   = Q4_K - Medium
5月 31 09:30:57 ai ollama[1180]: print_info: file size   = 4.86 GiB (5.10 BPW)
5月 31 09:30:57 ai ollama[1180]: load: special tokens cache size = 26
5月 31 09:30:57 ai ollama[1180]: load: token to piece cache size = 0.9311 MB
5月 31 09:30:57 ai ollama[1180]: print_info: arch             = qwen3
5月 31 09:30:57 ai ollama[1180]: print_info: vocab_only       = 1
5月 31 09:30:57 ai ollama[1180]: print_info: model type       = ?B
5月 31 09:30:57 ai ollama[1180]: print_info: model params     = 8.19 B
5月 31 09:30:57 ai ollama[1180]: print_info: general.name     = Qwen3 8B
5月 31 09:30:57 ai ollama[1180]: print_info: vocab type       = BPE
5月 31 09:30:57 ai ollama[1180]: print_info: n_vocab          = 151936
5月 31 09:30:57 ai ollama[1180]: print_info: n_merges         = 151387
5月 31 09:30:57 ai ollama[1180]: print_info: BOS token        = 151643 '<|endoftext|>'
5月 31 09:30:57 ai ollama[1180]: print_info: EOS token        = 151645 '<|im_end|>'
5月 31 09:30:57 ai ollama[1180]: print_info: EOT token        = 151645 '<|im_end|>'
5月 31 09:30:57 ai ollama[1180]: print_info: PAD token        = 151643 '<|endoftext|>'
5月 31 09:30:57 ai ollama[1180]: print_info: LF token         = 198 'Ċ'
5月 31 09:30:57 ai ollama[1180]: print_info: FIM PRE token    = 151659 '<|fim_prefix|>'
5月 31 09:30:57 ai ollama[1180]: print_info: FIM SUF token    = 151661 '<|fim_suffix|>'
5月 31 09:30:57 ai ollama[1180]: print_info: FIM MID token    = 151660 '<|fim_middle|>'
5月 31 09:30:57 ai ollama[1180]: print_info: FIM PAD token    = 151662 '<|fim_pad|>'
5月 31 09:30:57 ai ollama[1180]: print_info: FIM REP token    = 151663 '<|repo_name|>'
5月 31 09:30:57 ai ollama[1180]: print_info: FIM SEP token    = 151664 '<|file_sep|>'
5月 31 09:30:57 ai ollama[1180]: print_info: EOG token        = 151643 '<|endoftext|>'
5月 31 09:30:57 ai ollama[1180]: print_info: EOG token        = 151645 '<|im_end|>'
5月 31 09:30:57 ai ollama[1180]: print_info: EOG token        = 151662 '<|fim_pad|>'
5月 31 09:30:57 ai ollama[1180]: print_info: EOG token        = 151663 '<|repo_name|>'
5月 31 09:30:57 ai ollama[1180]: print_info: EOG token        = 151664 '<|file_sep|>'
5月 31 09:30:57 ai ollama[1180]: print_info: max token length = 256
5月 31 09:30:57 ai ollama[1180]: llama_model_load: vocab only - skipping tensors
5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.548+08:00 level=INFO source=server.go:431 msg="starting llama server" cmd="/usr/local/bin/ollama runner --model /mnt/dis>
5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.549+08:00 level=INFO source=sched.go:483 msg="loaded runners" count=2
5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.549+08:00 level=INFO source=server.go:591 msg="waiting for llama runner to start responding"
5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.550+08:00 level=INFO source=server.go:625 msg="waiting for server to become available" status="llm server not responding"
5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.596+08:00 level=INFO source=runner.go:815 msg="starting go runner"
5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.597+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc)
5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.604+08:00 level=INFO source=runner.go:874 msg="Server listening on 127.0.0.1:34567"
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: loaded meta data with 28 key-value pairs and 399 tensors from /mnt/diskb/ollama_models/models/blobs/sha256-a3de86cd1c132c8224>
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   0:                       general.architecture str              = qwen3
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   1:                               general.type str              = model
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   2:                               general.name str              = Qwen3 8B
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   3:                           general.basename str              = Qwen3
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   4:                         general.size_label str              = 8B
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   5:                            general.license str              = apache-2.0
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   6:                          qwen3.block_count u32              = 36
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   7:                       qwen3.context_length u32              = 40960
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   8:                     qwen3.embedding_length u32              = 4096
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv   9:                  qwen3.feed_forward_length u32              = 12288
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  10:                 qwen3.attention.head_count u32              = 32
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  11:              qwen3.attention.head_count_kv u32              = 8
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  12:                       qwen3.rope.freq_base f32              = 1000000.000000
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  13:     qwen3.attention.layer_norm_rms_epsilon f32              = 0.000001
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  14:                 qwen3.attention.key_length u32              = 128
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  15:               qwen3.attention.value_length u32              = 128
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  16:                       tokenizer.ggml.model str              = gpt2
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  17:                         tokenizer.ggml.pre str              = qwen2
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  18:                      tokenizer.ggml.tokens arr[str,151936]  = ["!", "\"", "#", "$", "%", "&", "'", ...
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  19:                  tokenizer.ggml.token_type arr[i32,151936]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  20:                      tokenizer.ggml.merges arr[str,151387]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  21:                tokenizer.ggml.eos_token_id u32              = 151645
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  22:            tokenizer.ggml.padding_token_id u32              = 151643
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  23:                tokenizer.ggml.bos_token_id u32              = 151643
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  24:               tokenizer.ggml.add_bos_token bool             = false
