time=2026-06-10T09:32:34.073+08:00 level=INFO source=routes.go:1919 msg="server config" env="map[HTTPS_PROXY: HTTP_PROXY: LLAMA_ARG_FIT: LLAMA_ARG_FIT_TARGET: NO_PROXY: OLLAMA_CONTEXT_LENGTH:0 OLLAMA_DEBUG:INFO OLLAMA_DEBUG_LOG_REQUESTS:false OLLAMA_EDITOR: OLLAMA_FLASH_ATTENTION:false OLLAMA_GO_TEMPLATE:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_IGPU_ENABLE: OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_TRANSFER_STREAMS:4 OLLAMA_MODELS:/Users/admin/.ollama/models OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NO_CLOUD:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false http_proxy: https_proxy: no_proxy:]" time=2026-06-10T09:32:34.073+08:00 level=INFO source=routes.go:1921 msg="Ollama cloud disabled: false" time=2026-06-10T09:32:34.074+08:00 level=INFO source=images.go:864 msg="total blobs: 5" time=2026-06-10T09:32:34.074+08:00 level=INFO source=images.go:871 msg="total unused blobs removed: 0" time=2026-06-10T09:32:34.074+08:00 level=INFO source=routes.go:1981 msg="Listening on 127.0.0.1:11434 (version 0.30.7)" time=2026-06-10T09:32:34.076+08:00 level=INFO source=runner.go:60 msg="discovering available GPUs..." time=2026-06-10T09:32:34.076+08:00 level=INFO source=model_list_cache.go:111 msg="model list cache hydration complete" models=2 failures=0 elapsed=2.011667ms time=2026-06-10T09:32:34.214+08:00 level=INFO source=runner.go:513 msg="failure during llama-server GPU discovery" OLLAMA_LIBRARY_PATH=[/Applications/Ollama.app/Contents/Resources] extra_envs=map[] error="llama-server --list-devices failed: exit status 1" detail="error: in directory_iterator::directory_iterator(...): Operation not permitted [\"/Users/admin/Downloads\"]" time=2026-06-10T09:32:34.214+08:00 level=INFO source=types.go:50 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="16.0 GiB" available="5.9 GiB" time=2026-06-10T09:32:34.214+08:00 level=INFO source=routes.go:2031 msg="vram-based default context" total_vram="0 B" default_num_ctx=4096 [GIN] 2026/06/10 - 09:32:34 | 200 | 64.5µs | 127.0.0.1 | GET "/api/version" [GIN] 2026/06/10 - 09:32:34 | 200 | 83.917µs | 127.0.0.1 | GET "/api/version" [GIN] 2026/06/10 - 09:32:34 | 401 | 630.512792ms | 127.0.0.1 | POST "/api/me" time=2026-06-10T09:32:34.846+08:00 level=INFO source=model_recommendations.go:177 msg="model recommendations cache sleep scheduled" wait=4h0m26.438174461s consecutive_failures=0 [GIN] 2026/06/10 - 09:32:50 | 200 | 2.390792ms | 127.0.0.1 | GET "/api/tags" time=2026-06-10T09:32:54.728+08:00 level=INFO source=sched.go:1148 msg="disabling mmap for llama-server load by default" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 reason=cpu time=2026-06-10T09:32:54.728+08:00 level=INFO source=server.go:109 msg="using llama-server for model" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T09:32:54.729+08:00 level=INFO source=llama_server.go:403 msg="starting llama-server" cmd="/Applications/Ollama.app/Contents/Resources/llama-server --model /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 --port 63600 --host 127.0.0.1 --no-webui --offline -c 4096 -np 1 --log-verbosity 4 --no-log-prefix --no-log-timestamps --no-jinja --chat-template chatml --no-mmap --flash-attn auto -b 512 -ub 512 --context-shift --keep 4" time=2026-06-10T09:32:54.731+08:00 level=INFO source=sched.go:613 msg="system memory" total="16.0 GiB" free="5.9 GiB" free_swap="0 B" time=2026-06-10T09:32:54.731+08:00 level=INFO source=llama_server.go:886 msg="loading model via llama-server" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T09:32:54.731+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T09:32:54.731+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server not responding" common_params_print_info: build 1 (6f3a9f3de) with AppleClang 21.0.0.21000099 for Darwin arm64 log_info: verbosity = 4 (adjust with the `-lv N` CLI arg) device_info: - MTL0 : Apple M2 Pro (12124 MiB, 12123 MiB free) - BLAS : Accelerate (0 MiB, 0 MiB free) - CPU : Apple M2 Pro (16384 MiB, 16384 MiB free) system_info: n_threads = 6 (n_threads_batch = 6) / 10 | MTL : EMBED_LIBRARY = 1 | CPU : NEON = 1 | ARM_FMA = 1 | FP16_VA = 1 | DOTPROD = 1 | LLAMAFILE = 1 | ACCELERATE = 1 | REPACK = 1 | srv init: using 9 threads for HTTP server srv init: The UI is disabled srv init: Use --ui/--no-ui (or deprecated --webui/--no-webui) to enable/disable srv start: binding port with default address family srv llama_server: loading model srv load_model: loading model '/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4' common_init_result: fitting params to device memory ... common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on) common_params_fit_impl: getting device memory data for initial parameters: common_memory_breakdown_print: | memory breakdown [MiB] | total free self model context compute unaccounted | common_memory_breakdown_print: | - MTL0 (Apple M2 Pro) | 12124 = 12123 + (1109 = 934 + 112 + 62) + -1108 | common_memory_breakdown_print: | - Host | 202 = 182 + 0 + 20 | common_params_fit_impl: projected to use 1109 MiB of device memory vs. 12123 MiB of free device memory common_params_fit_impl: will leave 11014 >= 1024 MiB of free device memory, no changes needed common_fit_params: successfully fit params to free device memory common_fit_params: fitting params to free memory took 0.14 seconds llama_model_loader: loaded meta data with 34 key-value pairs and 338 tensors from /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = qwen2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen2.5 1.5B Instruct llama_model_loader: - kv 3: general.finetune str = Instruct llama_model_loader: - kv 4: general.basename str = Qwen2.5 llama_model_loader: - kv 5: general.size_label str = 1.5B llama_model_loader: - kv 6: general.license str = apache-2.0 llama_model_loader: - kv 7: general.license.link str = https://huggingface.co/Qwen/Qwen2.5-1... llama_model_loader: - kv 8: general.base_model.count u32 = 1 llama_model_loader: - kv 9: general.base_model.0.name str = Qwen2.5 1.5B llama_model_loader: - kv 10: general.base_model.0.organization str = Qwen llama_model_loader: - kv 11: general.base_model.0.repo_url str = https://huggingface.co/Qwen/Qwen2.5-1.5B llama_model_loader: - kv 12: general.tags arr[str,2] = ["chat", "text-generation"] llama_model_loader: - kv 13: general.languages arr[str,1] = ["en"] llama_model_loader: - kv 14: qwen2.block_count u32 = 28 llama_model_loader: - kv 15: qwen2.context_length u32 = 32768 llama_model_loader: - kv 16: qwen2.embedding_length u32 = 1536 llama_model_loader: - kv 17: qwen2.feed_forward_length u32 = 8960 llama_model_loader: - kv 18: qwen2.attention.head_count u32 = 12 llama_model_loader: - kv 19: qwen2.attention.head_count_kv u32 = 2 llama_model_loader: - kv 20: qwen2.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 21: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 22: general.file_type u32 = 15 llama_model_loader: - kv 23: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 24: tokenizer.ggml.pre str = qwen2 llama_model_loader: - kv 25: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 26: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 27: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 28: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 29: tokenizer.ggml.padding_token_id u32 = 151643 llama_model_loader: - kv 30: tokenizer.ggml.bos_token_id u32 = 151643 llama_model_loader: - kv 31: tokenizer.ggml.add_bos_token bool = false llama_model_loader: - kv 32: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>... llama_model_loader: - kv 33: general.quantization_version u32 = 2 llama_model_loader: - type f32: 141 tensors llama_model_loader: - type q4_K: 168 tensors llama_model_loader: - type q6_K: 29 tensors print_info: file format = GGUF V3 (latest) print_info: file type = Q4_K - Medium print_info: file size = 934.69 MiB (5.08 BPW) llama_prepare_model_devices: using device MTL0 (Apple M2 Pro) (unknown id) - 12123 MiB free time=2026-06-10T09:32:54.984+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server loading model" load: 0 unused tokens load: control-looking token: 128247 '' was not control-type; this is probably a bug in the model. its type will be overridden load: printing all EOG tokens: load: - 128247 ('') load: - 151643 ('<|endoftext|>') load: - 151645 ('<|im_end|>') load: - 151662 ('<|fim_pad|>') load: - 151663 ('<|repo_name|>') load: - 151664 ('<|file_sep|>') load: special tokens cache size = 23 load: token to piece cache size = 0.9310 MB print_info: arch = qwen2 print_info: vocab_only = 0 print_info: no_alloc = 0 print_info: n_ctx_train = 32768 print_info: n_embd = 1536 print_info: n_embd_inp = 1536 print_info: n_layer = 28 print_info: n_head = 12 print_info: n_head_kv = 2 print_info: n_rot = 128 print_info: n_swa = 0 print_info: is_swa_any = 0 print_info: n_embd_head_k = 128 print_info: n_embd_head_v = 128 print_info: n_gqa = 6 print_info: n_embd_k_gqa = 256 print_info: n_embd_v_gqa = 256 print_info: f_norm_eps = 0.0e+00 print_info: f_norm_rms_eps = 1.0e-06 print_info: f_clamp_kqv = 0.0e+00 print_info: f_max_alibi_bias = 0.0e+00 print_info: f_logit_scale = 0.0e+00 print_info: f_attn_scale = 0.0e+00 print_info: f_attn_value_scale = 0.0000 print_info: n_ff = 8960 print_info: n_expert = 0 print_info: n_expert_used = 0 print_info: n_expert_groups = 0 print_info: n_group_used = 0 print_info: causal attn = 1 print_info: pooling type = -1 print_info: rope type = 2 print_info: rope scaling = linear print_info: freq_base_train = 1000000.0 print_info: freq_scale_train = 1 print_info: n_ctx_orig_yarn = 32768 print_info: rope_yarn_log_mul = 0.0000 print_info: rope_finetuned = unknown print_info: model type = 1.5B print_info: model params = 1.54 B print_info: general.name = Qwen2.5 1.5B Instruct print_info: vocab type = BPE print_info: n_vocab = 151936 print_info: n_merges = 151387 print_info: BOS token = 151643 '<|endoftext|>' print_info: EOS token = 151645 '<|im_end|>' print_info: EOT token = 151645 '<|im_end|>' print_info: PAD token = 151643 '<|endoftext|>' print_info: LF token = 198 'Ċ' print_info: FIM PRE token = 151659 '<|fim_prefix|>' print_info: FIM SUF token = 151661 '<|fim_suffix|>' print_info: FIM MID token = 151660 '<|fim_middle|>' print_info: FIM PAD token = 151662 '<|fim_pad|>' print_info: FIM REP token = 151663 '<|repo_name|>' print_info: FIM SEP token = 151664 '<|file_sep|>' print_info: EOG token = 128247 '' print_info: EOG token = 151643 '<|endoftext|>' print_info: EOG token = 151645 '<|im_end|>' print_info: EOG token = 151662 '<|fim_pad|>' print_info: EOG token = 151663 '<|repo_name|>' print_info: EOG token = 151664 '<|file_sep|>' print_info: max token length = 256 load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false) load_tensors: offloading output layer to GPU load_tensors: offloading 27 repeating layers to GPU load_tensors: offloaded 29/29 layers to GPU load_tensors: CPU model buffer size = 182.57 MiB load_tensors: MTL0 model buffer size = 934.69 MiB ...................................................................... common_init_result: added logit bias = -inf common_init_result: added <|endoftext|> logit bias = -inf common_init_result: added <|im_end|> logit bias = -inf common_init_result: added <|fim_pad|> logit bias = -inf common_init_result: added <|repo_name|> logit bias = -inf common_init_result: added <|file_sep|> logit bias = -inf llama_context: constructing llama_context llama_context: n_seq_max = 1 llama_context: n_ctx = 4096 llama_context: n_ctx_seq = 4096 llama_context: n_batch = 512 llama_context: n_ubatch = 512 llama_context: causal_attn = 1 llama_context: flash_attn = auto llama_context: kv_unified = false llama_context: freq_base = 1000000.0 llama_context: freq_scale = 1 llama_context: n_rs_seq = 0 llama_context: n_outputs_max = 1 llama_context: n_ctx_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized ggml_metal_init: allocating ggml_metal_init: found device: Apple M2 Pro ggml_metal_init: picking default device: Apple M2 Pro ggml_metal_init: use fusion = true ggml_metal_init: use concurrency = true ggml_metal_init: use graph optimize = true llama_context: CPU output buffer size = 0.58 MiB llama_kv_cache: MTL0 KV buffer size = 112.00 MiB llama_kv_cache: size = 112.00 MiB ( 4096 cells, 28 layers, 1/1 seqs), K (f16): 56.00 MiB, V (f16): 56.00 MiB llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 128 llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 128 sched_reserve: reserving ... sched_reserve: Flash Attention was auto, set to enabled sched_reserve: resolving fused Gated Delta Net support: sched_reserve: fused Gated Delta Net (autoregressive) enabled sched_reserve: fused Gated Delta Net (chunked) enabled sched_reserve: MTL0 compute buffer size = 62.51 MiB sched_reserve: CPU compute buffer size = 10.01 MiB sched_reserve: graph nodes = 958 sched_reserve: graph splits = 2 sched_reserve: reserve took 5.58 ms, sched copies = 1 common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable) srv load_model: initializing slots, n_slots = 1 common_speculative_init: no implementations specified for speculative decoding slot load_model: id 0 | task -1 | new slot, n_ctx = 4096 srv load_model: prompt cache is enabled, size limit: 8192 MiB srv load_model: use `--cache-ram 0` to disable the prompt cache srv load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391 srv load_model: context checkpoints enabled, max = 32, min spacing = 256 srv init: --cache-idle-slots requires --kv-unified, disabling init: chat template, example_format: '<|im_start|>system You are a helpful assistant<|im_end|> <|im_start|>user Hello<|im_end|> <|im_start|>assistant Hi there<|im_end|> <|im_start|>user How are you?