github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9436] could not run phi4-mini:3.8b-fp16 #6151

New Issue
Closed
opened 2026-04-12 17:30:04 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 17:30:04 -05:00
Owner
Copy Link

Originally created by @husy8 on GitHub (Mar 1, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9436

What is the issue?

I got a runtime error with following:

(base) user@localhost:~$ ollama run phi4-mini:3.8b-fp16
Error: llama runner process has terminated: error loading model: missing tensor 'output.weight'
llama_load_model_from_file: failed to load model

the model downloaded from https://ollama.com/library/phi4-mini:3.8b-fp16, last update time should be around at 2025.03.01 04:00 UTC.

Relevant log output

Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.132+08:00 level=WARN source=ggml.go:132 msg="key not found" key=phi3.attention.key_length default=128
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.132+08:00 level=WARN source=ggml.go:132 msg="key not found" key=phi3.attention.value_length default=128
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.132+08:00 level=INFO source=sched.go:715 msg="new model will fit in available VRAM in single GPU, loading" model=/usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97 gpu=GPU-09638d0b-28bf-bec6-5473-76a11a69be3a parallel=4 available=12346589184 required="9.3 GiB"
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.272+08:00 level=INFO source=server.go:97 msg="system memory" total="31.2 GiB" free="29.0 GiB" free_swap="0 B"
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=WARN source=ggml.go:132 msg="key not found" key=phi3.attention.key_length default=128
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=WARN source=ggml.go:132 msg="key not found" key=phi3.attention.value_length default=128
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=server.go:130 msg=offload library=cuda layers.requested=-1 layers.model=33 layers.offload=33 layers.split="" memory.available="[11.5 GiB]" memory.gpu_overhead="0 B" memory.required.full="9.3 GiB" memory.required.partial="9.3 GiB" memory.required.kv="1.0 GiB" memory.required.allocations="[9.3 GiB]" memory.weights.total="7.0 GiB" memory.weights.repeating="5.9 GiB" memory.weights.nonrepeating="1.1 GiB" memory.graph.full="512.0 MiB" memory.graph.partial="512.0 MiB"
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=server.go:380 msg="starting llama server" cmd="/usr/local/bin/ollama runner --model /usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97 --ctx-size 8192 --batch-size 512 --n-gpu-layers 33 --threads 6 --parallel 4 --port 43815"
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=sched.go:450 msg="loaded runners" count=1
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=server.go:557 msg="waiting for llama runner to start responding"
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=server.go:591 msg="waiting for server to become available" status="llm server error"
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.283+08:00 level=INFO source=runner.go:932 msg="starting go runner"
Mar 01 12:45:30 localhost ollama[3824]: ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
Mar 01 12:45:30 localhost ollama[3824]: ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
Mar 01 12:45:30 localhost ollama[3824]: ggml_cuda_init: found 1 CUDA devices:
Mar 01 12:45:30 localhost ollama[3824]:   Device 0: NVIDIA GeForce RTX 3060, compute capability 8.6, VMM: yes
Mar 01 12:45:30 localhost ollama[3824]: load_backend: loaded CUDA backend from /usr/local/lib/ollama/cuda_v12/libggml-cuda.so
Mar 01 12:45:30 localhost ollama[3824]: load_backend: loaded CPU backend from /usr/local/lib/ollama/libggml-cpu-haswell.so
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.314+08:00 level=INFO source=runner.go:935 msg=system info="CPU : LLAMAFILE = 1 | CPU : LLAMAFILE = 1 | CUDA : ARCHS = 600,610,620,700,720,750,800,860,870,890,900 | USE_GRAPHS = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | LLAMAFILE = 1 | cgo(gcc)" threads=6
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.315+08:00 level=INFO source=runner.go:993 msg="Server listening on 127.0.0.1:43815"
Mar 01 12:45:30 localhost ollama[3824]: llama_load_model_from_file: using device CUDA0 (NVIDIA GeForce RTX 3060) - 11774 MiB free
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: loaded meta data with 36 key-value pairs and 196 tensors from /usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97 (version GGUF V3 (latest))
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   0:                       general.architecture str              = phi3
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   1:              phi3.rope.scaling.attn_factor f32              = 1.190238
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   2:                               general.type str              = model
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   3:                               general.name str              = Phi 4 Mini Instruct
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   4:                           general.finetune str              = instruct
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   5:                           general.basename str              = Phi-4