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  25:                    tokenizer.chat_template str              = {%- if tools %}\n    {{- '<|im_start|>...
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  26:               general.quantization_version u32              = 2
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv  27:                          general.file_type u32              = 15
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type  f32:  145 tensors
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type  f16:   36 tensors
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type q4_K:  199 tensors
5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type q6_K:   19 tensors
5月 31 09:30:57 ai ollama[1180]: print_info: file format = GGUF V3 (latest)
5月 31 09:30:57 ai ollama[1180]: print_info: file type   = Q4_K - Medium
5月 31 09:30:57 ai ollama[1180]: print_info: file size   = 4.86 GiB (5.10 BPW)
5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.802+08:00 level=INFO source=server.go:625 msg="waiting for server to become available" status="llm server loading model"
5月 31 09:30:57 ai ollama[1180]: load: special tokens cache size = 26
5月 31 09:30:58 ai ollama[1180]: load: token to piece cache size = 0.9311 MB
5月 31 09:30:58 ai ollama[1180]: print_info: arch             = qwen3
5月 31 09:30:58 ai ollama[1180]: print_info: vocab_only       = 0
5月 31 09:30:58 ai ollama[1180]: print_info: n_ctx_train      = 40960
5月 31 09:30:58 ai ollama[1180]: print_info: n_embd           = 4096
5月 31 09:30:58 ai ollama[1180]: print_info: n_layer          = 36
5月 31 09:30:58 ai ollama[1180]: print_info: n_head           = 32
5月 31 09:30:58 ai ollama[1180]: print_info: n_head_kv        = 8
5月 31 09:30:58 ai ollama[1180]: print_info: n_rot            = 128
5月 31 09:30:58 ai ollama[1180]: print_info: n_swa            = 0
5月 31 09:30:58 ai ollama[1180]: print_info: n_swa_pattern    = 1
5月 31 09:30:58 ai ollama[1180]: print_info: n_embd_head_k    = 128
5月 31 09:30:58 ai ollama[1180]: print_info: n_embd_head_v    = 128
5月 31 09:30:58 ai ollama[1180]: print_info: n_gqa            = 4
5月 31 09:30:58 ai ollama[1180]: print_info: n_embd_k_gqa     = 1024
5月 31 09:30:58 ai ollama[1180]: print_info: n_embd_v_gqa     = 1024
5月 31 09:30:58 ai ollama[1180]: print_info: f_norm_eps       = 0.0e+00
5月 31 09:30:58 ai ollama[1180]: print_info: f_norm_rms_eps   = 1.0e-06
5月 31 09:30:58 ai ollama[1180]: print_info: f_clamp_kqv      = 0.0e+00
5月 31 09:30:58 ai ollama[1180]: print_info: f_max_alibi_bias = 0.0e+00
5月 31 09:30:58 ai ollama[1180]: print_info: f_logit_scale    = 0.0e+00
5月 31 09:30:58 ai ollama[1180]: print_info: f_attn_scale     = 0.0e+00
5月 31 09:30:58 ai ollama[1180]: print_info: n_ff             = 12288
5月 31 09:30:58 ai ollama[1180]: print_info: n_expert         = 0
5月 31 09:30:58 ai ollama[1180]: print_info: n_expert_used    = 0
5月 31 09:30:58 ai ollama[1180]: print_info: causal attn      = 1
5月 31 09:30:58 ai ollama[1180]: print_info: pooling type     = 0
5月 31 09:30:58 ai ollama[1180]: print_info: rope type        = 2
5月 31 09:30:58 ai ollama[1180]: print_info: rope scaling     = linear
5月 31 09:30:58 ai ollama[1180]: print_info: freq_base_train  = 1000000.0
5月 31 09:30:58 ai ollama[1180]: print_info: freq_scale_train = 1
5月 31 09:30:58 ai ollama[1180]: print_info: n_ctx_orig_yarn  = 40960
5月 31 09:30:58 ai ollama[1180]: print_info: rope_finetuned   = unknown
5月 31 09:30:58 ai ollama[1180]: print_info: ssm_d_conv       = 0
5月 31 09:30:58 ai ollama[1180]: print_info: ssm_d_inner      = 0
5月 31 09:30:58 ai ollama[1180]: print_info: ssm_d_state      = 0
5月 31 09:30:58 ai ollama[1180]: print_info: ssm_dt_rank      = 0
5月 31 09:30:58 ai ollama[1180]: print_info: ssm_dt_b_c_rms   = 0
5月 31 09:30:58 ai ollama[1180]: print_info: model type       = 8B
5月 31 09:30:58 ai ollama[1180]: print_info: model params     = 8.19 B
5月 31 09:30:58 ai ollama[1180]: print_info: general.name     = Qwen3 8B
5月 31 09:30:58 ai ollama[1180]: print_info: vocab type       = BPE
5月 31 09:30:58 ai ollama[1180]: print_info: n_vocab          = 151936
5月 31 09:30:58 ai ollama[1180]: print_info: n_merges         = 151387
5月 31 09:30:58 ai ollama[1180]: print_info: BOS token        = 151643 '<|endoftext|>'
5月 31 09:30:58 ai ollama[1180]: print_info: EOS token        = 151645 '<|im_end|>'
5月 31 09:30:58 ai ollama[1180]: print_info: EOT token        = 151645 '<|im_end|>'
5月 31 09:30:58 ai ollama[1180]: print_info: PAD token        = 151643 '<|endoftext|>'
5月 31 09:30:58 ai ollama[1180]: print_info: LF token         = 198 'Ċ'
5月 31 09:30:58 ai ollama[1180]: print_info: FIM PRE token    = 151659 '<|fim_prefix|>'
5月 31 09:30:58 ai ollama[1180]: print_info: FIM SUF token    = 151661 '<|fim_suffix|>'
5月 31 09:30:58 ai ollama[1180]: print_info: FIM MID token    = 151660 '<|fim_middle|>'
5月 31 09:30:58 ai ollama[1180]: print_info: FIM PAD token    = 151662 '<|fim_pad|>'
5月 31 09:30:58 ai ollama[1180]: print_info: FIM REP token    = 151663 '<|repo_name|>'
5月 31 09:30:58 ai ollama[1180]: print_info: FIM SEP token    = 151664 '<|file_sep|>'
5月 31 09:30:58 ai ollama[1180]: print_info: EOG token        = 151643 '<|endoftext|>'
5月 31 09:30:58 ai ollama[1180]: print_info: EOG token        = 151645 '<|im_end|>'
5月 31 09:30:58 ai ollama[1180]: print_info: EOG token        = 151662 '<|fim_pad|>'
5月 31 09:30:58 ai ollama[1180]: print_info: EOG token        = 151663 '<|repo_name|>'
5月 31 09:30:58 ai ollama[1180]: print_info: EOG token        = 151664 '<|file_sep|>'
5月 31 09:30:58 ai ollama[1180]: print_info: max token length = 256
5月 31 09:30:58 ai ollama[1180]: load_tensors: loading model tensors, this can take a while... (mmap = true)
5月 31 09:31:04 ai ollama[1180]: [GIN] 2025/05/31 - 09:31:04 | 200 |     397.047µs |       127.0.0.1 | HEAD     "/"
5月 31 09:31:04 ai ollama[1180]: [GIN] 2025/05/31 - 09:31:04 | 200 |     477.301µs |       127.0.0.1 | GET      "/api/ps"
5月 31 09:31:56 ai ollama[1180]: [GIN] 2025/05/31 - 09:31:56 | 200 |     350.568µs |       127.0.0.1 | HEAD     "/"
5月 31 09:31:56 ai ollama[1180]: [GIN] 2025/05/31 - 09:31:56 | 200 |     690.382µs |       127.0.0.1 | GET      "/api/ps"
5月 31 09:32:25 ai ollama[1180]: load_tensors:   CPU_Mapped model buffer size =  4977.62 MiB
5月 31 09:32:25 ai ollama[1180]: llama_context: constructing llama_context
5月 31 09:32:25 ai ollama[1180]: llama_context: n_seq_max     = 2
5月 31 09:32:25 ai ollama[1180]: llama_context: n_ctx         = 8192
5月 31 09:32:25 ai ollama[1180]: llama_context: n_ctx_per_seq = 4096
5月 31 09:32:25 ai ollama[1180]: llama_context: n_batch       = 1024
 ~ ▓▒░