<|im_end|> <|im_start|>assistant ' srv init: init: chat template, thinking = 0 srv llama_server: model loaded srv llama_server: server is listening on http://127.0.0.1:63600 srv update_slots: all slots are idle time=2026-06-10T09:32:55.740+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 1.01 seconds" time=2026-06-10T09:32:55.770+08:00 level=INFO source=images.go:354 msg="template selection" model=registry.ollama.ai/library/qwen2.5:1.5b selected=go_template renderer="" parser="" go_template="[completion tools]" chat_template="[tools completion]" harmony=null renderer_parser=null time=2026-06-10T09:32:55.770+08:00 level=INFO source=sched.go:729 msg="loaded runners" count=1 time=2026-06-10T09:32:55.770+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T09:32:55.770+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 1.04 seconds" srv log_server_r: done request: POST /tokenize 127.0.0.1 200 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = -1 srv get_availabl: updating prompt cache srv load: - looking for better prompt, base f_keep = -1.000, sim = 0.000 srv update: - cache state: 0 prompts, 0.000 MiB (limits: 8192.000 MiB, 4096 tokens, 8589934592 est) srv get_availabl: prompt cache update took 0.00 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> top-p -> ?min-p -> ?xtc -> temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 0.900, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 0.800 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 0 | processing task, is_child = 0 slot update_slots: id 0 | task 0 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 33 slot update_slots: id 0 | task 0 | cached n_tokens = 0, memory_seq_rm [0, end) slot init_sampler: id 0 | task 0 | init sampler, took 0.01 ms, tokens: text = 33, total = 33 srv log_server_r: done request: POST /completion 127.0.0.1 200 slot print_timing: id 0 | task 0 | prompt eval time = 221.31 ms / 33 tokens ( 6.71 ms per token, 149.12 tokens per second) slot print_timing: id 0 | task 0 | eval time = 227.40 ms / 24 tokens ( 9.48 ms per token, 105.54 tokens per second) slot print_timing: id 0 | task 0 | total time = 448.71 ms / 57 tokens slot print_timing: id 0 | task 0 | graphs reused = 23 slot release: id 0 | task 0 | stop processing: n_tokens = 56, truncated = 0 srv update_slots: all slots are idle srv log_server_r: done request: POST /tokenize 127.0.0.1 200 [GIN] 2026/06/10 - 09:32:56 | 200 | 1.673121083s | 127.0.0.1 | POST "/api/generate" time=2026-06-10T09:47:13.091+08:00 level=INFO source=sched.go:1148 msg="disabling mmap for llama-server load by default" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 reason=cpu time=2026-06-10T09:47:13.091+08:00 level=INFO source=server.go:109 msg="using llama-server for model" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T09:47:13.092+08:00 level=INFO source=llama_server.go:403 msg="starting llama-server" cmd="/Applications/Ollama.app/Contents/Resources/llama-server --model /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 --port 64176 --host 127.0.0.1 --no-webui --offline -c 4096 -np 1 --log-verbosity 4 --no-log-prefix --no-log-timestamps --no-jinja --chat-template chatml --no-mmap --flash-attn auto -b 512 -ub 512 --context-shift --keep 4" time=2026-06-10T09:47:13.094+08:00 level=INFO source=sched.go:613 msg="system memory" total="16.0 GiB" free="6.9 GiB" free_swap="0 B" time=2026-06-10T09:47:13.094+08:00 level=INFO source=llama_server.go:886 msg="loading model via llama-server" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T09:47:13.094+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T09:47:13.094+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server not responding" common_params_print_info: build 1 (6f3a9f3de) with AppleClang 21.0.0.21000099 for Darwin arm64 log_info: verbosity = 4 (adjust with the `-lv N` CLI arg) device_info: - MTL0 : Apple M2 Pro (12124 MiB, 12123 MiB free) - BLAS : Accelerate (0 MiB, 0 MiB free) - CPU : Apple M2 Pro (16384 MiB, 16384 MiB free) system_info: n_threads = 6 (n_threads_batch = 6) / 10 | MTL : EMBED_LIBRARY = 1 | CPU : NEON = 1 | ARM_FMA = 1 | FP16_VA = 1 | DOTPROD = 1 | LLAMAFILE = 1 | ACCELERATE = 1 | REPACK = 1 | srv init: using 9 threads for HTTP server srv init: The UI is disabled srv init: Use --ui/--no-ui (or deprecated --webui/--no-webui) to enable/disable srv start: binding port with default address family srv llama_server: loading model srv load_model: loading model '/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4' common_init_result: fitting params to device memory ... common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on) common_params_fit_impl: getting device memory data for initial parameters: common_memory_breakdown_print: | memory breakdown [MiB] | total free self model context compute unaccounted | common_memory_breakdown_print: | - MTL0 (Apple M2 Pro) | 12124 = 12123 + (1109 = 934 + 112 + 62) + -1108 | common_memory_breakdown_print: | - Host | 202 = 182 + 0 + 20 | common_params_fit_impl: projected to use 1109 MiB of device memory vs. 12123 MiB of free device memory common_params_fit_impl: will leave 11014 >= 1024 MiB of free device memory, no changes needed common_fit_params: successfully fit params to free device memory common_fit_params: fitting params to free memory took 0.14 seconds llama_model_loader: loaded meta data with 34 key-value pairs and 338 tensors from /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = qwen2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen2.5 1.5B Instruct llama_model_loader: - kv 3: general.finetune str = Instruct llama_model_loader: - kv 4: general.basename str = Qwen2.5 llama_model_loader: - kv 5: general.size_label str = 1.5B llama_model_loader: - kv 6: general.license str = apache-2.0 llama_model_loader: - kv 7: general.license.link str = https://huggingface.co/Qwen/Qwen2.5-1... llama_model_loader: - kv 8: general.base_model.count u32 = 1 llama_model_loader: - kv 9: general.base_model.0.name str = Qwen2.5 1.5B llama_model_loader: - kv 10: general.base_model.0.organization str = Qwen llama_model_loader: - kv 11: general.base_model.0.repo_url str = https://huggingface.co/Qwen/Qwen2.5-1.5B llama_model_loader: - kv 12: general.tags arr[str,2] = ["chat", "text-generation"] llama_model_loader: - kv 13: general.languages arr[str,1] = ["en"] llama_model_loader: - kv 14: qwen2.block_count u32 = 28 llama_model_loader: - kv 15: qwen2.context_length u32 = 32768 llama_model_loader: - kv 16: qwen2.embedding_length u32 = 1536 llama_model_loader: - kv 17: qwen2.feed_forward_length u32 = 8960 llama_model_loader: - kv 18: qwen2.attention.head_count u32 = 12 llama_model_loader: - kv 19: qwen2.attention.head_count_kv u32 = 2 llama_model_loader: - kv 20: qwen2.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 21: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 22: general.file_type u32 = 15 llama_model_loader: - kv 23: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 24: tokenizer.ggml.pre str = qwen2 llama_model_loader: - kv 25: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 26: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 27: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 28: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 29: tokenizer.ggml.padding_token_id u32 = 151643 llama_model_loader: - kv 30: tokenizer.ggml.bos_token_id u32 = 151643 llama_model_loader: - kv 31: tokenizer.ggml.add_bos_token bool = false llama_model_loader: - kv 32: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>... llama_model_loader: - kv 33: general.quantization_version u32 = 2 llama_model_loader: - type f32: 141 tensors llama_model_loader: - type q4_K: 168 tensors llama_model_loader: - type q6_K: 29 tensors print_info: file format = GGUF V3 (latest) print_info: file type = Q4_K - Medium print_info: file size = 934.69 MiB (5.08 BPW) llama_prepare_model_devices: using device MTL0 (Apple M2 Pro) (unknown id) - 12123 MiB free time=2026-06-10T09:47:13.346+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server loading model" load: 0 unused tokens load: control-looking token: 128247 '' was not control-type; this is probably a bug in the model. its type will be overridden load: printing all EOG tokens: load: - 128247 ('') load: - 151643 ('<|endoftext|>') load: - 151645 ('<|im_end|>') load: - 151662 ('<|fim_pad|>') load: - 151663 ('<|repo_name|>') load: - 151664 ('<|file_sep|>') load: special tokens cache size = 23 load: token to piece cache size = 0.9310 MB print_info: arch = qwen2 print_info: vocab_only = 0 print_info: no_alloc = 0 print_info: n_ctx_train = 32768 print_info: n_embd = 1536 print_info: n_embd_inp = 1536 print_info: n_layer = 28 print_info: n_head = 12 print_info: n_head_kv = 2 print_info: n_rot = 128 print_info: n_swa = 0 print_info: is_swa_any = 0 print_info: n_embd_head_k = 128 print_info: n_embd_head_v = 128 print_info: n_gqa = 6 print_info: n_embd_k_gqa = 256 print_info: n_embd_v_gqa = 256 print_info: f_norm_eps = 0.0e+00 print_info: f_norm_rms_eps = 1.0e-06 print_info: f_clamp_kqv = 0.0e+00 print_info: f_max_alibi_bias = 0.0e+00 print_info: f_logit_scale = 0.0e+00 print_info: f_attn_scale = 0.0e+00 print_info: f_attn_value_scale = 0.0000 print_info: n_ff = 8960 print_info: n_expert = 0 print_info: n_expert_used = 0 print_info: n_expert_groups = 0 print_info: n_group_used = 0 print_info: causal attn = 1 print_info: pooling type = -1 print_info: rope type = 2 print_info: rope scaling = linear print_info: freq_base_train = 1000000.0 print_info: freq_scale_train = 1 print_info: n_ctx_orig_yarn = 32768 print_info: rope_yarn_log_mul = 0.0000 print_info: rope_finetuned = unknown print_info: model type = 1.5B print_info: model params = 1.54 B print_info: general.name = Qwen2.5 1.5B Instruct print_info: vocab type = BPE print_info: n_vocab = 151936 print_info: n_merges = 151387 print_info: BOS token = 151643 '<|endoftext|>' print_info: EOS token = 151645 '<|im_end|>' print_info: EOT token = 151645 '<|im_end|>' print_info: PAD token = 151643 '<|endoftext|>' print_info: LF token = 198 'Ċ' print_info: FIM PRE token = 151659 '<|fim_prefix|>' print_info: FIM SUF token = 151661 '<|fim_suffix|>' print_info: FIM MID token = 151660 '<|fim_middle|>' print_info: FIM PAD token = 151662 '<|fim_pad|>' print_info: FIM REP token = 151663 '<|repo_name|>' print_info: FIM SEP token = 151664 '<|file_sep|>' print_info: EOG token = 128247 '' print_info: EOG token = 151643 '<|endoftext|>' print_info: EOG token = 151645 '<|im_end|>' print_info: EOG token = 151662 '<|fim_pad|>' print_info: EOG token = 151663 '<|repo_name|>' print_info: EOG token = 151664 '<|file_sep|>' print_info: max token length = 256 load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false) load_tensors: offloading output layer to GPU load_tensors: offloading 27 repeating layers to GPU load_tensors: offloaded 29/29 layers to GPU load_tensors: CPU model buffer size = 182.57 MiB load_tensors: MTL0 model buffer size = 934.69 MiB ...................................................................... common_init_result: added logit bias = -inf common_init_result: added <|endoftext|> logit bias = -inf common_init_result: added <|im_end|> logit bias = -inf common_init_result: added <|fim_pad|> logit bias = -inf common_init_result: added <|repo_name|> logit bias = -inf common_init_result: added <|file_sep|> logit bias = -inf llama_context: constructing