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   6:                         general.size_label str              = mini
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   7:                            general.license str              = mit
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   8:                       general.license.link str              = https://huggingface.co/microsoft/Phi-...
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv   9:                               general.tags arr[str,3]       = ["nlp", "code", "text-generation"]
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  10:                          general.languages arr[str,24]      = ["multilingual", "ar", "zh", "cs", "d...
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  11:                        phi3.context_length u32              = 131072
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  12:  phi3.rope.scaling.original_context_length u32              = 4096
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  13:                      phi3.embedding_length u32              = 3072
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  14:                   phi3.feed_forward_length u32              = 8192
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  15:                           phi3.block_count u32              = 32
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  16:                  phi3.attention.head_count u32              = 24
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  17:               phi3.attention.head_count_kv u32              = 8
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  18:      phi3.attention.layer_norm_rms_epsilon f32              = 0.000010
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  19:                  phi3.rope.dimension_count u32              = 96
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  20:                        phi3.rope.freq_base f32              = 10000.000000
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  21:                          general.file_type u32              = 1
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  22:              phi3.attention.sliding_window u32              = 262144
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  23:                       tokenizer.ggml.model str              = gpt2
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  24:                         tokenizer.ggml.pre str              = gpt-4o
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  25:                      tokenizer.ggml.tokens arr[str,200064]  = ["!", "\"", "#", "$", "%", "&", "'", ...
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  26:                  tokenizer.ggml.token_type arr[i32,200064]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  27:                      tokenizer.ggml.merges arr[str,199742]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "e r", ...
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  28:                tokenizer.ggml.bos_token_id u32              = 199999
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  29:                tokenizer.ggml.eos_token_id u32              = 199999
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  30:            tokenizer.ggml.unknown_token_id u32              = 199999
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  31:            tokenizer.ggml.padding_token_id u32              = 199999
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  32:               tokenizer.ggml.add_bos_token bool             = false
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  33:               tokenizer.ggml.add_eos_token bool             = false
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  34:                    tokenizer.chat_template str              = {% for message in messages %}{% if me...
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv  35:               general.quantization_version u32              = 2
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - type  f32:   67 tensors
Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - type  f16:  129 tensors
Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.525+08:00 level=INFO source=server.go:591 msg="waiting for server to become available" status="llm server loading model"
Mar 01 12:45:30 localhost ollama[3824]: llm_load_vocab: missing or unrecognized pre-tokenizer type, using: 'default'
Mar 01 12:45:30 localhost ollama[3824]: llm_load_vocab: special tokens cache size = 12
Mar 01 12:45:30 localhost ollama[3824]: llm_load_vocab: token to piece cache size = 1.3333 MB
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: format           = GGUF V3 (latest)
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: arch             = phi3
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: vocab type       = BPE
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_vocab          = 200064
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_merges         = 199742
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: vocab_only       = 0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_ctx_train      = 131072
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd           = 3072