OS

No response

GPU

No response

CPU

No response

Ollama version

No response

Originally created by @wlsoft2006 on GitHub (May 31, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/10923 ### What is the issue? <img width="713" alt="Image" src="https://github.com/user-attachments/assets/07df8d11-abc5-4130-ade4-68ca83630326" /> <img width="844" alt="Image" src="https://github.com/user-attachments/assets/006b03e2-6280-40a2-abb7-ad26add63db9" /> ### Relevant log output ```shell 5月 30 22:16:51 ai systemd[1]: Started Ollama Service. 5月 30 22:17:14 ai ollama[1180]: time=2025-05-30T22:17:13.277+08:00 level=INFO source=routes.go:1234 msg="server config" env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VI> 5月 30 22:17:15 ai ollama[1180]: time=2025-05-30T22:17:15.272+08:00 level=INFO source=images.go:479 msg="total blobs: 62" 5月 30 22:17:15 ai ollama[1180]: time=2025-05-30T22:17:15.275+08:00 level=INFO source=images.go:486 msg="total unused blobs removed: 0" 5月 30 22:17:15 ai ollama[1180]: time=2025-05-30T22:17:15.276+08:00 level=INFO source=routes.go:1287 msg="Listening on [::]:11434 (version 0.9.0)" 5月 30 22:17:15 ai ollama[1180]: time=2025-05-30T22:17:15.384+08:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs" 5月 30 22:17:28 ai ollama[1180]: time=2025-05-30T22:17:28.371+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-9ef6fcbc-ad1a-6752-6985-09e1e601e662 library=cud> 5月 30 22:17:28 ai ollama[1180]: time=2025-05-30T22:17:28.372+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-083bdbe0-7f5c-38d0-aa01-2635abd8764e library=cud> 5月 31 08:44:56 ai ollama[1180]: [GIN] 2025/05/31 - 08:44:56 | 200 | 26.883628ms | 127.0.0.1 | HEAD "/" 5月 31 08:44:56 ai ollama[1180]: [GIN] 2025/05/31 - 08:44:56 | 200 | 22.248672ms | 127.0.0.1 | GET "/api/ps" 5月 31 08:45:00 ai ollama[1180]: [GIN] 2025/05/31 - 08:45:00 | 200 | 68.544µs | 127.0.0.1 | HEAD "/" 5月 31 08:45:01 ai ollama[1180]: [GIN] 2025/05/31 - 08:45:01 | 200 | 620.481972ms | 127.0.0.1 | POST "/api/show" 5月 31 08:45:02 ai ollama[1180]: time=2025-05-31T08:45:02.009+08:00 level=INFO source=sched.go:788 msg="new model will fit in available VRAM in single GPU, loading" model=/mnt/di> 5月 31 08:45:02 ai ollama[1180]: time=2025-05-31T08:45:02.640+08:00 level=INFO source=server.go:135 msg="system memory" total="54.9 GiB" free="50.8 GiB" free_swap="4.0 GiB" 5月 31 08:45:02 ai ollama[1180]: time=2025-05-31T08:45:02.641+08:00 level=INFO source=server.go:168 msg=offload library=cuda layers.requested=-1 layers.model=37 layers.offload=37> 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: loaded meta data with 32 key-value pairs and 399 tensors from /mnt/diskb/ollama_models/models/blobs/sha256-e6a7edc1a4d7d9b2de> 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 0: general.architecture str = qwen3 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 1: general.type str = model 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 2: general.name str = DeepSeek R1 0528 Qwen3 8B 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 3: general.basename str = DeepSeek-R1-0528-Qwen3 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 4: general.size_label str = 8B 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 5: general.license str = mit 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 6: qwen3.block_count u32 = 36 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 7: qwen3.context_length u32 = 131072 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 8: qwen3.embedding_length u32 = 4096 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 9: qwen3.feed_forward_length u32 = 12288 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 10: qwen3.attention.head_count u32 = 32 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 11: qwen3.attention.head_count_kv u32 = 8 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 12: qwen3.rope.freq_base f32 = 1000000.000000 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 13: qwen3.attention.layer_norm_rms_epsilon f32 = 0.000001 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 14: qwen3.attention.key_length u32 = 128 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 15: qwen3.attention.value_length u32 = 128 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 16: qwen3.rope.scaling.type str = yarn 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 17: qwen3.rope.scaling.factor f32 = 4.000000 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 18: qwen3.rope.scaling.original_context_length u32 = 32768 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 19: tokenizer.ggml.model str = gpt2 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 20: tokenizer.ggml.pre str = qwen2 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 21: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 22: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 23: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 24: tokenizer.ggml.bos_token_id u32 = 151643 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 25: tokenizer.ggml.eos_token_id u32 = 151645 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 26: tokenizer.ggml.padding_token_id u32 = 151645 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 27: tokenizer.ggml.add_bos_token bool = false 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 28: tokenizer.ggml.add_eos_token bool = false 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 29: tokenizer.chat_template str = {% if not add_generation_prompt is de... 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 30: general.quantization_version u32 = 2 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - kv 31: general.file_type u32 = 15 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - type f32: 145 tensors 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - type f16: 36 tensors 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - type q4_K: 199 tensors 5月 31 08:45:02 ai ollama[1180]: llama_model_loader: - type q6_K: 19 tensors 5月 31 08:45:02 ai ollama[1180]: print_info: file format = GGUF V3 (latest) 5月 31 08:45:02 ai ollama[1180]: print_info: file type = Q4_K - Medium 5月 31 08:45:02 ai ollama[1180]: print_info: file size = 4.86 GiB (5.10 BPW) 5月 31 08:45:03 ai ollama[1180]: load: special_eos_id is not in special_eog_ids - the tokenizer config may be incorrect 5月 31 08:45:03 ai ollama[1180]: load: special tokens cache size = 28 5月 31 08:45:03 ai ollama[1180]: load: token to piece cache size = 0.9311 MB 5月 31 08:45:03 ai ollama[1180]: print_info: arch = qwen3 5月 31 08:45:03 ai ollama[1180]: print_info: vocab_only = 1 5月 31 08:45:03 ai ollama[1180]: print_info: model type = ?B 5月 31 08:45:03 ai ollama[1180]: print_info: model params = 8.19 B 5月 31 08:45:03 ai ollama[1180]: print_info: general.name = DeepSeek R1 0528 Qwen3 8B 5月 31 08:45:03 ai ollama[1180]: print_info: vocab type = BPE 5月 31 08:45:03 ai ollama[1180]: print_info: n_vocab = 151936 5月 31 08:45:03 ai ollama[1180]: print_info: n_merges = 151387 5月 31 08:45:03 ai ollama[1180]: print_info: BOS token = 151643 '<｜begin▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOS token = 151645 '<｜end▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOT token = 151645 '<｜end▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: PAD token = 151645 '<｜end▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: LF token = 198 'Ċ' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM PRE token = 151659 '<|fim_prefix|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM SUF token = 151661 '<|fim_suffix|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM MID token = 151660 '<|fim_middle|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM PAD token = 151662 '<|fim_pad|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM REP token = 151663 '<|repo_name|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM SEP token = 151664 '<|file_sep|>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOG token = 151645 '<｜end▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOG token = 151662 '<|fim_pad|>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOG token = 151663 '<|repo_name|>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOG token = 151664 '<|file_sep|>' 5月 31 08:45:03 ai ollama[1180]: print_info: max token length = 256 5月 31 08:45:03 ai ollama[1180]: llama_model_load: vocab only - skipping tensors 5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.101+08:00 level=INFO source=server.go:431 msg="starting llama server" cmd="/usr/local/bin/ollama runner --model /mnt/dis> 5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.102+08:00 level=INFO source=sched.go:483 msg="loaded runners" count=1 5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.102+08:00 level=INFO source=server.go:591 msg="waiting for llama runner to start responding" 5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.103+08:00 level=INFO source=server.go:625 msg="waiting for server to become available" status="llm server not responding" 5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.126+08:00 