llama_context llama_context: n_seq_max = 1 llama_context: n_ctx = 4096 llama_context: n_ctx_seq = 4096 llama_context: n_batch = 512 llama_context: n_ubatch = 512 llama_context: causal_attn = 1 llama_context: flash_attn = auto llama_context: kv_unified = false llama_context: freq_base = 1000000.0 llama_context: freq_scale = 1 llama_context: n_rs_seq = 0 llama_context: n_outputs_max = 1 llama_context: n_ctx_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized ggml_metal_init: allocating ggml_metal_init: found device: Apple M2 Pro ggml_metal_init: picking default device: Apple M2 Pro ggml_metal_init: use fusion = true ggml_metal_init: use concurrency = true ggml_metal_init: use graph optimize = true llama_context: CPU output buffer size = 0.58 MiB llama_kv_cache: MTL0 KV buffer size = 112.00 MiB llama_kv_cache: size = 112.00 MiB ( 4096 cells, 28 layers, 1/1 seqs), K (f16): 56.00 MiB, V (f16): 56.00 MiB llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 128 llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 128 sched_reserve: reserving ... sched_reserve: Flash Attention was auto, set to enabled sched_reserve: resolving fused Gated Delta Net support: sched_reserve: fused Gated Delta Net (autoregressive) enabled sched_reserve: fused Gated Delta Net (chunked) enabled sched_reserve: MTL0 compute buffer size = 62.51 MiB sched_reserve: CPU compute buffer size = 10.01 MiB sched_reserve: graph nodes = 958 sched_reserve: graph splits = 2 sched_reserve: reserve took 6.23 ms, sched copies = 1 common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable) srv load_model: initializing slots, n_slots = 1 common_speculative_init: no implementations specified for speculative decoding slot load_model: id 0 | task -1 | new slot, n_ctx = 4096 srv load_model: prompt cache is enabled, size limit: 8192 MiB srv load_model: use `--cache-ram 0` to disable the prompt cache srv load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391 srv load_model: context checkpoints enabled, max = 32, min spacing = 256 srv init: --cache-idle-slots requires --kv-unified, disabling init: chat template, example_format: '<|im_start|>system You are a helpful assistant<|im_end|> <|im_start|>user Hello<|im_end|> <|im_start|>assistant Hi there<|im_end|> <|im_start|>user How are you?<|im_end|> <|im_start|>assistant ' srv init: init: chat template, thinking = 0 srv llama_server: model loaded srv llama_server: server is listening on http://127.0.0.1:64176 srv update_slots: all slots are idle time=2026-06-10T09:47:13.849+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 0.76 seconds" time=2026-06-10T09:47:13.868+08:00 level=INFO source=images.go:354 msg="template selection" model=registry.ollama.ai/library/qwen2.5:1.5b selected=go_template renderer="" parser="" go_template="[completion tools]" chat_template="[tools completion]" harmony=null renderer_parser=null time=2026-06-10T09:47:13.868+08:00 level=INFO source=sched.go:729 msg="loaded runners" count=1 time=2026-06-10T09:47:13.868+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T09:47:13.869+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 0.78 seconds" srv log_server_r: done request: POST /tokenize 127.0.0.1 200 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = -1 srv get_availabl: updating prompt cache srv load: - looking for better prompt, base f_keep = -1.000, sim = 0.000 srv update: - cache state: 0 prompts, 0.000 MiB (limits: 8192.000 MiB, 4096 tokens, 8589934592 est) srv get_availabl: prompt cache update took 0.00 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> top-p -> ?min-p -> ?xtc -> temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 0.900, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 0.800 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 0 | processing task, is_child = 0 slot update_slots: id 0 | task 0 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 35 slot update_slots: id 0 | task 0 | cached n_tokens = 0, memory_seq_rm [0, end) slot init_sampler: id 0 | task 0 | init sampler, took 0.01 ms, tokens: text = 35, total = 35 srv log_server_r: done request: POST /completion 127.0.0.1 200 slot print_timing: id 0 | task 0 | prompt eval time = 56.92 ms / 35 tokens ( 1.63 ms per token, 614.88 tokens per second) slot print_timing: id 0 | task 0 | eval time = 154.35 ms / 15 tokens ( 10.29 ms per token, 97.18 tokens per second) slot print_timing: id 0 | task 0 | total time = 211.28 ms / 50 tokens slot print_timing: id 0 | task 0 | graphs reused = 14 slot release: id 0 | task 0 | stop processing: n_tokens = 49, truncated = 0 srv update_slots: all slots are idle srv log_server_r: done request: POST /tokenize 127.0.0.1 200 [GIN] 2026/06/10 - 09:47:14 | 200 | 1.150395333s | 127.0.0.1 | POST "/api/generate" [GIN] 2026/06/10 - 10:39:49 | 200 | 4.339958ms | 127.0.0.1 | GET "/api/tags" time=2026-06-10T10:39:55.243+08:00 level=INFO source=sched.go:1148 msg="disabling mmap for llama-server load by default" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 reason=cpu time=2026-06-10T10:39:55.243+08:00 level=INFO source=server.go:109 msg="using llama-server for model" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T10:39:55.244+08:00 level=INFO source=llama_server.go:403 msg="starting llama-server" cmd="/Applications/Ollama.app/Contents/Resources/llama-server --model /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 --port 49216 --host 127.0.0.1 --no-webui --offline -c 4096 -np 1 --log-verbosity 4 --no-log-prefix --no-log-timestamps --no-jinja --chat-template chatml --no-mmap --flash-attn auto -b 512 -ub 512 --context-shift --keep 4" time=2026-06-10T10:39:55.250+08:00 level=INFO source=sched.go:613 msg="system memory" total="16.0 GiB" free="5.8 GiB" free_swap="0 B" time=2026-06-10T10:39:55.250+08:00 level=INFO source=llama_server.go:886 msg="loading model via llama-server" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T10:39:55.250+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T10:39:55.250+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server not responding" common_params_print_info: build 1 (6f3a9f3de) with AppleClang 21.0.0.21000099 for Darwin arm64 log_info: verbosity = 4 (adjust with the `-lv N` CLI arg) device_info: - MTL0 : Apple M2 Pro (12124 MiB, 12123 MiB free) - BLAS : Accelerate (0 MiB, 0 MiB free) - CPU : Apple M2 Pro (16384 MiB, 16384 MiB free) system_info: n_threads = 6 (n_threads_batch = 6) / 10 | MTL : EMBED_LIBRARY = 1 | CPU : NEON = 1 | ARM_FMA = 1 | FP16_VA = 1 | DOTPROD = 1 | LLAMAFILE = 1 | ACCELERATE = 1 | REPACK = 1 | srv init: using 9 threads for HTTP server srv init: The UI is disabled srv init: Use --ui/--no-ui (or deprecated --webui/--no-webui) to enable/disable srv start: binding port with default address family srv llama_server: loading model srv load_model: loading model '/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4' common_init_result: fitting params to device memory ... common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on) common_params_fit_impl: getting device memory data for initial parameters: common_memory_breakdown_print: | memory breakdown [MiB] | total free self model context compute unaccounted | common_memory_breakdown_print: | - MTL0 (Apple M2 Pro) | 12124 = 12123 + (1109 = 934 + 112 + 62) + -1108 | common_memory_breakdown_print: | - Host | 202 = 182 + 0 + 20 | common_params_fit_impl: projected to use 1109 MiB of device memory vs. 12123 MiB of free device memory common_params_fit_impl: will leave 11014 >= 1024 MiB of free device memory, no changes needed common_fit_params: successfully fit params to free device memory common_fit_params: fitting params to free memory took 0.14 seconds llama_model_loader: loaded meta data with 34 key-value pairs and 338 tensors from /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = qwen2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen2.5 1.5B Instruct llama_model_loader: - kv 3: general.finetune str = Instruct llama_model_loader: - kv 4: general.basename str = Qwen2.5 llama_model_loader: - kv 5: general.size_label str = 1.5B llama_model_loader: - kv 6: general.license str = apache-2.0 llama_model_loader: - kv 7: general.license.link str = https://huggingface.co/Qwen/Qwen2.5-1... llama_model_loader: - kv 8: general.base_model.count u32 = 1 llama_model_loader: - kv 9: general.base_model.0.name str = Qwen2.5 1.5B llama_model_loader: - kv 10: general.base_model.0.organization str = Qwen llama_model_loader: - kv 11: general.base_model.0.repo_url str = https://huggingface.co/Qwen/Qwen2.5-1.5B llama_model_loader: - kv 12: general.tags arr[str,2] = ["chat", "text-generation"] llama_model_loader: - kv 13: general.languages arr[str,1] = ["en"] llama_model_loader: - kv 14: qwen2.block_count u32 = 28 llama_model_loader: - kv 15: qwen2.context_length u32 = 32768 llama_model_loader: - kv 16: qwen2.embedding_length u32 = 1536 llama_model_loader: - kv 17: qwen2.feed_forward_length u32 = 8960 llama_model_loader: - kv 18: qwen2.attention.head_count u32 = 12 llama_model_loader: - kv 19: qwen2.attention.head_count_kv u32 = 2 llama_model_loader: - kv 20: qwen2.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 21: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 22: general.file_type u32 = 15 llama_model_loader: - kv 23: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 24: tokenizer.ggml.pre str = qwen2 llama_model_loader: - kv 25: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 26: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 27: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 28: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 29: tokenizer.ggml.padding_token_id u32 = 151643 llama_model_loader: - kv 30: tokenizer.ggml.bos_token_id u32 = 151643 llama_model_loader: - kv 31: tokenizer.ggml.add_bos_token bool = false llama_model_loader: - kv 32: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>... llama_model_loader: - kv 33: general.quantization_version u32 = 2 llama_model_loader: - type f32: 141 tensors llama_model_loader: - type q4_K: 168 tensors llama_model_loader: - type q6_K: 29 tensors print_info: file format = GGUF V3 (latest) print_info: file type = Q4_K - Medium print_info: file size = 934.69 MiB (5.08 BPW) llama_prepare_model_devices: using device MTL0 (Apple M2 Pro) (unknown id) - 12123 MiB free time=2026-06-10T10:39:56.006+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server loading model" load: 0 unused tokens load: control-looking token: 128247 '' was not control-type; this is probably a bug in the model. its type will be overridden load: printing all EOG tokens: load: - 128247 ('') load: - 151643 ('<|endoftext|>') load: - 151645 ('<|im_end|>') load: - 151662 ('<|fim_pad|>') load: - 151663 ('<|repo_name|>') load: - 151664 ('<|file_sep|>') load: special tokens cache size = 23 load: token to piece cache size = 0.9310 MB print_info: arch = qwen2 print_info: vocab_only = 0 print_info: no_alloc = 0 print_info: n_ctx_train = 32768 print_info: n_embd = 1536 print_info: n_embd_inp = 1536 print_info: n_layer = 28 print_info: n_head = 12 print_info: n_head_kv = 2 print_info: n_rot = 128 print_info: n_swa = 0 print_info: is_swa_any = 0 print_info: n_embd_head_k = 128 print_info: n_embd_head_v = 128 print_info: n_gqa = 6 print_info: n_embd_k_gqa = 256 print_info: n_embd_v_gqa = 256 print_info: f_norm_eps = 0.0e+00 print_info: f_norm_rms_eps = 1.0e-06 print_info: f_clamp_kqv = 0.0e+00 print_info: f_max_alibi_bias = 0.0e+00 print_info: f_logit_scale = 0.0e+00 print_info: f_attn_scale = 0.0e+00 print_info: f_attn_value_scale = 0.0000 print_info: n_ff = 8960 print_info: n_expert = 0 print_info: n_expert_used = 0 print_info: n_expert_groups = 0 print_info: n_group_used = 0 print_info: causal attn = 1 print_info: pooling type = -1 print_info: rope type = 2 print_info: rope scaling = linear print_info: freq_base_train = 1000000.0 print_info: freq_scale_train = 1 print_info: n_ctx_orig_yarn = 32768 print_info: rope_yarn_log_mul = 0.0000 print_info: rope_finetuned = unknown print_info: model type = 1.5B print_info: model params = 1.54 B print_info: general.name = Qwen2.5 1.5B Instruct print_info: vocab type = BPE print_info: n_vocab = 151936 print_info: n_merges = 151387 print_info: BOS token = 151643 '<|endoftext|>' print_info: EOS token = 151645 '<|im_end|>' print_info: EOT token = 151645 '<|im_end|>' print_info: PAD token = 151643 '<|endoftext|>' print_info: LF token = 198 'Ċ' print_info: FIM PRE token = 151659 '<|fim_prefix|>' print_info: FIM SUF token = 151661 '<|fim_suffix|>' print_info: FIM MID