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_layer          = 32
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_head           = 24
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_head_kv        = 8
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_rot            = 96
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_swa            = 262144
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd_head_k    = 128
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd_head_v    = 128
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_gqa            = 3
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd_k_gqa     = 1024
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd_v_gqa     = 1024
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_norm_eps       = 0.0e+00
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_norm_rms_eps   = 1.0e-05
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_clamp_kqv      = 0.0e+00
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_max_alibi_bias = 0.0e+00
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_logit_scale    = 0.0e+00
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_ff             = 8192
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_expert         = 0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_expert_used    = 0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: causal attn      = 1
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: pooling type     = 0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: rope type        = 2
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: rope scaling     = linear
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: freq_base_train  = 10000.0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: freq_scale_train = 1
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_ctx_orig_yarn  = 4096
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: rope_finetuned   = unknown
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_d_conv       = 0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_d_inner      = 0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_d_state      = 0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_dt_rank      = 0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_dt_b_c_rms   = 0
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: model type       = 3B
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: model ftype      = F16
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: model params     = 3.84 B
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: model size       = 7.15 GiB (16.00 BPW)
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: general.name     = Phi 4 Mini Instruct
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: BOS token        = 199999 '<|endoftext|>'
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: EOS token        = 199999 '<|endoftext|>'
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: EOT token        = 199999 '<|endoftext|>'
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: UNK token        = 199999 '<|endoftext|>'
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: PAD token        = 199999 '<|endoftext|>'
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: LF token         = 128 'Ä'
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: EOG token        = 199999 '<|endoftext|>'
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: EOG token        = 200020 '<|end|>'
Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: max token length = 256
Mar 01 12:45:31 localhost ollama[3824]: llama_model_load: error loading model: missing tensor 'output.weight'
Mar 01 12:45:31 localhost ollama[3824]: llama_load_model_from_file: failed to load model
Mar 01 12:45:31 localhost ollama[3824]: panic: unable to load model: /usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97
Mar 01 12:45:31 localhost ollama[3824]: goroutine 50 [running]:
Mar 01 12:45:31 localhost ollama[3824]: github.com/ollama/ollama/runner/llamarunner.(*Server).loadModel(0xc0005c8000, {0x21, 0x0, 0x1, 0x0, {0x0, 0x0, 0x0}, 0xc000116020, 0x0}, ...)
Mar 01 12:45:31 localhost ollama[3824]:         github.com/ollama/ollama/runner/llamarunner/runner.go:851 +0x38d
Mar 01 12:45:31 localhost ollama[3824]: created by github.com/ollama/ollama/runner/llamarunner.Execute in goroutine 1
Mar 01 12:45:31 localhost ollama[3824]:         github.com/ollama/ollama/runner/llamarunner/runner.go:968 +0xcd5
Mar 01 12:45:31 localhost ollama[3824]: time=2025-03-01T12:45:31.149+08:00 level=ERROR source=server.go:421 msg="llama runner terminated" error="exit status 2"
Mar 01 12:45:31 localhost ollama[3824]: time=2025-03-01T12:45:31.279+08:00 level=ERROR source=sched.go:456 msg="error loading llama server" error="llama runner process has terminated: error loading model: missing tensor 'output.weight'\nllama_load_model_from_file: failed to load model"
Mar 01 12:45:31 localhost ollama[3824]: [GIN] 2025/03/01 - 12:45:31 | 500 |  1.351642665s |       127.0.0.1 | POST     "/api/generate"
Mar 01 12:45:36 localhost ollama[3824]: time=2025-03-01T12:45:36.428+08:00 level=WARN source=sched.go:647 msg="gpu VRAM usage didn't recover within timeout" seconds=5.149761061 model=/usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97