level=INFO source=runner.go:815 msg="starting go runner" 5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.126+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc) 5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.132+08:00 level=INFO source=runner.go:874 msg="Server listening on 127.0.0.1:40057" 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: loaded meta data with 32 key-value pairs and 399 tensors from /mnt/diskb/ollama_models/models/blobs/sha256-e6a7edc1a4d7d9b2de> 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 0: general.architecture str = qwen3 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 1: general.type str = model 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 2: general.name str = DeepSeek R1 0528 Qwen3 8B 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 3: general.basename str = DeepSeek-R1-0528-Qwen3 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 4: general.size_label str = 8B 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 5: general.license str = mit 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 6: qwen3.block_count u32 = 36 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 7: qwen3.context_length u32 = 131072 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 8: qwen3.embedding_length u32 = 4096 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 9: qwen3.feed_forward_length u32 = 12288 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 10: qwen3.attention.head_count u32 = 32 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 11: qwen3.attention.head_count_kv u32 = 8 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 12: qwen3.rope.freq_base f32 = 1000000.000000 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 13: qwen3.attention.layer_norm_rms_epsilon f32 = 0.000001 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 14: qwen3.attention.key_length u32 = 128 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 15: qwen3.attention.value_length u32 = 128 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 16: qwen3.rope.scaling.type str = yarn 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 17: qwen3.rope.scaling.factor f32 = 4.000000 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 18: qwen3.rope.scaling.original_context_length u32 = 32768 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 19: tokenizer.ggml.model str = gpt2 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 20: tokenizer.ggml.pre str = qwen2 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 21: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 22: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 23: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 24: tokenizer.ggml.bos_token_id u32 = 151643 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 25: tokenizer.ggml.eos_token_id u32 = 151645 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 26: tokenizer.ggml.padding_token_id u32 = 151645 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 27: tokenizer.ggml.add_bos_token bool = false 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 28: tokenizer.ggml.add_eos_token bool = false 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 29: tokenizer.chat_template str = {% if not add_generation_prompt is de... 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 30: general.quantization_version u32 = 2 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - kv 31: general.file_type u32 = 15 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - type f32: 145 tensors 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - type f16: 36 tensors 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - type q4_K: 199 tensors 5月 31 08:45:03 ai ollama[1180]: llama_model_loader: - type q6_K: 19 tensors 5月 31 08:45:03 ai ollama[1180]: print_info: file format = GGUF V3 (latest) 5月 31 08:45:03 ai ollama[1180]: print_info: file type = Q4_K - Medium 5月 31 08:45:03 ai ollama[1180]: print_info: file size = 4.86 GiB (5.10 BPW) 5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.356+08:00 level=INFO source=server.go:625 msg="waiting for server to become available" status="llm server loading model" 5月 31 08:45:03 ai ollama[1180]: load: special_eos_id is not in special_eog_ids - the tokenizer config may be incorrect 5月 31 08:45:03 ai ollama[1180]: load: special tokens cache size = 28 5月 31 08:45:03 ai ollama[1180]: load: token to piece cache size = 0.9311 MB 5月 31 08:45:03 ai ollama[1180]: print_info: arch = qwen3 5月 31 08:45:03 ai ollama[1180]: print_info: vocab_only = 0 5月 31 08:45:03 ai ollama[1180]: print_info: n_ctx_train = 131072 5月 31 08:45:03 ai ollama[1180]: print_info: n_embd = 4096 5月 31 08:45:03 ai ollama[1180]: print_info: n_layer = 36 5月 31 08:45:03 ai ollama[1180]: print_info: n_head = 32 5月 31 08:45:03 ai ollama[1180]: print_info: n_head_kv = 8 5月 31 08:45:03 ai ollama[1180]: print_info: n_rot = 128 5月 31 08:45:03 ai ollama[1180]: print_info: n_swa = 0 5月 31 08:45:03 ai ollama[1180]: print_info: n_swa_pattern = 1 5月 31 08:45:03 ai ollama[1180]: print_info: n_embd_head_k = 128 5月 31 08:45:03 ai ollama[1180]: print_info: n_embd_head_v = 128 5月 31 08:45:03 ai ollama[1180]: print_info: n_gqa = 4 5月 31 08:45:03 ai ollama[1180]: print_info: n_embd_k_gqa = 1024 5月 31 08:45:03 ai ollama[1180]: print_info: n_embd_v_gqa = 1024 5月 31 08:45:03 ai ollama[1180]: print_info: f_norm_eps = 0.0e+00 5月 31 08:45:03 ai ollama[1180]: print_info: f_norm_rms_eps = 1.0e-06 5月 31 08:45:03 ai ollama[1180]: print_info: f_clamp_kqv = 0.0e+00 5月 31 08:45:03 ai ollama[1180]: print_info: f_max_alibi_bias = 0.0e+00 5月 31 08:45:03 ai ollama[1180]: print_info: f_logit_scale = 0.0e+00 5月 31 08:45:03 ai ollama[1180]: print_info: f_attn_scale = 0.0e+00 5月 31 08:45:03 ai ollama[1180]: print_info: n_ff = 12288 5月 31 08:45:03 ai ollama[1180]: print_info: n_expert = 0 5月 31 08:45:03 ai ollama[1180]: print_info: n_expert_used = 0 5月 31 08:45:03 ai ollama[1180]: print_info: causal attn = 1 5月 31 08:45:03 ai ollama[1180]: print_info: pooling type = 0 5月 31 08:45:03 ai ollama[1180]: print_info: rope type = 2 5月 31 08:45:03 ai ollama[1180]: print_info: rope scaling = yarn 5月 31 08:45:03 ai ollama[1180]: print_info: freq_base_train = 1000000.0 5月 31 08:45:03 ai ollama[1180]: print_info: freq_scale_train = 0.25 5月 31 08:45:03 ai ollama[1180]: print_info: n_ctx_orig_yarn = 32768 5月 31 08:45:03 ai ollama[1180]: print_info: rope_finetuned = unknown 5月 31 08:45:03 ai ollama[1180]: print_info: ssm_d_conv = 0 5月 31 08:45:03 ai ollama[1180]: print_info: ssm_d_inner = 0 5月 31 08:45:03 ai ollama[1180]: print_info: ssm_d_state = 0 5月 31 08:45:03 ai ollama[1180]: print_info: ssm_dt_rank = 0 5月 31 08:45:03 ai ollama[1180]: print_info: ssm_dt_b_c_rms = 0 5月 31 08:45:03 ai ollama[1180]: print_info: model type = 8B 5月 31 08:45:03 ai ollama[1180]: print_info: model params = 8.19 B 5月 31 08:45:03 ai ollama[1180]: print_info: general.name = DeepSeek R1 0528 Qwen3 8B 5月 31 08:45:03 ai ollama[1180]: print_info: vocab type = BPE 5月 31 08:45:03 ai ollama[1180]: print_info: n_vocab = 151936 5月 31 08:45:03 ai ollama[1180]: print_info: n_merges = 151387 5月 31 08:45:03 ai ollama[1180]: print_info: BOS token = 151643 '<｜begin▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOS token = 151645 '<｜end▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOT token = 151645 '<｜end▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: PAD token = 151645 '<｜end▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: LF token = 198 'Ċ' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM PRE token = 151659 '<|fim_prefix|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM SUF token = 151661 '<|fim_suffix|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM MID token = 151660 '<|fim_middle|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM PAD token = 151662 '<|fim_pad|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM REP token = 151663 '<|repo_name|>' 5月 31 08:45:03 ai ollama[1180]: print_info: FIM SEP token = 151664 '<|file_sep|>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOG token = 151645 '<｜end▁of▁sentence｜>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOG token = 151662 '<|fim_pad|>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOG token = 151663 '<|repo_name|>' 5月 31 08:45:03 ai ollama[1180]: print_info: EOG token = 151664 '<|file_sep|>' 5月 31 08:45:03 ai ollama[1180]: print_info: max token length = 256 5月 31 08:45:03 ai ollama[1180]: load_tensors: loading model tensors, this can take a while... (mmap = true) 5月 31 08:46:30 ai ollama[1180]: load_tensors: CPU_Mapped model buffer size = 4977.62 MiB 5月 31 08:46:30 ai ollama[1180]: llama_context: constructing llama_context 5月 31 08:46:30 ai ollama[1180]: llama_context: n_seq_max = 2 5月 31 08:46:30 ai ollama[1180]: llama_context: n_ctx = 8192 5月 31 08:46:30 ai ollama[1180]: llama_context: n_ctx_per_seq = 4096 5月 31 08:46:30 ai ollama[1180]: llama_context: n_batch = 1024 5月 31 08:46:30 ai ollama[1180]: llama_context: n_ubatch = 512 5月 31 08:46:30 ai ollama[1180]: llama_context: causal_attn = 1 5月 31 08:46:30 ai ollama[1180]: llama_context: flash_attn = 0 5月 31 08:46:30 ai ollama[1180]: llama_context: freq_base = 1000000.0 5月 31 08:46:30 ai ollama[1180]: llama_context: freq_scale = 0.25 5月 31 08:46:30 ai ollama[1180]: llama_context: n_ctx_per_seq (4096) < n_ctx_train (131072) -- the full capacity of the model will not be utilized 5月 31 08:46:30 ai ollama[1180]: llama_context: CPU output buffer size = 1.19 MiB 5月 31 08:46:30 ai ollama[1180]: llama_kv_cache_unified: kv_size = 8192, type_k = 'f16', type_v = 'f16', n_layer = 36, can_shift = 1, padding = 32 5月 31 08:46:32 ai ollama[1180]: llama_kv_cache_unified: CPU KV buffer size = 1152.00 MiB 5月 31 08:46:32 ai ollama[1180]: llama_kv_cache_unified: KV self size = 1152.00 MiB, K (f16): 576.00 MiB, V (f16): 576.00 MiB 5月 31 08:46:32 ai ollama[1180]: llama_context: CPU compute buffer size = 560.01 MiB 5月 31 08:46:32 ai ollama[1180]: llama_context: graph nodes = 1374 5月 31 08:46:32 ai ollama[1180]: llama_context: graph splits = 1 5月 31 08:46:33 ai ollama[1180]: time=2025-05-31T08:46:33.014+08:00 level=INFO source=server.go:630 msg="llama runner started in 89.91 seconds" 5月 31 08:46:33 ai ollama[1180]: [GIN] 2025/05/31 - 08:46:33 | 200 | 1m31s | 127.0.0.1 | POST "/api/generate" 5月 31 08:48:41 ai ollama[1180]: [GIN] 2025/05/31 - 08:48:41 | 200 | 1m55s | 127.0.0.1 | POST "/api/chat" 5月 31 09:30:29 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:29 | 200 | 405.308µs | 127.0.0.1 | HEAD "/" 5月 31 09:30:29 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:29 | 200 | 90.556909ms | 127.0.0.1 | POST "/api/show" 5月 31 09:30:29 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:29 | 200 | 49.681837ms | 127.0.0.1 | POST "/api/generate" 5月 31 09:30:41 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:41 | 200 | 83.707µs | 127.0.0.1 | HEAD "/" 5月 31 09:30:42 ai ollama[1180]: time=2025-05-31T09:30:42.206+08:00 level=WARN source=routes.go:920 msg="bad manifest filepath" name=registry.ollama.ai/library/gemma3:latest erro> 5月 31 09:30:42 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:42 | 200 | 326.388416ms | 127.0.0.1 | GET "/api/tags" 5月 31 09:30:55 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:55 | 200 | 39.07µs | 127.0.0.1 | HEAD "/" 5月 31 09:30:55 ai ollama[1180]: [GIN] 2025/05/31 - 09:30:55 | 200 | 437.917824ms | 127.0.0.1 | POST "/api/show" 5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.436+08:00 level=INFO source=sched.go:548 msg="updated VRAM based on existing loaded models" gpu=GPU-9ef6fcbc-ad1a-6752-6> 5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.436+08:00 level=INFO source=sched.go:548 msg="updated VRAM based on existing loaded models" gpu=GPU-083bdbe0-7f5c-38d0-a> 5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.437+08:00 level=INFO source=sched.go:788 msg="new model will fit in available VRAM in single GPU, loading" model=/mnt/di> 5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.990+08:00 level=INFO source=server.go:135 msg="system memory" total="54.9 GiB" free="49.5 GiB" free_swap="4.0 GiB" 5月 31 09:30:56 ai ollama[1180]: time=2025-05-31T09:30:56.991+08:00 level=INFO source=server.go:168 msg=offload library=cuda layers.requested=-1 layers.model=37 layers.offload=37> 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: loaded meta data with 28 key-value pairs and 399 tensors from /mnt/diskb/ollama_models/models/blobs/sha256-a3de86cd1c132c8224> 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 0: general.architecture str = qwen3 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 1: general.type str = model 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 2: general.name str = Qwen3 8B 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 3: general.basename str = Qwen3 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 4: general.size_label str = 8B 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 5: general.license str = apache-2.0 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 6: qwen3.block_count u32 = 36 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 7: qwen3.context_length u32 = 40960 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 8: qwen3.embedding_length u32 = 4096 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 9: qwen3.feed_forward_length u32 = 12288 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 10: qwen3.attention.head_count u32 = 32 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 11: qwen3.attention.head_count_kv u32 = 8 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 12: qwen3.rope.freq_base f32 = 1000000.000000 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 13: qwen3.attention.layer_norm_rms_epsilon f32 = 0.000001 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 14: qwen3.attention.key_length u32 = 128 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 15: qwen3.attention.value_length u32 = 128 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 16: tokenizer.ggml.model str = gpt2 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 17: tokenizer.ggml.pre str = qwen2 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 18: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 19: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 20: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 21: tokenizer.ggml.eos_token_id u32 = 151645 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 22: tokenizer.ggml.padding_token_id u32 = 151643 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 23: tokenizer.ggml.bos_token_id u32 = 151643 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 24: tokenizer.ggml.add_bos_token bool = false 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 25: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>... 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 26: general.quantization_version u32 = 2 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 27: general.file_type u32 = 15 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type f32: 145 tensors 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type f16: 36 tensors 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type q4_K: 199 tensors 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type q6_K: 19 tensors 5月 31 09:30:57 ai ollama[1180]: print_info: file format = GGUF V3 (latest) 5月 31 09:30:57 ai ollama[1180]: print_info: file type = Q4_K - Medium 5月 31 09:30:57 ai ollama[1180]: print_info: file size = 4.86 GiB (5.10 BPW) 5月 31 09:30:57 ai ollama[1180]: load: special tokens cache size = 26 5月 31 09:30:57 ai ollama[1180]: load: token to piece cache size = 0.9311 MB 5月 31 09:30:57 ai ollama[1180]: print_info: arch = qwen3 5月 31 09:30:57 ai ollama[1180]: print_info: vocab_only = 1 5月 31 09:30:57 ai ollama[1180]: print_info: model type = ?B 5月 31 09:30:57 ai ollama[1180]: print_info: model params = 8.19 B 5月 31 09:30:57 ai ollama[1180]: print_info: general.name = Qwen3 8B 5月 31 09:30:57 ai ollama[1180]: print_info: vocab type = BPE 5月 31 09:30:57 ai ollama[1180]: print_info: n_vocab = 151936 5月 31 09:30:57 ai ollama[1180]: print_info: n_merges = 151387 5月 31 09:30:57 ai ollama[1180]: print_info: BOS token = 151643 '<|endoftext|>' 5月 31 09:30:57 ai ollama[1180]: print_info: EOS token = 151645 '<|im_end|>' 5月 31 09:30:57 ai ollama[1180]: print_info: EOT token = 151645 '<|im_end|>' 5月 31 09:30:57 ai ollama[1180]: print_info: PAD token = 151643 '<|endoftext|>' 5月 31 09:30:57 ai ollama[1180]: print_info: LF token = 198 'Ċ' 5月 31 09:30:57 ai ollama[1180]: print_info: FIM PRE token = 151659 '<|fim_prefix|>' 5月 31 09:30:57 ai ollama[1180]: print_info: FIM SUF token = 151661 '<|fim_suffix|>' 5月 31 09:30:57 ai ollama[1180]: print_info: FIM MID token = 151660 '<|fim_middle|>' 5月 31 09:30:57 ai ollama[1180]: print_info: FIM PAD token = 151662 '<|fim_pad|>' 5月 31 09:30:57 ai ollama[1180]: print_info: FIM REP token = 151663 '<|repo_name|>' 5月 31 09:30:57 ai ollama[1180]: print_info: FIM SEP token = 151664 '<|file_sep|>' 5月 31 09:30:57 ai ollama[1180]: print_info: EOG token = 151643 '<|endoftext|>' 5月 31 09:30:57 ai ollama[1180]: print_info: EOG token = 151645 '<|im_end|>' 5月 31 09:30:57 ai ollama[1180]: print_info: EOG token = 151662 '<|fim_pad|>' 5月 31 09:30:57 ai ollama[1180]: print_info: EOG token = 151663 '<|repo_name|>' 5月 31 09:30:57 ai ollama[1180]: print_info: EOG token = 151664 '<|file_sep|>' 5月 31 09:30:57 ai ollama[1180]: print_info: max token length = 256 5月 31 09:30:57 ai ollama[1180]: llama_model_load: vocab only - skipping tensors 5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.548+08:00 level=INFO source=server.go:431 msg="starting llama server" cmd="/usr/local/bin/ollama runner --model /mnt/dis> 5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.549+08:00 level=INFO source=sched.go:483 msg="loaded runners" count=2 5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.549+08:00 level=INFO source=server.go:591 msg="waiting for llama runner to start responding" 5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.550+08:00 level=INFO source=server.go:625 