token = 151660 '<|fim_middle|>' print_info: FIM PAD token = 151662 '<|fim_pad|>' print_info: FIM REP token = 151663 '<|repo_name|>' print_info: FIM SEP token = 151664 '<|file_sep|>' print_info: EOG token = 128247 '' print_info: EOG token = 151643 '<|endoftext|>' print_info: EOG token = 151645 '<|im_end|>' print_info: EOG token = 151662 '<|fim_pad|>' print_info: EOG token = 151663 '<|repo_name|>' print_info: EOG token = 151664 '<|file_sep|>' print_info: max token length = 256 load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false) load_tensors: offloading output layer to GPU load_tensors: offloading 27 repeating layers to GPU load_tensors: offloaded 29/29 layers to GPU load_tensors: CPU model buffer size = 182.57 MiB load_tensors: MTL0 model buffer size = 934.69 MiB ...................................................................... common_init_result: added logit bias = -inf common_init_result: added <|endoftext|> logit bias = -inf common_init_result: added <|im_end|> logit bias = -inf common_init_result: added <|fim_pad|> logit bias = -inf common_init_result: added <|repo_name|> logit bias = -inf common_init_result: added <|file_sep|> logit bias = -inf llama_context: constructing llama_context llama_context: n_seq_max = 1 llama_context: n_ctx = 4096 llama_context: n_ctx_seq = 4096 llama_context: n_batch = 512 llama_context: n_ubatch = 512 llama_context: causal_attn = 1 llama_context: flash_attn = auto llama_context: kv_unified = false llama_context: freq_base = 1000000.0 llama_context: freq_scale = 1 llama_context: n_rs_seq = 0 llama_context: n_outputs_max = 1 llama_context: n_ctx_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized ggml_metal_init: allocating ggml_metal_init: found device: Apple M2 Pro ggml_metal_init: picking default device: Apple M2 Pro ggml_metal_init: use fusion = true ggml_metal_init: use concurrency = true ggml_metal_init: use graph optimize = true llama_context: CPU output buffer size = 0.58 MiB llama_kv_cache: MTL0 KV buffer size = 112.00 MiB llama_kv_cache: size = 112.00 MiB ( 4096 cells, 28 layers, 1/1 seqs), K (f16): 56.00 MiB, V (f16): 56.00 MiB llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 128 llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 128 sched_reserve: reserving ... sched_reserve: Flash Attention was auto, set to enabled sched_reserve: resolving fused Gated Delta Net support: sched_reserve: fused Gated Delta Net (autoregressive) enabled sched_reserve: fused Gated Delta Net (chunked) enabled sched_reserve: MTL0 compute buffer size = 62.51 MiB sched_reserve: CPU compute buffer size = 10.01 MiB sched_reserve: graph nodes = 958 sched_reserve: graph splits = 2 sched_reserve: reserve took 7.12 ms, sched copies = 1 common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable) srv load_model: initializing slots, n_slots = 1 common_speculative_init: no implementations specified for speculative decoding slot load_model: id 0 | task -1 | new slot, n_ctx = 4096 srv load_model: prompt cache is enabled, size limit: 8192 MiB srv load_model: use `--cache-ram 0` to disable the prompt cache srv load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391 srv load_model: context checkpoints enabled, max = 32, min spacing = 256 srv init: --cache-idle-slots requires --kv-unified, disabling init: chat template, example_format: '<|im_start|>system You are a helpful assistant<|im_end|> <|im_start|>user Hello<|im_end|> <|im_start|>assistant Hi there<|im_end|> <|im_start|>user How are you?<|im_end|> <|im_start|>assistant ' srv init: init: chat template, thinking = 0 srv llama_server: model loaded srv llama_server: server is listening on http://127.0.0.1:49216 srv update_slots: all slots are idle time=2026-06-10T10:39:56.761+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 1.51 seconds" time=2026-06-10T10:39:56.793+08:00 level=INFO source=images.go:354 msg="template selection" model=registry.ollama.ai/library/qwen2.5:1.5b selected=go_template renderer="" parser="" go_template="[completion tools]" chat_template="[tools completion]" harmony=null renderer_parser=null time=2026-06-10T10:39:56.793+08:00 level=INFO source=sched.go:729 msg="loaded runners" count=1 time=2026-06-10T10:39:56.793+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T10:39:56.793+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 1.54 seconds" srv log_server_r: done request: POST /tokenize 127.0.0.1 200 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = -1 srv get_availabl: updating prompt cache srv load: - looking for better prompt, base f_keep = -1.000, sim = 0.000 srv update: - cache state: 0 prompts, 0.000 MiB (limits: 8192.000 MiB, 4096 tokens, 8589934592 est) srv get_availabl: prompt cache update took 0.00 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> top-p -> ?min-p -> ?xtc -> temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 0.900, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 0.800 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 0 | processing task, is_child = 0 slot update_slots: id 0 | task 0 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 34 slot update_slots: id 0 | task 0 | cached n_tokens = 0, memory_seq_rm [0, end) slot init_sampler: id 0 | task 0 | init sampler, took 0.01 ms, tokens: text = 34, total = 34 srv log_server_r: done request: POST /completion 127.0.0.1 200 slot print_timing: id 0 | task 0 | prompt eval time = 61.33 ms / 34 tokens ( 1.80 ms per token, 554.37 tokens per second) slot print_timing: id 0 | task 0 | eval time = 639.30 ms / 66 tokens ( 9.69 ms per token, 103.24 tokens per second) slot print_timing: id 0 | task 0 | total time = 700.63 ms / 100 tokens slot print_timing: id 0 | task 0 | graphs reused = 65 slot release: id 0 | task 0 | stop processing: n_tokens = 99, truncated = 0 srv update_slots: all slots are idle srv log_server_r: done request: POST /tokenize 127.0.0.1 200 [GIN] 2026/06/10 - 10:39:57 | 200 | 2.435702375s | 127.0.0.1 | POST "/api/generate" [GIN] 2026/06/10 - 11:53:47 | 200 | 61µs | 127.0.0.1 | HEAD "/" [GIN] 2026/06/10 - 11:53:47 | 200 | 2.595166ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 11:53:49 | 200 | 65.75µs | 127.0.0.1 | HEAD "/" time=2026-06-10T11:54:07.504+08:00 level=INFO source=download.go:179 msg="downloading c5396e06af29 in 4 100 MB part(s)" time=2026-06-10T11:55:17.130+08:00 level=INFO source=download.go:179 msg="downloading 005f95c74751 in 1 490 B part(s)" [GIN] 2026/06/10 - 11:55:18 | 200 | 1m29s | 127.0.0.1 | POST "/api/pull" [GIN] 2026/06/10 - 11:55:23 | 200 | 64.542µs | 127.0.0.1 | HEAD "/" [GIN] 2026/06/10 - 11:55:23 | 200 | 111.321958ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 11:55:26 | 200 | 130.792µs | 127.0.0.1 | HEAD "/" [GIN] 2026/06/10 - 11:55:26 | 200 | 2.505959ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:15:07 | 200 | 1.683786708s | 127.0.0.1 | POST "/api/pull" [GIN] 2026/06/10 - 12:15:09 | 200 | 1.920583ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:16:00 | 200 | 27.708µs | 127.0.0.1 | HEAD "/" [GIN] 2026/06/10 - 12:16:00 | 200 | 683.125µs | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:16:39 | 200 | 66.75µs | 127.0.0.1 | HEAD "/" [GIN] 2026/06/10 - 12:16:39 | 200 | 3.4835ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:16:39 | 200 | 38.16775ms | 127.0.0.1 | POST "/api/generate" [GIN] 2026/06/10 - 12:16:39 | 200 | 8.560792ms | 127.0.0.1 | DELETE "/api/delete" time=2026-06-10T12:19:44.723+08:00 level=INFO source=sched.go:1148 msg="disabling mmap for llama-server load by default" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 reason=cpu time=2026-06-10T12:19:44.724+08:00 level=INFO source=server.go:109 msg="using llama-server for model" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T12:19:44.725+08:00 level=INFO source=llama_server.go:403 msg="starting llama-server" cmd="/Applications/Ollama.app/Contents/Resources/llama-server --model /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 --port 51256 --host 127.0.0.1 --no-webui --offline -c 4096 -np 1 --log-verbosity 4 --no-log-prefix --no-log-timestamps --no-jinja --chat-template chatml --no-mmap --flash-attn auto -b 512 -ub 512 --context-shift --keep 4" time=2026-06-10T12:19:44.727+08:00 level=INFO source=sched.go:613 msg="system memory" total="16.0 GiB" free="4.5 GiB" free_swap="0 B" time=2026-06-10T12:19:44.727+08:00 level=INFO source=llama_server.go:886 msg="loading model via llama-server" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T12:19:44.727+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T12:19:44.727+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server not responding" common_params_print_info: build 1 (6f3a9f3de) with AppleClang 21.0.0.21000099 for Darwin arm64 log_info: verbosity = 4 (adjust with the `-lv N` CLI arg) device_info: - MTL0 : Apple M2 Pro (12124 MiB, 12123 MiB free) - BLAS : Accelerate (0 MiB, 0 MiB free) - CPU : Apple M2 Pro (16384 MiB, 16384 MiB free) system_info: n_threads = 6 (n_threads_batch = 6) / 10 | MTL : EMBED_LIBRARY = 1 | CPU : NEON = 1 | ARM_FMA = 1 | FP16_VA = 1 | DOTPROD = 1 | LLAMAFILE = 1 | ACCELERATE = 1 | REPACK = 1 | srv init: using 9 threads for HTTP server srv init: The UI is disabled srv init: Use --ui/--no-ui (or deprecated --webui/--no-webui) to enable/disable srv start: binding port with default address family srv llama_server: loading model srv load_model: loading model '/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4' common_init_result: fitting params to device memory ... common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on) common_params_fit_impl: getting device memory data for initial parameters: common_memory_breakdown_print: | memory breakdown [MiB] | total free self model context compute unaccounted | common_memory_breakdown_print: | - MTL0 (Apple M2 Pro) | 12124 = 12123 + (1109 = 934 + 112 + 62) + -1108 | common_memory_breakdown_print: | - Host | 202 = 182 + 0 + 20 | common_params_fit_impl: projected to use 1109 MiB of device memory vs. 12123 MiB of free device memory common_params_fit_impl: will leave 11014 >= 1024 MiB of free device memory, no changes needed common_fit_params: successfully fit params to free device memory common_fit_params: fitting params to free memory took 0.15 seconds llama_model_loader: loaded meta data with 34 key-value pairs and 338 tensors from /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = qwen2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen2.5 1.5B Instruct llama_model_loader: - kv 3: general.finetune str = Instruct llama_model_loader: - kv 4: general.basename str = Qwen2.5 llama_model_loader: - kv 5: general.size_label str = 1.5B llama_model_loader: - kv 6: general.license str = apache-2.0 llama_model_loader: - kv 7: general.license.link str = https://huggingface.co/Qwen/Qwen2.5-1... llama_model_loader: - kv 8: general.base_model.count u32 = 1 llama_model_loader: - kv 9: general.base_model.0.name str = Qwen2.5 1.5B llama_model_loader: - kv 10: general.base_model.0.organization str = Qwen llama_model_loader: - kv 11: general.base_model.0.repo_url str = https://huggingface.co/Qwen/Qwen2.5-1.5B llama_model_loader: - kv 12: general.tags arr[str,2] = ["chat", "text-generation"] llama_model_loader: - kv 13: general.languages arr[str,1] = ["en"] llama_model_loader: - kv 14: qwen2.block_count u32 = 28 llama_model_loader: - kv 15: qwen2.context_length u32 = 32768 llama_model_loader: - kv 16: qwen2.embedding_length u32 = 1536 llama_model_loader: - kv 17: qwen2.feed_forward_length u32 = 8960 llama_model_loader: - kv 18: qwen2.attention.head_count u32 = 12 llama_model_loader: - kv 19: qwen2.attention.head_count_kv u32 = 2 llama_model_loader: - kv 20: qwen2.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 21: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 22: general.file_type u32 = 15 llama_model_loader: - kv 23: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 24: tokenizer.ggml.pre str = qwen2 time=2026-06-10T12:19:44.981+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server loading model" llama_model_loader: - kv 25: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 26: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 27: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 28: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 29: tokenizer.ggml.padding_token_id u32 = 151643 llama_model_loader: - kv 30: tokenizer.ggml.bos_token_id u32 = 151643 llama_model_loader: - kv 31: tokenizer.ggml.add_bos_token bool = false llama_model_loader: - kv 32: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>... llama_model_loader: - kv 33: general.quantization_version u32 = 2 llama_model_loader: - type f32: 141 tensors llama_model_loader: - type q4_K: 168 tensors llama_model_loader: - type q6_K: 29 tensors print_info: file format = GGUF V3 (latest) print_info: file type = Q4_K - Medium print_info: file size = 934.69 MiB (5.08 BPW) llama_prepare_model_devices: using device MTL0 (Apple M2 Pro) (unknown id) - 12123 MiB free load: 0 unused tokens load: control-looking token: 128247 '' was not control-type; this is probably a bug in the model. its type will be overridden load: printing all EOG tokens: load: - 128247 ('') load: - 151643 ('<|endoftext|>') load: - 151645 ('<|im_end|>') load: - 151662 ('<|fim_pad|>') load: - 151663 ('<|repo_name|>') load: - 151664 ('<|file_sep|>') load: special tokens cache size = 23 load: token to piece cache size = 0.9310 MB print_info: arch = qwen2 print_info: vocab_only = 0 print_info: no_alloc = 0 print_info: n_ctx_train = 32768 print_info: n_embd = 1536 print_info: n_embd_inp = 1536 print_info: n_layer = 28 print_info: n_head = 12 print_info: n_head_kv = 2 print_info: n_rot = 128 print_info: n_swa = 0 print_info: is_swa_any = 0 print_info: n_embd_head_k = 128 print_info: n_embd_head_v = 128 print_info: n_gqa = 6 print_info: n_embd_k_gqa = 256 print_info: n_embd_v_gqa = 256 print_info: f_norm_eps = 0.0e+00 print_info: f_norm_rms_eps = 1.0e-06 print_info: f_clamp_kqv = 0.0e+00 print_info: f_max_alibi_bias = 0.0e+00 print_info: f_logit_scale = 0.0e+00 print_info: f_attn_scale = 0.0e+00 print_info: f_attn_value_scale = 0.0000 print_info: n_ff = 8960 print_info: n_expert = 0 print_info: n_expert_used = 0 print_info: n_expert_groups = 0 print_info: n_group_used = 0 print_info: causal attn = 1 print_info: pooling type = -1 print_info: rope type = 2 print_info: rope scaling = linear print_info: freq_base_train = 1000000.0 print_info: freq_scale_train = 1 print_info: n_ctx_orig_yarn = 32768 print_info: rope_yarn_log_mul = 0.0000 print_info: rope_finetuned = unknown print_info: model type = 1.5B print_info: model params = 1.54 B print_info: general.name = Qwen2.5 1.5B Instruct print_info: vocab type = BPE print_info: n_vocab = 151936 print_info: n_merges = 151387 print_info: BOS token = 151643 '<|endoftext|>' print_info: EOS token = 151645 '<|im_end|>' print_info: EOT token = 151645 '<|im_end|>' print_info: PAD token = 151643 '<|endoftext|>' print_info: LF token = 198 'Ċ' print_info: FIM PRE token = 151659 '<|fim_prefix|>' print_info: FIM SUF token = 151661 '<|fim_suffix|>' print_info: FIM MID token = 151660 '<|fim_middle|>' print_info: FIM PAD token = 151662 '<|fim_pad|>' print_info: FIM REP token = 151663 '<|repo_name|>' print_info: FIM SEP token = 151664 '<|file_sep|>' print_info: EOG token = 128247 '' print_info: EOG token = 151643 '<|endoftext|>' print_info: EOG token = 151645 '<|im_end|>' print_info: EOG token = 151662 '<|fim_pad|>' print_info: EOG token = 151663 '<|repo_name|>' print_info: EOG token = 151664 '<|file_sep|>' print_info: max token length = 256 load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false) load_tensors: offloading output layer to GPU load_tensors: offloading 27 repeating layers to GPU load_tensors: offloaded 29/29 layers to GPU load_tensors: CPU model buffer size = 182.57 MiB load_tensors: MTL0 model buffer size = 934.69 MiB ...................................................................... common_init_result: added logit bias = -inf common_init_result: added <|endoftext|> logit bias = -inf common_init_result: added <|im_end|> logit bias = -inf common_init_result: added <|fim_pad|> logit bias = -inf common_init_result: added <|repo_name|> logit bias = -inf common_init_result: added <|file_sep|> logit bias = -inf llama_context: constructing llama_context llama_context: n_seq_max = 1 llama_context: n_ctx = 4096 llama_context: n_ctx_seq = 4096 llama_context: n_batch = 512 llama_context: n_ubatch = 512 llama_context: causal_attn = 1 llama_context: flash_attn = auto llama_context: kv_unified = false llama_context: freq_base = 1000000.0 llama_context: freq_scale = 1 llama_context: n_rs_seq = 0 llama_context: n_outputs_max = 1 llama_context: n_ctx_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized ggml_metal_init: allocating ggml_metal_init: found device: Apple M2 Pro ggml_metal_init: picking default device: Apple M2 Pro ggml_metal_init: use fusion = true ggml_metal_init: use concurrency = true ggml_metal_init: use graph optimize = true llama_context: CPU output buffer size = 0.58 MiB llama_kv_cache: MTL0 KV buffer size = 112.00 MiB llama_kv_cache: size = 112.00 MiB ( 4096 cells, 28 layers, 1/1 seqs), K (f16): 56.00 MiB, V (f16): 56.00 MiB llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 128 llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 128 sched_reserve: reserving ... sched_reserve: Flash Attention was auto, set to enabled sched_reserve: resolving fused Gated Delta Net support: sched_reserve: fused Gated Delta Net (autoregressive) enabled sched_reserve: fused Gated Delta Net (chunked) enabled sched_reserve: MTL0 compute buffer size = 62.51 MiB sched_reserve: CPU compute buffer size = 10.01 MiB sched_reserve: graph nodes = 958 sched_reserve: graph splits = 2 sched_reserve: reserve took 10.97 ms, sched copies = 1 common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable) srv load_model: initializing slots, n_slots = 1 common_speculative_init: no implementations specified for speculative decoding slot load_model: id 0 | task -1 | new slot, n_ctx = 4096 srv load_model: prompt cache is enabled, size limit: 8192 MiB srv load_model: use `--cache-ram 0` to disable the prompt cache srv load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391 srv load_model: context checkpoints enabled, max = 32, min spacing = 256 srv init: --cache-idle-slots requires --kv-unified, disabling init: chat template, example_format: '<|im_start|>system You are a helpful assistant<|im_end|> <|im_start|>user Hello<|im_end|> <|im_start|>assistant Hi there<|im_end|> <|im_start|>user How are you?<|im_end|> <|im_start|>assistant ' srv init: init: chat template, thinking = 0 srv llama_server: model loaded srv llama_server: server is listening on http://127.0.0.1:51256 srv update_slots: all slots are idle time=2026-06-10T12:19:45.988+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 1.26 seconds" time=2026-06-10T12:19:46.017+08:00 level=INFO source=images.go:354 msg="template selection" model=registry.ollama.ai/library/qwen2.5:1.5b selected=go_template renderer="" parser="" go_template="[completion tools]" chat_template="[tools completion]" harmony=null renderer_parser=null time=2026-06-10T12:19:46.017+08:00 level=INFO source=sched.go:729 msg="loaded runners" count=1 time=2026-06-10T12:19:46.017+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T12:19:46.018+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 1.29 seconds" srv log_server_r: done request: POST /tokenize 127.0.0.1 200 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = -1 srv get_availabl: updating prompt cache srv load: - looking for better prompt, base f_keep = -1.000, sim = 0.000 srv update: - cache state: 0 prompts, 0.000 MiB (limits: 8192.000 MiB, 4096 tokens, 8589934592 est) srv get_availabl: prompt cache update took 0.00 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> top-p -> ?min-p -> ?xtc -> temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 0.900, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 0.800 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 0 | processing task, is_child = 0 slot update_slots: id 0 | task 0 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 35 slot update_slots: id 0 | task 0 | cached n_tokens = 0, memory_seq_rm [0, end) slot init_sampler: id 0 | task 0 | init sampler, took 0.01 ms, tokens: text = 35, total = 35 srv log_server_r: done request: POST /completion 127.0.0.1 200 slot print_timing: id 0 | task 0 | prompt eval time = 67.22 ms / 35 tokens ( 1.92 ms per token, 520.66 tokens per second) slot print_timing: id 0 | task 0 | eval time = 433.39 ms / 44 tokens ( 9.85 ms per token, 101.52 tokens per second) slot print_timing: id 0 | task 0 | total time = 500.61 ms / 79 tokens slot print_timing: id 0 | task 0 | graphs reused = 43 slot release: id 0 | task 0 | stop processing: n_tokens = 78, truncated = 0 srv update_slots: all slots are idle srv log_server_r: done request: POST /tokenize 127.0.0.1 200 [GIN] 2026/06/10 - 12:19:46 | 200 | 1.97899175s | 127.0.0.1 | POST "/api/generate" time=2026-06-10T12:38:12.978+08:00 level=INFO source=sched.go:1148 msg="disabling mmap for llama-server load by default" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 reason=cpu time=2026-06-10T12:38:12.979+08:00 level=INFO source=server.go:109 msg="using llama-server for model" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T12:38:12.980+08:00 level=INFO source=llama_server.go:403 msg="starting llama-server" cmd="/Applications/Ollama.app/Contents/Resources/llama-server --model /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 --port 51633 --host 127.0.0.1 --no-webui --offline -c 4096 -np 1 --log-verbosity 4 --no-log-prefix --no-log-timestamps --no-jinja --chat-template chatml --no-mmap --flash-attn auto -b 512 -ub 512 --context-shift --keep 4" time=2026-06-10T12:38:12.982+08:00 level=INFO source=sched.go:613 msg="system memory" total="16.0 GiB" free="5.5 GiB" free_swap="0 B" time=2026-06-10T12:38:12.982+08:00 level=INFO source=llama_server.go:886 msg="loading model via llama-server" model=/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 time=2026-06-10T12:38:12.982+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T12:38:12.982+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server not responding" common_params_print_info: build 1 (6f3a9f3de) with AppleClang 21.0.0.21000099 for Darwin arm64 log_info: verbosity = 4 (adjust with the `-lv N` CLI arg) device_info: - MTL0 : Apple M2 Pro (12124 MiB, 12123 MiB free) - BLAS : Accelerate (0 MiB, 0 MiB free) - CPU : Apple M2 Pro (16384 MiB, 16384 MiB free) system_info: n_threads = 6 (n_threads_batch = 6) / 10 | MTL : EMBED_LIBRARY = 1 | CPU : NEON = 1 | ARM_FMA = 1 | FP16_VA = 1 | DOTPROD = 1 | LLAMAFILE = 1 | ACCELERATE = 1 | REPACK = 1 | srv init: using 9 threads for HTTP server srv init: The UI is disabled srv init: Use --ui/--no-ui (or deprecated --webui/--no-webui) to enable/disable srv start: binding port with default address family srv llama_server: loading model srv load_model: loading model '/Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4' common_init_result: fitting params to device memory ... common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on) common_params_fit_impl: getting device memory data for initial parameters: common_memory_breakdown_print: | memory breakdown [MiB] | total free self model context compute unaccounted | common_memory_breakdown_print: | - MTL0 (Apple M2 Pro) | 12124 = 12123 + (1109 = 934 + 112 + 62) + -1108 | common_memory_breakdown_print: | - Host | 202 = 182 + 0 + 20 | common_params_fit_impl: projected to use 1109 MiB of device memory vs. 12123 MiB of free device memory common_params_fit_impl: will leave 11014 >= 1024 MiB of free device memory, no changes needed common_fit_params: successfully fit params to free device memory common_fit_params: fitting params to free memory took 0.14 seconds llama_model_loader: loaded meta data with 34 key-value pairs and 338 tensors from /Users/admin/.ollama/models/blobs/sha256-183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = qwen2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen2.5 1.5B Instruct llama_model_loader: - kv 3: general.finetune str = Instruct llama_model_loader: - kv 4: general.basename str = Qwen2.5 llama_model_loader: - kv 5: general.size_label str = 1.5B llama_model_loader: - kv 6: general.license str = apache-2.0 llama_model_loader: - kv 7: general.license.link str = https://huggingface.co/Qwen/Qwen2.5-1... llama_model_loader: - kv 8: general.base_model.count u32 = 1 llama_model_loader: - kv 9: general.base_model.0.name str = Qwen2.5 1.5B llama_model_loader: - kv 10: general.base_model.0.organization str = Qwen llama_model_loader: - kv 11: general.base_model.0.repo_url str = https://huggingface.co/Qwen/Qwen2.5-1.5B llama_model_loader: - kv 12: general.tags arr[str,2] = ["chat", "text-generation"] llama_model_loader: - kv 13: general.languages arr[str,1] = ["en"] llama_model_loader: - kv 14: qwen2.block_count u32 = 28 llama_model_loader: - kv 15: qwen2.context_length u32 = 32768 llama_model_loader: - kv 16: qwen2.embedding_length u32 = 1536 llama_model_loader: - kv 17: qwen2.feed_forward_length u32 = 8960 llama_model_loader: - kv 18: qwen2.attention.head_count u32 = 12 llama_model_loader: - kv 19: qwen2.attention.head_count_kv u32 = 2 llama_model_loader: - kv 20: qwen2.