msg="waiting for server to become available" status="llm server not responding" 5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.596+08:00 level=INFO source=runner.go:815 msg="starting go runner" 5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.597+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc) 5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.604+08:00 level=INFO source=runner.go:874 msg="Server listening on 127.0.0.1:34567" 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: loaded meta data with 28 key-value pairs and 399 tensors from /mnt/diskb/ollama_models/models/blobs/sha256-a3de86cd1c132c8224> 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 0: general.architecture str = qwen3 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 1: general.type str = model 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 2: general.name str = Qwen3 8B 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 3: general.basename str = Qwen3 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 4: general.size_label str = 8B 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 5: general.license str = apache-2.0 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 6: qwen3.block_count u32 = 36 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 7: qwen3.context_length u32 = 40960 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 8: qwen3.embedding_length u32 = 4096 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 9: qwen3.feed_forward_length u32 = 12288 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 10: qwen3.attention.head_count u32 = 32 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 11: qwen3.attention.head_count_kv u32 = 8 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 12: qwen3.rope.freq_base f32 = 1000000.000000 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 13: qwen3.attention.layer_norm_rms_epsilon f32 = 0.000001 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 14: qwen3.attention.key_length u32 = 128 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 15: qwen3.attention.value_length u32 = 128 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 16: tokenizer.ggml.model str = gpt2 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 17: tokenizer.ggml.pre str = qwen2 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 18: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 19: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 20: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 21: tokenizer.ggml.eos_token_id u32 = 151645 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 22: tokenizer.ggml.padding_token_id u32 = 151643 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 23: tokenizer.ggml.bos_token_id u32 = 151643 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 24: tokenizer.ggml.add_bos_token bool = false 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 25: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>... 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 26: general.quantization_version u32 = 2 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - kv 27: general.file_type u32 = 15 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type f32: 145 tensors 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type f16: 36 tensors 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type q4_K: 199 tensors 5月 31 09:30:57 ai ollama[1180]: llama_model_loader: - type q6_K: 19 tensors 5月 31 09:30:57 ai ollama[1180]: print_info: file format = GGUF V3 (latest) 5月 31 09:30:57 ai ollama[1180]: print_info: file type = Q4_K - Medium 5月 31 09:30:57 ai ollama[1180]: print_info: file size = 4.86 GiB (5.10 BPW) 5月 31 09:30:57 ai ollama[1180]: time=2025-05-31T09:30:57.802+08:00 level=INFO source=server.go:625 msg="waiting for server to become available" status="llm server loading model" 5月 31 09:30:57 ai ollama[1180]: load: special tokens cache size = 26 5月 31 09:30:58 ai ollama[1180]: load: token to piece cache size = 0.9311 MB 5月 31 09:30:58 ai ollama[1180]: print_info: arch = qwen3 5月 31 09:30:58 ai ollama[1180]: print_info: vocab_only = 0 5月 31 09:30:58 ai ollama[1180]: print_info: n_ctx_train = 40960 5月 31 09:30:58 ai ollama[1180]: print_info: n_embd = 4096 5月 31 09:30:58 ai ollama[1180]: print_info: n_layer = 36 5月 31 09:30:58 ai ollama[1180]: print_info: n_head = 32 5月 31 09:30:58 ai ollama[1180]: print_info: n_head_kv = 8 5月 31 09:30:58 ai ollama[1180]: print_info: n_rot = 128 5月 31 09:30:58 ai ollama[1180]: print_info: n_swa = 0 5月 31 09:30:58 ai ollama[1180]: print_info: n_swa_pattern = 1 5月 31 09:30:58 ai ollama[1180]: print_info: n_embd_head_k = 128 5月 31 09:30:58 ai ollama[1180]: print_info: n_embd_head_v = 128 5月 31 09:30:58 ai ollama[1180]: print_info: n_gqa = 4 5月 31 09:30:58 ai ollama[1180]: print_info: n_embd_k_gqa = 1024 5月 31 09:30:58 ai ollama[1180]: print_info: n_embd_v_gqa = 1024 5月 31 09:30:58 ai ollama[1180]: print_info: f_norm_eps = 0.0e+00 5月 31 09:30:58 ai ollama[1180]: print_info: f_norm_rms_eps = 1.0e-06 5月 31 09:30:58 ai ollama[1180]: print_info: f_clamp_kqv = 0.0e+00 5月 31 09:30:58 ai ollama[1180]: print_info: f_max_alibi_bias = 0.0e+00 5月 31 09:30:58 ai ollama[1180]: print_info: f_logit_scale = 0.0e+00 5月 31 09:30:58 ai ollama[1180]: print_info: f_attn_scale = 0.0e+00 5月 31 09:30:58 ai ollama[1180]: print_info: n_ff = 12288 5月 31 09:30:58 ai ollama[1180]: print_info: n_expert = 0 5月 31 09:30:58 ai ollama[1180]: print_info: n_expert_used = 0 5月 31 09:30:58 ai ollama[1180]: print_info: causal attn = 1 5月 31 09:30:58 ai ollama[1180]: print_info: pooling type = 0 5月 31 09:30:58 ai ollama[1180]: print_info: rope type = 2 5月 31 09:30:58 ai ollama[1180]: print_info: rope scaling = linear 5月 31 09:30:58 ai ollama[1180]: print_info: freq_base_train = 1000000.0 5月 31 09:30:58 ai ollama[1180]: print_info: freq_scale_train = 1 5月 31 09:30:58 ai ollama[1180]: print_info: n_ctx_orig_yarn = 40960 5月 31 09:30:58 ai ollama[1180]: print_info: rope_finetuned = unknown 5月 31 09:30:58 ai ollama[1180]: print_info: ssm_d_conv = 0 5月 31 09:30:58 ai ollama[1180]: print_info: ssm_d_inner = 0 5月 31 09:30:58 ai ollama[1180]: print_info: ssm_d_state = 0 5月 31 09:30:58 ai ollama[1180]: print_info: ssm_dt_rank = 0 5月 31 09:30:58 ai ollama[1180]: print_info: ssm_dt_b_c_rms = 0 5月 31 09:30:58 ai ollama[1180]: print_info: model type = 8B 5月 31 09:30:58 ai ollama[1180]: print_info: model params = 8.19 B 5月 31 09:30:58 ai ollama[1180]: print_info: general.name = Qwen3 8B 5月 31 09:30:58 ai ollama[1180]: print_info: vocab type = BPE 5月 31 09:30:58 ai ollama[1180]: print_info: n_vocab = 151936 5月 31 09:30:58 ai ollama[1180]: print_info: n_merges = 151387 5月 31 09:30:58 ai ollama[1180]: print_info: BOS token = 151643 '<|endoftext|>' 5月 31 09:30:58 ai ollama[1180]: print_info: EOS token = 151645 '<|im_end|>' 5月 31 09:30:58 ai ollama[1180]: print_info: EOT token = 151645 '<|im_end|>' 5月 31 09:30:58 ai ollama[1180]: print_info: PAD token = 151643 '<|endoftext|>' 5月 31 09:30:58 ai ollama[1180]: print_info: LF token = 198 'Ċ' 5月 31 09:30:58 ai ollama[1180]: print_info: FIM PRE token = 151659 '<|fim_prefix|>' 5月 31 09:30:58 ai ollama[1180]: print_info: FIM SUF token = 151661 '<|fim_suffix|>' 5月 31 09:30:58 ai ollama[1180]: print_info: FIM MID token = 151660 '<|fim_middle|>' 5月 31 09:30:58 ai ollama[1180]: print_info: FIM PAD token = 151662 '<|fim_pad|>' 5月 31 09:30:58 ai ollama[1180]: print_info: FIM REP token = 151663 '<|repo_name|>' 5月 31 09:30:58 ai ollama[1180]: print_info: FIM SEP token = 151664 '<|file_sep|>' 5月 31 09:30:58 ai ollama[1180]: print_info: EOG token = 151643 '<|endoftext|>' 5月 31 09:30:58 ai ollama[1180]: print_info: EOG token = 151645 '<|im_end|>' 5月 31 09:30:58 ai ollama[1180]: print_info: EOG token = 151662 '<|fim_pad|>' 5月 31 09:30:58 ai ollama[1180]: print_info: EOG token = 151663 '<|repo_name|>' 5月 31 09:30:58 ai ollama[1180]: print_info: EOG token = 151664 '<|file_sep|>' 5月 31 09:30:58 ai ollama[1180]: print_info: max token length = 256 5月 31 09:30:58 ai ollama[1180]: load_tensors: loading model tensors, this can take a while... (mmap = true) 5月 31 09:31:04 ai ollama[1180]: [GIN] 2025/05/31 - 09:31:04 | 200 | 397.047µs | 127.0.0.1 | HEAD "/" 5月 31 09:31:04 ai ollama[1180]: [GIN] 2025/05/31 - 09:31:04 | 200 | 477.301µs | 127.0.0.1 | GET "/api/ps" 5月 31 09:31:56 ai ollama[1180]: [GIN] 2025/05/31 - 09:31:56 | 200 | 350.568µs | 127.0.0.1 | HEAD "/" 5月 31 09:31:56 ai ollama[1180]: [GIN] 2025/05/31 - 09:31:56 | 200 | 690.382µs | 127.0.0.1 | GET "/api/ps" 5月 31 09:32:25 ai ollama[1180]: load_tensors: CPU_Mapped model buffer size = 4977.62 MiB 5月 31 09:32:25 ai ollama[1180]: llama_context: constructing llama_context 5月 31 09:32:25 ai ollama[1180]: llama_context: n_seq_max = 2 5月 31 09:32:25 ai ollama[1180]: llama_context: n_ctx = 8192 5月 31 09:32:25 ai ollama[1180]: llama_context: n_ctx_per_seq = 4096 5月 31 09:32:25 ai ollama[1180]: llama_context: n_batch = 1024 ~ ▓▒░ ``` ### OS _No response_ ### GPU _No response_ ### CPU _No response_ ### Ollama version _No response_