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 21: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 22: general.file_type u32 = 15 llama_model_loader: - kv 23: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 24: tokenizer.ggml.pre str = qwen2 llama_model_loader: - kv 25: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 26: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 27: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 28: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 29: tokenizer.ggml.padding_token_id u32 = 151643 llama_model_loader: - kv 30: tokenizer.ggml.bos_token_id u32 = 151643 llama_model_loader: - kv 31: tokenizer.ggml.add_bos_token bool = false llama_model_loader: - kv 32: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>... llama_model_loader: - kv 33: general.quantization_version u32 = 2 llama_model_loader: - type f32: 141 tensors llama_model_loader: - type q4_K: 168 tensors llama_model_loader: - type q6_K: 29 tensors print_info: file format = GGUF V3 (latest) print_info: file type = Q4_K - Medium print_info: file size = 934.69 MiB (5.08 BPW) llama_prepare_model_devices: using device MTL0 (Apple M2 Pro) (unknown id) - 12123 MiB free time=2026-06-10T12:38:13.233+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server loading model" load: 0 unused tokens load: control-looking token: 128247 '' was not control-type; this is probably a bug in the model. its type will be overridden load: printing all EOG tokens: load: - 128247 ('') load: - 151643 ('<|endoftext|>') load: - 151645 ('<|im_end|>') load: - 151662 ('<|fim_pad|>') load: - 151663 ('<|repo_name|>') load: - 151664 ('<|file_sep|>') load: special tokens cache size = 23 load: token to piece cache size = 0.9310 MB print_info: arch = qwen2 print_info: vocab_only = 0 print_info: no_alloc = 0 print_info: n_ctx_train = 32768 print_info: n_embd = 1536 print_info: n_embd_inp = 1536 print_info: n_layer = 28 print_info: n_head = 12 print_info: n_head_kv = 2 print_info: n_rot = 128 print_info: n_swa = 0 print_info: is_swa_any = 0 print_info: n_embd_head_k = 128 print_info: n_embd_head_v = 128 print_info: n_gqa = 6 print_info: n_embd_k_gqa = 256 print_info: n_embd_v_gqa = 256 print_info: f_norm_eps = 0.0e+00 print_info: f_norm_rms_eps = 1.0e-06 print_info: f_clamp_kqv = 0.0e+00 print_info: f_max_alibi_bias = 0.0e+00 print_info: f_logit_scale = 0.0e+00 print_info: f_attn_scale = 0.0e+00 print_info: f_attn_value_scale = 0.0000 print_info: n_ff = 8960 print_info: n_expert = 0 print_info: n_expert_used = 0 print_info: n_expert_groups = 0 print_info: n_group_used = 0 print_info: causal attn = 1 print_info: pooling type = -1 print_info: rope type = 2 print_info: rope scaling = linear print_info: freq_base_train = 1000000.0 print_info: freq_scale_train = 1 print_info: n_ctx_orig_yarn = 32768 print_info: rope_yarn_log_mul = 0.0000 print_info: rope_finetuned = unknown print_info: model type = 1.5B print_info: model params = 1.54 B print_info: general.name = Qwen2.5 1.5B Instruct print_info: vocab type = BPE print_info: n_vocab = 151936 print_info: n_merges = 151387 print_info: BOS token = 151643 '<|endoftext|>' print_info: EOS token = 151645 '<|im_end|>' print_info: EOT token = 151645 '<|im_end|>' print_info: PAD token = 151643 '<|endoftext|>' print_info: LF token = 198 'Ċ' print_info: FIM PRE token = 151659 '<|fim_prefix|>' print_info: FIM SUF token = 151661 '<|fim_suffix|>' print_info: FIM MID token = 151660 '<|fim_middle|>' print_info: FIM PAD token = 151662 '<|fim_pad|>' print_info: FIM REP token = 151663 '<|repo_name|>' print_info: FIM SEP token = 151664 '<|file_sep|>' print_info: EOG token = 128247 '' print_info: EOG token = 151643 '<|endoftext|>' print_info: EOG token = 151645 '<|im_end|>' print_info: EOG token = 151662 '<|fim_pad|>' print_info: EOG token = 151663 '<|repo_name|>' print_info: EOG token = 151664 '<|file_sep|>' print_info: max token length = 256 load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false) load_tensors: offloading output layer to GPU load_tensors: offloading 27 repeating layers to GPU load_tensors: offloaded 29/29 layers to GPU load_tensors: CPU model buffer size = 182.57 MiB load_tensors: MTL0 model buffer size = 934.69 MiB ...................................................................... common_init_result: added logit bias = -inf common_init_result: added <|endoftext|> logit bias = -inf common_init_result: added <|im_end|> logit bias = -inf common_init_result: added <|fim_pad|> logit bias = -inf common_init_result: added <|repo_name|> logit bias = -inf common_init_result: added <|file_sep|> logit bias = -inf llama_context: constructing llama_context llama_context: n_seq_max = 1 llama_context: n_ctx = 4096 llama_context: n_ctx_seq = 4096 llama_context: n_batch = 512 llama_context: n_ubatch = 512 llama_context: causal_attn = 1 llama_context: flash_attn = auto llama_context: kv_unified = false llama_context: freq_base = 1000000.0 llama_context: freq_scale = 1 llama_context: n_rs_seq = 0 llama_context: n_outputs_max = 1 llama_context: n_ctx_seq (4096) < n_ctx_train (32768) -- the full capacity of the model will not be utilized ggml_metal_init: allocating ggml_metal_init: found device: Apple M2 Pro ggml_metal_init: picking default device: Apple M2 Pro ggml_metal_init: use fusion = true ggml_metal_init: use concurrency = true ggml_metal_init: use graph optimize = true llama_context: CPU output buffer size = 0.58 MiB llama_kv_cache: MTL0 KV buffer size = 112.00 MiB llama_kv_cache: size = 112.00 MiB ( 4096 cells, 28 layers, 1/1 seqs), K (f16): 56.00 MiB, V (f16): 56.00 MiB llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 128 llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 128 sched_reserve: reserving ... sched_reserve: Flash Attention was auto, set to enabled sched_reserve: resolving fused Gated Delta Net support: sched_reserve: fused Gated Delta Net (autoregressive) enabled sched_reserve: fused Gated Delta Net (chunked) enabled sched_reserve: MTL0 compute buffer size = 62.51 MiB sched_reserve: CPU compute buffer size = 10.01 MiB sched_reserve: graph nodes = 958 sched_reserve: graph splits = 2 sched_reserve: reserve took 52.13 ms, sched copies = 1 common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable) srv load_model: initializing slots, n_slots = 1 common_speculative_init: no implementations specified for speculative decoding slot load_model: id 0 | task -1 | new slot, n_ctx = 4096 srv load_model: prompt cache is enabled, size limit: 8192 MiB srv load_model: use `--cache-ram 0` to disable the prompt cache srv load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391 srv load_model: context checkpoints enabled, max = 32, min spacing = 256 srv init: --cache-idle-slots requires --kv-unified, disabling init: chat template, example_format: '<|im_start|>system You are a helpful assistant<|im_end|> <|im_start|>user Hello<|im_end|> <|im_start|>assistant Hi there<|im_end|> <|im_start|>user How are you?<|im_end|> <|im_start|>assistant ' srv init: init: chat template, thinking = 0 srv llama_server: model loaded srv llama_server: server is listening on http://127.0.0.1:51633 srv update_slots: all slots are idle time=2026-06-10T12:38:13.986+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 1.00 seconds" time=2026-06-10T12:38:14.008+08:00 level=INFO source=images.go:354 msg="template selection" model=registry.ollama.ai/library/qwen2.5:1.5b selected=go_template renderer="" parser="" go_template="[completion tools]" chat_template="[tools completion]" harmony=null renderer_parser=null time=2026-06-10T12:38:14.008+08:00 level=INFO source=sched.go:729 msg="loaded runners" count=1 time=2026-06-10T12:38:14.008+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T12:38:14.008+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 1.03 seconds" srv log_server_r: done request: POST /tokenize 127.0.0.1 200 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = -1 srv get_availabl: updating prompt cache srv load: - looking for better prompt, base f_keep = -1.000, sim = 0.000 srv update: - cache state: 0 prompts, 0.000 MiB (limits: 8192.000 MiB, 4096 tokens, 8589934592 est) srv get_availabl: prompt cache update took 0.00 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> top-p -> ?min-p -> ?xtc -> temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 0.900, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 0.800 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 0 | processing task, is_child = 0 slot update_slots: id 0 | task 0 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 30 slot update_slots: id 0 | task 0 | cached n_tokens = 0, memory_seq_rm [0, end) slot init_sampler: id 0 | task 0 | init sampler, took 0.01 ms, tokens: text = 30, total = 30 srv log_server_r: done request: POST /completion 127.0.0.1 200 slot print_timing: id 0 | task 0 | prompt eval time = 39.89 ms / 30 tokens ( 1.33 ms per token, 752.11 tokens per second) slot print_timing: id 0 | task 0 | eval time = 90.07 ms / 10 tokens ( 9.01 ms per token, 111.03 tokens per second) slot print_timing: id 0 | task 0 | total time = 129.96 ms / 40 tokens slot print_timing: id 0 | task 0 | graphs reused = 9 slot release: id 0 | task 0 | stop processing: n_tokens = 39, truncated = 0 srv update_slots: all slots are idle srv log_server_r: done request: POST /tokenize 127.0.0.1 200 [GIN] 2026/06/10 - 12:38:14 | 200 | 1.341979958s | 127.0.0.1 | POST "/api/generate" srv log_server_r: done request: POST /tokenize 127.0.0.1 200 slot get_availabl: id 0 | task -1 | selected slot by LCP similarity, sim_best = 1.000 (> 0.100 thold), f_keep = 0.769 slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> ?top-p -> ?min-p -> ?xtc -> ?temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 1.000, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 1.000 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 11 | processing task, is_child = 0 slot update_slots: id 0 | task 11 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 30 slot update_slots: id 0 | task 11 | need to evaluate at least 1 token for each active slot (n_past = 30, task.n_tokens() = 30) slot update_slots: id 0 | task 11 | n_past was set to 29 slot update_slots: id 0 | task 11 | cached n_tokens = 29, memory_seq_rm [29, end) slot init_sampler: id 0 | task 11 | init sampler, took 0.01 ms, tokens: text = 30, total = 30 srv log_server_r: done request: POST /completion 127.0.0.1 200 slot print_timing: id 0 | task 11 | prompt eval time = 20.70 ms / 1 tokens ( 20.70 ms per token, 48.30 tokens per second) slot print_timing: id 0 | task 11 | eval time = 86.36 ms / 10 tokens ( 8.64 ms per token, 115.80 tokens per second) slot print_timing: id 0 | task 11 | total time = 107.06 ms / 11 tokens slot print_timing: id 0 | task 11 | graphs reused = 19 slot release: id 0 | task 11 | stop processing: n_tokens = 39, truncated = 0 srv update_slots: all slots are idle [GIN] 2026/06/10 - 12:38:41 | 200 | 244.639166ms | 127.0.0.1 | POST "/v1/chat/completions" [GIN] 2026/06/10 - 12:40:08 | 404 | 15.75µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:40:08 | 200 | 2.840166ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:40:08 | 200 | 1.672375ms | 127.0.0.1 | GET "/v1/models" [GIN] 2026/06/10 - 12:40:08 | 404 | 1.299292ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:40:09 | 404 | 12.083µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:40:09 | 200 | 909.334µs | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:40:09 | 404 | 2.896334ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:40:09 | 404 | 11.083µs | 127.0.0.1 | GET "/v1/models/local-qwen/qwen2.5:1.5b" [GIN] 2026/06/10 - 12:40:09 | 200 | 1.569792ms | 127.0.0.1 | GET "/v1/models" [GIN] 2026/06/10 - 12:40:35 | 404 | 2.636958ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:40:35 | 404 | 10.667µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:40:36 | 200 | 1.565ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:40:36 | 404 | 2.33825ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:40:36 | 404 | 10.917µs | 127.0.0.1 | GET "/v1/models/local-qwen/qwen2.5:1.5b" [GIN] 2026/06/10 - 12:40:36 | 200 | 1.410417ms | 127.0.0.1 | GET "/v1/models" [GIN] 2026/06/10 - 12:40:54 | 404 | 1.71525ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:40:54 | 404 | 10.75µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:40:54 | 200 | 1.544209ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:40:54 | 404 | 2.0125ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:40:54 | 404 | 9.166µs | 127.0.0.1 | GET "/v1/models/local-qwen/qwen2.5:1.5b" [GIN] 2026/06/10 - 12:40:54 | 200 | 1.55825ms | 127.0.0.1 | GET "/v1/models" [GIN] 2026/06/10 - 12:44:32 | 200 | 2.366708ms | 127.0.0.1 | GET "/v1/models" [GIN] 2026/06/10 - 12:44:32 | 200 | 1.903875ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:47:11 | 200 | 1.279209ms | 127.0.0.1 | GET "/v1/models" [GIN] 2026/06/10 - 12:47:19 | 200 | 1.996625ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:48:06 | 200 | 1.767375ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:51:26 | 200 | 1.643625ms | 127.0.0.1 | GET "/v1/models" [GIN] 2026/06/10 - 12:51:33 | 404 | 10.083µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:51:33 | 200 | 676.834µs | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:51:33 | 200 | 1.838458ms | 127.0.0.1 | GET "/v1/models" [GIN] 2026/06/10 - 12:51:33 | 200 | 113.000417ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:51:34 | 404 | 6.291µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:51:34 | 200 | 795.334µs | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:51:34 | 200 | 2.523166ms | 127.0.0.1 | POST "/api/show" time=2026-06-10T12:51:35.574+08:00 level=INFO source=sched.go:1148 msg="disabling mmap for llama-server load by default" model=/Users/admin/.ollama/models/blobs/sha256-96c415656d377afbff962f6cdb2394ab092ccbcbaab4b82525bc4ca800fe8a49 reason=cpu time=2026-06-10T12:51:35.574+08:00 level=INFO source=server.go:109 msg="using llama-server for model" model=/Users/admin/.ollama/models/blobs/sha256-96c415656d377afbff962f6cdb2394ab092ccbcbaab4b82525bc4ca800fe8a49 time=2026-06-10T12:51:35.574+08:00 level=INFO source=llama_server.go:403 msg="starting llama-server" cmd="/Applications/Ollama.app/Contents/Resources/llama-server --model /Users/admin/.ollama/models/blobs/sha256-96c415656d377afbff962f6cdb2394ab092ccbcbaab4b82525bc4ca800fe8a49 --port 51990 --host 127.0.0.1 --no-webui --offline -c 4096 -np 1 --log-verbosity 4 --no-log-prefix --no-log-timestamps --no-mmap --flash-attn auto -b 512 -ub 512 --context-shift --keep 4" time=2026-06-10T12:51:35.576+08:00 level=INFO source=sched.go:613 msg="system memory" total="16.0 GiB" free="5.2 GiB" free_swap="0 B" time=2026-06-10T12:51:35.576+08:00 level=INFO source=llama_server.go:886 msg="loading model via llama-server" model=/Users/admin/.ollama/models/blobs/sha256-96c415656d377afbff962f6cdb2394ab092ccbcbaab4b82525bc4ca800fe8a49 time=2026-06-10T12:51:35.576+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T12:51:35.576+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server not responding" common_params_print_info: build 1 (6f3a9f3de) with AppleClang 21.0.0.21000099 for Darwin arm64 log_info: verbosity = 4 (adjust with the `-lv N` CLI arg) device_info: - MTL0 : Apple M2 Pro (12124 MiB, 12123 MiB free) - BLAS : Accelerate (0 MiB, 0 MiB free) - CPU : Apple M2 Pro (16384 MiB, 16384 MiB free) system_info: n_threads = 6 (n_threads_batch = 6) / 10 | MTL : EMBED_LIBRARY = 1 | CPU : NEON = 1 | ARM_FMA = 1 | FP16_VA = 1 | DOTPROD = 1 | LLAMAFILE = 1 | ACCELERATE = 1 | REPACK = 1 | srv init: using 9 threads for HTTP server srv init: The UI is disabled srv init: Use --ui/--no-ui (or deprecated --webui/--no-webui) to enable/disable srv start: binding port with default address family srv llama_server: loading model srv load_model: loading model '/Users/admin/.ollama/models/blobs/sha256-96c415656d377afbff962f6cdb2394ab092ccbcbaab4b82525bc4ca800fe8a49' common_init_result: fitting params to device memory ... common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on) common_params_fit_impl: getting device memory data for initial parameters: common_memory_breakdown_print: | memory breakdown [MiB] | total free self model context compute unaccounted | common_memory_breakdown_print: | - MTL0 (Apple M2 Pro) | 12124 = 12123 + (4521 = 4168 + 224 + 129) + -4520 | common_memory_breakdown_print: | - Host | 328 = 292 + 0 + 36 | common_params_fit_impl: projected to use 4521 MiB of device memory vs. 12123 MiB of free device memory common_params_fit_impl: will leave 7602 >= 1024 MiB of free device memory, no changes needed common_fit_params: successfully fit params to free device memory common_fit_params: fitting params to free memory took 0.14 seconds llama_model_loader: loaded meta data with 26 key-value pairs and 339 tensors from /Users/admin/.ollama/models/blobs/sha256-96c415656d377afbff962f6cdb2394ab092ccbcbaab4b82525bc4ca800fe8a49 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = qwen2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = DeepSeek R1 Distill Qwen 7B llama_model_loader: - kv 3: general.basename str = DeepSeek-R1-Distill-Qwen llama_model_loader: - kv 4: general.size_label str = 7B llama_model_loader: - kv 5: qwen2.block_count u32 = 28 llama_model_loader: - kv 6: qwen2.context_length u32 = 131072 llama_model_loader: - kv 7: qwen2.embedding_length u32 = 3584 llama_model_loader: - kv 8: qwen2.feed_forward_length u32 = 18944 llama_model_loader: - kv 9: qwen2.attention.head_count u32 = 28 llama_model_loader: - kv 10: qwen2.attention.head_count_kv u32 = 4 llama_model_loader: - kv 11: qwen2.rope.freq_base f32 = 10000.000000 llama_model_loader: - kv 12: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 13: general.file_type u32 = 15 llama_model_loader: - kv 14: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 15: tokenizer.ggml.pre str = qwen2 llama_model_loader: - kv 16: tokenizer.ggml.tokens arr[str,152064] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 17: tokenizer.ggml.token_type arr[i32,152064] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 18: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 19: tokenizer.ggml.bos_token_id u32 = 151646 llama_model_loader: - kv 20: tokenizer.ggml.eos_token_id u32 = 151643 llama_model_loader: - kv 21: tokenizer.ggml.padding_token_id u32 = 151643 llama_model_loader: - kv 22: tokenizer.ggml.add_bos_token bool = true llama_model_loader: - kv 23: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 24: tokenizer.chat_template str = {% if not add_generation_prompt is de... llama_model_loader: - kv 25: general.quantization_version u32 = 2 llama_model_loader: - type f32: 141 tensors llama_model_loader: - type q4_K: 169 tensors llama_model_loader: - type q6_K: 29 tensors print_info: file format = GGUF V3 (latest) print_info: file type = Q4_K - Medium print_info: file size = 4.36 GiB (4.91 BPW) llama_prepare_model_devices: using device MTL0 (Apple M2 Pro) (unknown id) - 12123 MiB free time=2026-06-10T12:51:35.828+08:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server loading model" load: 0 unused tokens load: control-looking token: 128247 '' was not control-type; this is probably a bug in the model. its type will be overridden load: printing all EOG tokens: load: - 128247 ('') load: - 151643 ('<｜end▁of▁sentence｜>') load: - 151662 ('<|fim_pad|>') load: - 151663 ('<|repo_name|>') load: - 151664 ('<|file_sep|>') load: special tokens cache size = 23 load: token to piece cache size = 0.9310 MB print_info: arch = qwen2 print_info: vocab_only = 0 print_info: no_alloc = 0 print_info: n_ctx_train = 131072 print_info: n_embd = 3584 print_info: n_embd_inp = 3584 print_info: n_layer = 28 print_info: n_head = 28 print_info: n_head_kv = 4 print_info: n_rot = 128 print_info: n_swa = 0 print_info: is_swa_any = 0 print_info: n_embd_head_k = 128 print_info: n_embd_head_v = 128 print_info: n_gqa = 7 print_info: n_embd_k_gqa = 512 print_info: n_embd_v_gqa = 512 print_info: f_norm_eps = 0.0e+00 print_info: f_norm_rms_eps = 1.0e-06 print_info: f_clamp_kqv = 0.0e+00 print_info: f_max_alibi_bias = 0.0e+00 print_info: f_logit_scale = 0.0e+00 print_info: f_attn_scale = 0.0e+00 print_info: f_attn_value_scale = 0.0000 print_info: n_ff = 18944 print_info: n_expert = 0 print_info: n_expert_used = 0 print_info: n_expert_groups = 0 print_info: n_group_used = 0 print_info: causal attn = 1 print_info: pooling type = -1 print_info: rope type = 2 print_info: rope scaling = linear print_info: freq_base_train = 10000.0 print_info: freq_scale_train = 1 print_info: n_ctx_orig_yarn = 131072 print_info: rope_yarn_log_mul = 0.0000 print_info: rope_finetuned = unknown print_info: model type = 7B print_info: model params = 7.62 B print_info: general.name = DeepSeek R1 Distill Qwen 7B print_info: vocab type = BPE print_info: n_vocab = 152064 print_info: n_merges = 151387 print_info: BOS token = 151646 '<｜begin▁of▁sentence｜>' print_info: EOS token = 151643 '<｜end▁of▁sentence｜>' print_info: EOT token = 151643 '<｜end▁of▁sentence｜>' print_info: PAD token = 151643 '<｜end▁of▁sentence｜>' print_info: LF token = 198 'Ċ' print_info: FIM PRE token = 151659 '<|fim_prefix|>' print_info: FIM SUF token = 151661 '<|fim_suffix|>' print_info: FIM MID token = 151660 '<|fim_middle|>' print_info: FIM PAD token = 151662 '<|fim_pad|>' print_info: FIM REP token = 151663 '<|repo_name|>' print_info: FIM SEP token = 151664 '<|file_sep|>' print_info: EOG token = 128247 '' print_info: EOG token = 151643 '<｜end▁of▁sentence｜>' print_info: EOG token = 151662 '<|fim_pad|>' print_info: EOG token = 151663 '<|repo_name|>' print_info: EOG token = 151664 '<|file_sep|>' print_info: max token length = 256 load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false) load_tensors: offloading output layer to GPU load_tensors: offloading 27 repeating layers to GPU load_tensors: offloaded 29/29 layers to GPU load_tensors: CPU model buffer size = 292.36 MiB load_tensors: MTL0 model buffer size = 4168.09 MiB .................................................................................... common_init_result: added logit bias = -inf common_init_result: added <｜end▁of▁sentence｜> logit bias = -inf common_init_result: added <|fim_pad|> logit bias = -inf common_init_result: added <|repo_name|> logit bias = -inf common_init_result: added <|file_sep|> logit bias = -inf llama_context: constructing llama_context llama_context: n_seq_max = 1 llama_context: n_ctx = 4096 llama_context: n_ctx_seq = 4096 llama_context: n_batch = 512 llama_context: n_ubatch = 512 llama_context: causal_attn = 1 llama_context: flash_attn = auto llama_context: kv_unified = false llama_context: freq_base = 10000.0 llama_context: freq_scale = 1 llama_context: n_rs_seq = 0 llama_context: n_outputs_max = 1 llama_context: n_ctx_seq (4096) < n_ctx_train (131072) -- the full capacity of the model will not be utilized ggml_metal_init: allocating ggml_metal_init: found device: Apple M2 Pro ggml_metal_init: picking default device: Apple M2 Pro ggml_metal_init: use fusion = true ggml_metal_init: use concurrency = true ggml_metal_init: use graph optimize = true llama_context: CPU output buffer size = 0.58 MiB llama_kv_cache: MTL0 KV buffer size = 224.00 MiB llama_kv_cache: size = 224.00 MiB ( 4096 cells, 28 layers, 1/1 seqs), K (f16): 112.00 MiB, V (f16): 112.00 MiB llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 128 llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 128 sched_reserve: reserving ... sched_reserve: Flash Attention was auto, set to enabled sched_reserve: resolving fused Gated Delta Net support: sched_reserve: fused Gated Delta Net (autoregressive) enabled sched_reserve: fused Gated Delta Net (chunked) enabled sched_reserve: MTL0 compute buffer size = 129.01 MiB sched_reserve: CPU compute buffer size = 18.01 MiB sched_reserve: graph nodes = 958 sched_reserve: graph splits = 2 sched_reserve: reserve took 20.65 ms, sched copies = 1 common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable) srv load_model: initializing slots, n_slots = 1 common_speculative_init: no implementations specified for speculative decoding slot load_model: id 0 | task -1 | new slot, n_ctx = 4096 srv load_model: prompt cache is enabled, size limit: 8192 MiB srv load_model: use `--cache-ram 0` to disable the prompt cache srv load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391 srv load_model: context checkpoints enabled, max = 32, min spacing = 256 srv init: --cache-idle-slots requires --kv-unified, disabling init: chat template, example_format: 'You are a helpful assistant<｜User｜>Hello<｜Assistant｜>Hi there<｜end▁of▁sentence｜><｜User｜>How are you?