GiteaMirror added the bug label 2026-05-04 17:34:48 -05:00

GiteaMirror closed this issue

2026-05-04 17:34:49 -05:00

GiteaMirror commented

2026-05-04 17:34:50 -05:00

@rick-github commented on GitHub (May 31, 2025):

5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.126+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc)

It looks like no backends were found. How did you install ollama?

@rick-github commented on GitHub (May 31, 2025): ``` 5月 31 08:45:03 ai ollama[1180]: time=2025-05-31T08:45:03.126+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc) ``` It looks like no backends were found. How did you install ollama?

GiteaMirror commented

2026-05-04 17:34:50 -05:00

@wlsoft2006 commented on GitHub (May 31, 2025):

curl -fsSL https://ollama.com/install.sh | sh

@wlsoft2006 commented on GitHub (May 31, 2025): curl -fsSL https://ollama.com/install.sh | sh

GiteaMirror commented

2026-05-04 17:34:50 -05:00

@rick-github commented on GitHub (May 31, 2025):

What's the output of

ls -lR /usr/local/lib/ollama

@rick-github commented on GitHub (May 31, 2025): What's the output of ``` ls -lR /usr/local/lib/ollama ```

GiteaMirror commented

2026-05-04 17:34:50 -05:00

@wlsoft2006 commented on GitHub (May 31, 2025):

What's the output of
ls -lR /usr/local/lib/ollama

/usr/local/lib/ollama:
总计 0

@wlsoft2006 commented on GitHub (May 31, 2025): > What's the output of > > ``` > ls -lR /usr/local/lib/ollama > ``` /usr/local/lib/ollama: 总计 0

GiteaMirror commented

2026-05-04 17:34:51 -05:00

@wlsoft2006 commented on GitHub (May 31, 2025):

Normal access to gpu when installing 0.71

@wlsoft2006 commented on GitHub (May 31, 2025): Normal access to gpu when installing 0.71

GiteaMirror commented

2026-05-04 17:34:51 -05:00

@rick-github commented on GitHub (May 31, 2025):

Your installation is incomplete. Try running the install again. If any errors are shown, paste them here.

@rick-github commented on GitHub (May 31, 2025): Your installation is incomplete. Try running the install again. If any errors are shown, paste them here.

GiteaMirror commented

2026-05-04 17:34:52 -05:00

@wlsoft2006 commented on GitHub (May 31, 2025):

Your installation is incomplete. Try running the install again. If any errors are shown, paste them here.

Thanks, I'll try reinstalling it

@wlsoft2006 commented on GitHub (May 31, 2025): > Your installation is incomplete. Try running the install again. If any errors are shown, paste them here. Thanks, I'll try reinstalling it

GiteaMirror commented

2026-05-04 17:34:53 -05:00

@wlsoft2006 commented on GitHub (May 31, 2025):

The problem has been solved, thank you

@wlsoft2006 commented on GitHub (May 31, 2025): The problem has been solved, thank you

Sign in to join this conversation.