<｜Assistant｜>' srv init: init: chat template, thinking = 1 srv llama_server: model loaded srv llama_server: server is listening on http://127.0.0.1:51990 srv update_slots: all slots are idle time=2026-06-10T12:51:38.859+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 3.28 seconds" time=2026-06-10T12:51:38.906+08:00 level=INFO source=images.go:354 msg="template selection" model=registry.ollama.ai/library/deepseek-r1:7b selected=gguf_chat_template renderer="" parser="" go_template=[completion] chat_template="[tools thinking completion]" harmony=null renderer_parser=null time=2026-06-10T12:51:38.907+08:00 level=INFO source=sched.go:729 msg="loaded runners" count=1 time=2026-06-10T12:51:38.907+08:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding" time=2026-06-10T12:51:38.908+08:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 3.33 seconds" srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv params_from_: Chat format: peg-native srv prompt_get_n: message_spans: last user message: byte_pos=21688, media=0, n_before_user=5212 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = -1 srv get_availabl: updating prompt cache srv load: - looking for better prompt, base f_keep = -1.000, sim = 0.000 srv update: - cache state: 0 prompts, 0.000 MiB (limits: 8192.000 MiB, 4096 tokens, 8589934592 est) srv get_availabl: prompt cache update took 0.00 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> ?top-p -> ?min-p -> ?xtc -> ?temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 1.000, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 1.000 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 0 | processing task, is_child = 0 slot update_slots: id 0 | task 0 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 5224 srv send_error: task id = 0, error: request (5224 tokens) exceeds the available context size (4096 tokens), try increasing it slot release: id 0 | task 0 | stop processing: n_tokens = 0, truncated = 0 srv update_slots: no tokens to decode srv update_slots: all slots are idle srv stop: cancel task, id_task = 0 srv update_slots: all slots are idle srv log_server_r: done request: POST /v1/chat/completions 127.0.0.1 400 [GIN] 2026/06/10 - 12:51:38 | 400 | 3.558687s | 127.0.0.1 | POST "/v1/chat/completions" [GIN] 2026/06/10 - 12:51:39 | 200 | 4.919042ms | 127.0.0.1 | POST "/api/show" srv params_from_: Chat format: peg-native srv prompt_get_n: message_spans: last user message: byte_pos=0, media=0, n_before_user=1 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = 51937852923 srv get_availabl: updating prompt cache srv load: - looking for better prompt, base f_keep = -1.000, sim = 0.000 srv update: - cache state: 0 prompts, 0.000 MiB (limits: 8192.000 MiB, 4096 tokens, 8589934592 est) srv get_availabl: prompt cache update took 0.00 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> ?top-p -> ?min-p -> ?xtc -> ?temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 1.000, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 1.000 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 3 | processing task, is_child = 0 slot update_slots: id 0 | task 3 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 8 slot update_slots: id 0 | task 3 | cached n_tokens = 0, memory_seq_rm [0, end) srv log_server_r: done request: POST /v1/chat/completions 127.0.0.1 200 slot update_slots: id 0 | task 3 | cached n_tokens = 1, memory_seq_rm [1, end) slot init_sampler: id 0 | task 3 | init sampler, took 0.00 ms, tokens: text = 8, total = 8 slot print_timing: id 0 | task 3 | prompt eval time = 216.79 ms / 8 tokens ( 27.10 ms per token, 36.90 tokens per second) slot print_timing: id 0 | task 3 | eval time = 496.82 ms / 17 tokens ( 29.22 ms per token, 34.22 tokens per second) slot print_timing: id 0 | task 3 | total time = 713.61 ms / 25 tokens slot print_timing: id 0 | task 3 | graphs reused = 16 slot release: id 0 | task 3 | stop processing: n_tokens = 24, truncated = 0 srv update_slots: all slots are idle [GIN] 2026/06/10 - 12:51:54 | 200 | 855.969791ms | 127.0.0.1 | POST "/v1/chat/completions" [GIN] 2026/06/10 - 12:52:48 | 404 | 18.958µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:52:48 | 200 | 2.448291ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:52:48 | 200 | 1.601958ms | 127.0.0.1 | POST "/api/show" srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv params_from_: Chat format: peg-native srv prompt_get_n: message_spans: last user message: byte_pos=21688, media=0, n_before_user=5212 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = 51953101445 srv get_availabl: updating prompt cache srv prompt_save: - saving prompt with length 24, total state size = 1.313 MiB (draft: 0.000 MiB) srv load: - looking for better prompt, base f_keep = 0.042, sim = 0.000 srv update: - cache state: 1 prompts, 1.313 MiB (limits: 8192.000 MiB, 4096 tokens, 149689 est) srv update: - prompt 0xbc31ccab0: 24 tokens, checkpoints: 0, 1.313 MiB srv get_availabl: prompt cache update took 0.95 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> ?top-p -> ?min-p -> ?xtc -> ?temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 1.000, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 1.000 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 22 | processing task, is_child = 0 slot update_slots: id 0 | task 22 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 5219 srv send_error: task id = 22, error: request (5219 tokens) exceeds the available context size (4096 tokens), try increasing it slot release: id 0 | task 22 | stop processing: n_tokens = 24, truncated = 0 srv update_slots: no tokens to decode srv update_slots: all slots are idle srv stop: cancel task, id_task = 22 srv update_slots: all slots are idle srv log_server_r: done request: POST /v1/chat/completions 127.0.0.1 400 [GIN] 2026/06/10 - 12:52:50 | 400 | 174.446834ms | 127.0.0.1 | POST "/v1/chat/completions" [GIN] 2026/06/10 - 12:53:27 | 404 | 22.125µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:53:27 | 200 | 2.483041ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:53:27 | 200 | 1.38775ms | 127.0.0.1 | POST "/api/show" srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv params_from_: Chat format: peg-native srv prompt_get_n: message_spans: last user message: byte_pos=20858, media=0, n_before_user=5037 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = 52008924900 srv get_availabl: updating prompt cache srv prompt_save: - saving prompt with length 24, total state size = 1.313 MiB (draft: 0.000 MiB) srv alloc: - prompt is already in the cache, skipping srv load: - looking for better prompt, base f_keep = 0.042, sim = 0.000 srv update: - cache state: 1 prompts, 1.313 MiB (limits: 8192.000 MiB, 4096 tokens, 149689 est) srv update: - prompt 0xbc31ccab0: 24 tokens, checkpoints: 0, 1.313 MiB srv get_availabl: prompt cache update took 0.02 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> ?top-p -> ?min-p -> ?xtc -> ?temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 1.000, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 1.000 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 25 | processing task, is_child = 0 slot update_slots: id 0 | task 25 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 5049 srv send_error: task id = 25, error: request (5049 tokens) exceeds the available context size (4096 tokens), try increasing it slot release: id 0 | task 25 | stop processing: n_tokens = 24, truncated = 0 srv update_slots: no tokens to decode srv update_slots: all slots are idle srv stop: cancel task, id_task = 25 srv update_slots: all slots are idle srv log_server_r: done request: POST /v1/chat/completions 127.0.0.1 400 [GIN] 2026/06/10 - 12:53:28 | 400 | 173.182459ms | 127.0.0.1 | POST "/v1/chat/completions" [GIN] 2026/06/10 - 12:55:18 | 404 | 14.333µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:55:18 | 200 | 1.911834ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:55:19 | 200 | 2.022584ms | 127.0.0.1 | POST "/api/show" srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv params_from_: Chat format: peg-native srv prompt_get_n: message_spans: last user message: byte_pos=39459, media=0, n_before_user=9560 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = 52047734445 srv get_availabl: updating prompt cache srv prompt_save: - saving prompt with length 24, total state size = 1.313 MiB (draft: 0.000 MiB) srv alloc: - prompt is already in the cache, skipping srv load: - looking for better prompt, base f_keep = 0.042, sim = 0.000 srv update: - cache state: 1 prompts, 1.313 MiB (limits: 8192.000 MiB, 4096 tokens, 149689 est) srv update: - prompt 0xbc31ccab0: 24 tokens, checkpoints: 0, 1.313 MiB srv get_availabl: prompt cache update took 0.02 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> ?top-p -> ?min-p -> ?xtc -> ?temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 1.000, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 1.000 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 28 | processing task, is_child = 0 slot update_slots: id 0 | task 28 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 9572 srv send_error: task id = 28, error: request (9572 tokens) exceeds the available context size (4096 tokens), try increasing it slot release: id 0 | task 28 | stop processing: n_tokens = 24, truncated = 0 srv update_slots: no tokens to decode srv update_slots: all slots are idle srv stop: cancel task, id_task = 28 srv update_slots: all slots are idle srv log_server_r: done request: POST /v1/chat/completions 127.0.0.1 400 [GIN] 2026/06/10 - 12:55:20 | 400 | 221.670375ms | 127.0.0.1 | POST "/v1/chat/completions" [GIN] 2026/06/10 - 12:55:41 | 404 | 19.625µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:55:41 | 200 | 3.232375ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:55:41 | 200 | 3.174792ms | 127.0.0.1 | POST "/api/show" srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv log_server_r: done request: POST /apply-template 127.0.0.1 200 srv log_server_r: done request: POST /tokenize 127.0.0.1 200 srv params_from_: Chat format: peg-native srv prompt_get_n: message_spans: last user message: byte_pos=39459, media=0, n_before_user=9560 slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = 52159405221 srv get_availabl: updating prompt cache srv prompt_save: - saving prompt with length 24, total state size = 1.313 MiB (draft: 0.000 MiB) srv alloc: - prompt is already in the cache, skipping srv load: - looking for better prompt, base f_keep = 0.042, sim = 0.000 srv update: - cache state: 1 prompts, 1.313 MiB (limits: 8192.000 MiB, 4096 tokens, 149689 est) srv update: - prompt 0xbc31ccab0: 24 tokens, checkpoints: 0, 1.313 MiB srv get_availabl: prompt cache update took 0.02 ms slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> ?top-p -> ?min-p -> ?xtc -> ?temp-ext -> dist slot launch_slot_: id 0 | task -1 | sampler params: repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000 dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096 top_k = 40, top_p = 1.000, min_p = 0.000, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 1.000 mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 slot launch_slot_: id 0 | task 31 | processing task, is_child = 0 slot update_slots: id 0 | task 31 | new prompt, n_ctx_slot = 4096, n_keep = 4, task.n_tokens = 9572 srv send_error: task id = 31, error: request (9572 tokens) exceeds the available context size (4096 tokens), try increasing it slot release: id 0 | task 31 | stop processing: n_tokens = 24, truncated = 0 srv update_slots: no tokens to decode srv update_slots: all slots are idle srv stop: cancel task, id_task = 31 srv update_slots: all slots are idle srv log_server_r: done request: POST /v1/chat/completions 127.0.0.1 400 [GIN] 2026/06/10 - 12:55:43 | 400 | 222.011ms | 127.0.0.1 | POST "/v1/chat/completions" [GIN] 2026/06/10 - 12:56:23 | 404 | 14.541µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:56:23 | 200 | 2.640833ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:56:23 | 200 | 1.756083ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:56:34 | 404 | 13.959µs | 127.0.0.1 | GET "/api/v1/models" [GIN] 2026/06/10 - 12:56:34 | 200 | 2.183667ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/06/10 - 12:56:34 | 200 | 2.157542ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:57:06 | 200 | 2.215583ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/06/10 - 12:57:18 | 200 | 606.166µs | 127.0.0.1 | GET "/api/version" time=2026-06-10T12:57:28.867+08:00 level=ERROR source=llama_server.go:861 msg="llama-server terminated" error="signal: killed" exit=unknown