Branches Tags

main

dhiltgen/ci

dhiltgen/llama-runner

hoyyeva/anthropic-local-image-path

hoyyeva/anthropic-reference-images-path

parth-anthropic-reference-images-path

brucemacd/download-before-remove

hoyyeva/editor-config-repair

parth-mlx-decode-checkpoints

parth-launch-codex-app

hoyyeva/fix-codex-model-metadata-warning

hoyyeva/qwen

parth/hide-claude-desktop-till-release

hoyyeva/opencode-image-modality

parth-add-claude-code-autoinstall

release_v0.22.0

pdevine/manifest-list

codex/fix-codex-model-metadata-warning

pdevine/addressable-manifest

brucemacd/launch-fetch-reccomended

jmorganca/llama-compat

launch-copilot-cli

hoyyeva/opencode-thinking

release_v0.20.7

parth-auto-save-backup

parth-test

jmorganca/gemma4-audio-replacements

fix-manifest-digest-on-pull

hoyyeva/vscode-improve

brucemacd/install-server-wait

parth/update-claude-docs

brucemac/start-ap-install

pdevine/mlx-update

pdevine/qwen35_vision

drifkin/api-show-fallback

mintlify/image-generation-1773352582

hoyyeva/server-context-length-local-config

jmorganca/faster-reptition-penalties

jmorganca/convert-nemotron

parth-pi-thinking

pdevine/sampling-penalties

jmorganca/fix-create-quantization-memory

dongchen/resumable_transfer_fix

pdevine/sampling-cache-error

jessegross/mlx-usage

hoyyeva/openclaw-config

hoyyeva/app-html

pdevine/qwen3next

brucemacd/sign-sh-install

brucemacd/tui-update

brucemacd/usage-api

jmorganca/launch-empty

fix-app-dist-embed

mxyng/mlx-compile

mxyng/mlx-quant

mxyng/mlx-glm4.7

mxyng/mlx

brucemacd/simplify-model-picker

jmorganca/qwen3-concurrent

fix-glm-4.7-flash-mla-config

drifkin/qwen3-coder-opening-tag

brucemacd/usage-cli

fix-cuda12-fattn-shmem

ollama-imagegen-docs

parth/fix-multiline-inputs

brucemacd/config-docs

mxyng/model-files

mxyng/simple-execute

fix-imagegen-ollama-models

mxyng/async-upload

jmorganca/lazy-no-dtype-changes

imagegen-auto-detect-create

parth/decrease-concurrent-download-hf

fix-mlx-quantize-init

jmorganca/x-cleanup

usage

imagegen-readme

jmorganca/glm-image

mlx-gpu-cd

jmorganca/imagegen-modelfile

parth/agent-skills

parth/agent-allowlist

parth/signed-in-offline

parth/agents

parth/fix-context-chopping

improve-cloud-flow

parth/add-models-websearch

parth/prompt-renderer-mcp

jmorganca/native-settings

jmorganca/download-stream-hash

jmorganca/client2-rebased

brucemacd/oai-chat-req-multipart

jessegross/multi_chunk_reserve

grace/additional-omit-empty

grace/mistral-3-large

mxyng/tokenizer2

mxyng/tokenizer

jessegross/flash

hoyyeva/windows-nacked-app

mxyng/cleanup-attention

grace/deepseek-parser

hoyyeva/remember-unsent-prompt

parth/add-lfs-pointer-error-conversion

parth/olmo2-test2

hoyyeva/ollama-launchagent-plist

nicole/olmo-model

parth/olmo-test

mxyng/remove-embedded

parth/render-template

jmorganca/intellect-3

parth/remove-prealloc-linter

jmorganca/cmd-eval

nicole/nomic-embed-text-fix

mxyng/lint-2

hoyyeva/add-gemini-3-pro-preview

hoyyeva/load-model-list

mxyng/expand-path

mxyng/environ-2

hoyyeva/deeplink-json-encoding

parth/improve-tool-calling-tests

hoyyeva/conversation

hoyyeva/assistant-edit-response

hoyyeva/thinking

origin/brucemacd/invalid-char-i-err

parth/improve-tool-calling

jmorganca/required-omitempty

grace/qwen3-vl-tests

mxyng/iter-client

parth/docs-readme

nicole/embed-test

pdevine/integration-benchstat

parth/remove-generate-cmd

parth/add-toolcall-id

mxyng/server-tests

jmorganca/glm-4.6

jmorganca/gin-h-compat

drifkin/stable-tool-args

pdevine/qwen3-more-thinking

parth/add-websearch-client

nicole/websearch_local

jmorganca/qwen3-coder-updates

grace/deepseek-v3-migration-tests

mxyng/fix-create

jmorganca/cloud-errors

pdevine/parser-tidy

revert-12233-parth/simplify-entrypoints-runner

parth/enable-so-gpt-oss

brucemacd/qwen3vl

jmorganca/readme-simplify

parth/gpt-oss-structured-outputs

revert-12039-jmorganca/tools-braces

mxyng/embeddings

mxyng/gguf

mxyng/benchmark

mxyng/types-null

parth/move-parsing

mxyng/gemma2

jmorganca/docs

mxyng/16-bit

mxyng/create-stdin

pdevine/authorizedkeys

mxyng/quant

parth/opt-in-error-context-window

brucemacd/cache-models

brucemacd/runner-completion

jmorganca/llama-update-6

brucemacd/benchmark-list

brucemacd/partial-read-caps

parth/deepseek-r1-tools

mxyng/omit-array

parth/tool-prefix-temp

brucemacd/runner-test

jmorganca/qwen25vl

brucemacd/model-forward-test-ext

parth/python-function-parsing

jmorganca/cuda-compression-none

drifkin/num-parallel

drifkin/chat-truncation-fix

jmorganca/sync

parth/python-tools-calling

drifkin/array-head-count

brucemacd/create-no-loop

parth/server-enable-content-stream-with-tools

qwen25omni

mxyng/v3

brucemacd/ropeconfig

jmorganca/silence-tokenizer

parth/sample-so-test

parth/sampling-structured-outputs

brucemacd/doc-go-engine

parth/constrained-sampling-json

jmorganca/mistral-wip

brucemacd/mistral-small-convert

parth/sample-unmarshal-json-for-params

brucemacd/jomorganca/mistral

pdevine/bfloat16

jmorganca/mistral

brucemacd/mistral

pdevine/logging

parth/sample-correctness-fix

parth/sample-fix-sorting

jmorgan/sample-fix-sorting-extras

jmorganca/temp-0-images

brucemacd/parallel-embed-models

brucemacd/shim-grammar

jmorganca/fix-gguf-error

bmizerany/nameswork

jmorganca/faster-releases

bmizerany/validatenames

brucemacd/err-no-vocab

brucemacd/rope-config

brucemacd/err-hint

brucemacd/qwen2_5

brucemacd/logprobs

brucemacd/new_runner_graph_bench

progress-flicker

brucemacd/forward-test

brucemacd/go_qwen2

pdevine/gemma2

jmorganca/add-missing-symlink-eval

mxyng/next-debug

parth/set-context-size-openai

brucemacd/next-bpe-bench

brucemacd/next-bpe-test

brucemacd/new_runner_e2e

brucemacd/new_runner_qwen2

pdevine/convert-cohere2

brucemacd/convert-cli

parth/log-probs

mxyng/next-mlx

mxyng/cmd-history

parth/templating

parth/tokenize-detokenize

brucemacd/check-key-register

bmizerany/grammar

jmorganca/vendor-081b29bd

mxyng/func-checks

jmorganca/fix-null-format

parth/fix-default-to-warn-json

jmorganca/qwen2vl

jmorganca/no-concat

parth/cmd-cleanup-SO

brucemacd/check-key-register-structured-err

parth/openai-stream-usage

parth/fix-referencing-so

stream-tools-stop

jmorganca/degin-1

brucemacd/install-path-clean

brucemacd/push-name-validation

brucemacd/browser-key-register

jmorganca/openai-fix-first-message

jmorganca/fix-proxy

jessegross/sample

parth/disallow-streaming-tools

dhiltgen/remove_submodule

jmorganca/ga

jmorganca/mllama

pdevine/newlines

pdevine/geems-2b

jmorganca/llama-bump

mxyng/modelname-7

mxyng/gin-slog

mxyng/modelname-6

jyan/convert-prog

jyan/quant5

paligemma-support

pdevine/import-docs

jmorganca/openai-context

jyan/paligemma

jyan/p2

jyan/palitest

bmizerany/embedspeedup

jmorganca/llama-vit

brucemacd/allow-ollama

royh/ep-methods

royh/whisper

mxyng/api-models

mxyng/fix-memory

jyan/q4_4/8

jyan/ollama-v

royh/stream-tools

roy-embed-parallel

bmizerany/hrm

revert-5963-revert-5924-mxyng/llama3.1-rope

royh/embed-viz

jyan/local2

jyan/auth

jyan/local

jyan/parse-temp

jmorganca/template-mistral

jyan/reord-g

royh-openai-suffixdocs

royh-imgembed

royh-embed-parallel

jyan/quant4

royh-precision

jyan/progress

pdevine/fix-template

jyan/quant3

pdevine/ggla

mxyng/update-registry-domain

jmorganca/ggml-static

mxyng/create-context

jyan/v0.146

mxyng/layers-from-files

build_dist

bmizerany/noseek

royh-ls

royh-name

timeout

mxyng/server-timestamp

bmizerany/nosillyggufslurps

royh-params

jmorganca/llama-cpp-7c26775

royh-openai-delete

royh-show-rigid

jmorganca/enable-fa

jmorganca/no-error-template

jyan/format

royh-testdelete

bmizerany/fastverify

language_support

pdevine/ps-glitches

brucemacd/tokenize

bruce/iq-quants

bmizerany/filepathwithcoloninhost

mxyng/split-bin

bmizerany/client-registry

jmorganca/if-none-match

native

jmorganca/native

jmorganca/batch-embeddings

jmorganca/initcmake

jmorganca/mm

pdevine/showggmlinfo

modenameenforcealphanum

bmizerany/modenameenforcealphanum

jmorganca/done-reason

jmorganca/llama-cpp-8960fe8

ollama.com

bmizerany/filepathnobuild

bmizerany/types/model/defaultfix

rmdisplaylong

nogogen

bmizerany/x

modelfile-readme

bmizerany/replacecolon

jmorganca/limit

jmorganca/execstack

jmorganca/replace-assets

mxyng/tune-concurrency

jmorganca/testing

whitespace-detection

jmorganca/options

upgrade-all

scratch

cuda-search

mattw/airenamer

mattw/allmodelsonhuggingface

mattw/quantcontext

mattw/whatneedstorun

brucemacd/llama-mem-calc

mattw/faq-context

mattw/communitylinks

mattw/noprune

mattw/python-functioncalling

rename

mxyng/install

pulse

remove-first

editor

mattw/selfqueryingretrieval

cgo

mattw/howtoquant

api

matt/streamingapi

format-config

mxyng/extra-args

shell

update-nous-hermes

cp-model

upload-progress

fix-unknown-model

fix-model-names

delete-fix

insecure-registry

ls

deletemodels

progressbar

readme-updates

license-layers

skip-list

list-models

modelpath

matt/examplemodelfiles

distribution

go-opts

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: github-starred/ollama#69246