OS

Linux localhost 6.1.0-31-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.128-1 (2025-02-07) x86_64 GNU/Linux

GPU

NVIDIA RTX 3060 x1

CPU

AMD Ryzen 5 5500

Ollama version

0.5.12

Originally created by @husy8 on GitHub (Mar 1, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9436 ### What is the issue? I got a runtime error with following: ```bash (base) user@localhost:~$ ollama run phi4-mini:3.8b-fp16 Error: llama runner process has terminated: error loading model: missing tensor 'output.weight' llama_load_model_from_file: failed to load model ``` the model downloaded from https://ollama.com/library/phi4-mini:3.8b-fp16, last update time should be around at 2025.03.01 04:00 UTC. ### Relevant log output ```shell Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.132+08:00 level=WARN source=ggml.go:132 msg="key not found" key=phi3.attention.key_length default=128 Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.132+08:00 level=WARN source=ggml.go:132 msg="key not found" key=phi3.attention.value_length default=128 Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.132+08:00 level=INFO source=sched.go:715 msg="new model will fit in available VRAM in single GPU, loading" model=/usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97 gpu=GPU-09638d0b-28bf-bec6-5473-76a11a69be3a parallel=4 available=12346589184 required="9.3 GiB" Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.272+08:00 level=INFO source=server.go:97 msg="system memory" total="31.2 GiB" free="29.0 GiB" free_swap="0 B" Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=WARN source=ggml.go:132 msg="key not found" key=phi3.attention.key_length default=128 Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=WARN source=ggml.go:132 msg="key not found" key=phi3.attention.value_length default=128 Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=server.go:130 msg=offload library=cuda layers.requested=-1 layers.model=33 layers.offload=33 layers.split="" memory.available="[11.5 GiB]" memory.gpu_overhead="0 B" memory.required.full="9.3 GiB" memory.required.partial="9.3 GiB" memory.required.kv="1.0 GiB" memory.required.allocations="[9.3 GiB]" memory.weights.total="7.0 GiB" memory.weights.repeating="5.9 GiB" memory.weights.nonrepeating="1.1 GiB" memory.graph.full="512.0 MiB" memory.graph.partial="512.0 MiB" Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=server.go:380 msg="starting llama server" cmd="/usr/local/bin/ollama runner --model /usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97 --ctx-size 8192 --batch-size 512 --n-gpu-layers 33 --threads 6 --parallel 4 --port 43815" Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=sched.go:450 msg="loaded runners" count=1 Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=server.go:557 msg="waiting for llama runner to start responding" Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.273+08:00 level=INFO source=server.go:591 msg="waiting for server to become available" status="llm server error" Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.283+08:00 level=INFO source=runner.go:932 msg="starting go runner" Mar 01 12:45:30 localhost ollama[3824]: ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no Mar 01 12:45:30 localhost ollama[3824]: ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no Mar 01 12:45:30 localhost ollama[3824]: ggml_cuda_init: found 1 CUDA devices: Mar 01 12:45:30 localhost ollama[3824]: Device 0: NVIDIA GeForce RTX 3060, compute capability 8.6, VMM: yes Mar 01 12:45:30 localhost ollama[3824]: load_backend: loaded CUDA backend from /usr/local/lib/ollama/cuda_v12/libggml-cuda.so Mar 01 12:45:30 localhost ollama[3824]: load_backend: loaded CPU backend from /usr/local/lib/ollama/libggml-cpu-haswell.so Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.314+08:00 level=INFO source=runner.go:935 msg=system info="CPU : LLAMAFILE = 1 | CPU : LLAMAFILE = 1 | CUDA : ARCHS = 600,610,620,700,720,750,800,860,870,890,900 | USE_GRAPHS = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | LLAMAFILE = 1 | cgo(gcc)" threads=6 Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.315+08:00 level=INFO source=runner.go:993 msg="Server listening on 127.0.0.1:43815" Mar 01 12:45:30 localhost ollama[3824]: llama_load_model_from_file: using device CUDA0 (NVIDIA GeForce RTX 3060) - 11774 MiB free Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: loaded meta data with 36 key-value pairs and 196 tensors from /usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97 (version GGUF V3 (latest)) Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 0: general.architecture str = phi3 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 1: phi3.rope.scaling.attn_factor f32 = 1.190238 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 2: general.type str = model Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 3: general.name str = Phi 4 Mini Instruct Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 4: general.finetune str = instruct Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 5: general.basename str = Phi-4 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 6: general.size_label str = mini Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 7: general.license str = mit Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 8: general.license.link str = https://huggingface.co/microsoft/Phi-... Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 9: general.tags arr[str,3] = ["nlp", "code", "text-generation"] Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 10: general.languages arr[str,24] = ["multilingual", "ar", "zh", "cs", "d... Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 11: phi3.context_length u32 = 131072 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 12: phi3.rope.scaling.original_context_length u32 = 4096 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 13: phi3.embedding_length u32 = 3072 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 14: phi3.feed_forward_length u32 = 8192 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 15: phi3.block_count u32 = 32 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 16: phi3.attention.head_count u32 = 24 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 17: phi3.attention.head_count_kv u32 = 8 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 18: phi3.attention.layer_norm_rms_epsilon f32 = 0.000010 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 19: phi3.rope.dimension_count u32 = 96 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 20: phi3.rope.freq_base f32 = 10000.000000 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 21: general.file_type u32 = 1 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 22: phi3.attention.sliding_window u32 = 262144 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 23: tokenizer.ggml.model str = gpt2 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 24: tokenizer.ggml.pre str = gpt-4o Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 25: tokenizer.ggml.tokens arr[str,200064] = ["!", "\"", "#", "$", "%", "&", "'", ... Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 26: tokenizer.ggml.token_type arr[i32,200064] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 27: tokenizer.ggml.merges arr[str,199742] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "e r", ... Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 28: tokenizer.ggml.bos_token_id u32 = 199999 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 29: tokenizer.ggml.eos_token_id u32 = 199999 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 30: tokenizer.ggml.unknown_token_id u32 = 199999 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 31: tokenizer.ggml.padding_token_id u32 = 199999 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 32: tokenizer.ggml.add_bos_token bool = false Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 33: tokenizer.ggml.add_eos_token bool = false Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 34: tokenizer.chat_template str = {% for message in messages %}{% if me... Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - kv 35: general.quantization_version u32 = 2 Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - type f32: 67 tensors Mar 01 12:45:30 localhost ollama[3824]: llama_model_loader: - type f16: 129 tensors Mar 01 12:45:30 localhost ollama[3824]: time=2025-03-01T12:45:30.525+08:00 level=INFO source=server.go:591 msg="waiting for server to become available" status="llm server loading model" Mar 01 12:45:30 localhost ollama[3824]: llm_load_vocab: missing or unrecognized pre-tokenizer type, using: 'default' Mar 01 12:45:30 localhost ollama[3824]: llm_load_vocab: special tokens cache size = 12 Mar 01 12:45:30 localhost ollama[3824]: llm_load_vocab: token to piece cache size = 1.3333 MB Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: format = GGUF V3 (latest) Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: arch = phi3 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: vocab type = BPE Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_vocab = 200064 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_merges = 199742 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: vocab_only = 0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_ctx_train = 131072 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd = 3072 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_layer = 32 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_head = 24 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_head_kv = 8 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_rot = 96 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_swa = 262144 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd_head_k = 128 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd_head_v = 128 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_gqa = 3 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd_k_gqa = 1024 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_embd_v_gqa = 1024 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_norm_eps = 0.0e+00 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_norm_rms_eps = 1.0e-05 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_clamp_kqv = 0.0e+00 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_max_alibi_bias = 0.0e+00 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: f_logit_scale = 0.0e+00 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_ff = 8192 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_expert = 0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_expert_used = 0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: causal attn = 1 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: pooling type = 0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: rope type = 2 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: rope scaling = linear Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: freq_base_train = 10000.0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: freq_scale_train = 1 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: n_ctx_orig_yarn = 4096 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: rope_finetuned = unknown Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_d_conv = 0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_d_inner = 0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_d_state = 0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_dt_rank = 0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: ssm_dt_b_c_rms = 0 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: model type = 3B Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: model ftype = F16 Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: model params = 3.84 B Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: model size = 7.15 GiB (16.00 BPW) Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: general.name = Phi 4 Mini Instruct Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: BOS token = 199999 '<|endoftext|>' Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: EOS token = 199999 '<|endoftext|>' Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: EOT token = 199999 '<|endoftext|>' Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: UNK token = 199999 '<|endoftext|>' Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: PAD token = 199999 '<|endoftext|>' Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: LF token = 128 'Ä' Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: EOG token = 199999 '<|endoftext|>' Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: EOG token = 200020 '<|end|>' Mar 01 12:45:30 localhost ollama[3824]: llm_load_print_meta: max token length = 256 Mar 01 12:45:31 localhost ollama[3824]: llama_model_load: error loading model: missing tensor 'output.weight' Mar 01 12:45:31 localhost ollama[3824]: llama_load_model_from_file: failed to load model Mar 01 12:45:31 localhost ollama[3824]: panic: unable to load model: /usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97 Mar 01 12:45:31 localhost ollama[3824]: goroutine 50 [running]: Mar 01 12:45:31 localhost ollama[3824]: github.com/ollama/ollama/runner/llamarunner.(*Server).loadModel(0xc0005c8000, {0x21, 0x0, 0x1, 0x0, {0x0, 0x0, 0x0}, 0xc000116020, 0x0}, ...) Mar 01 12:45:31 localhost ollama[3824]: github.com/ollama/ollama/runner/llamarunner/runner.go:851 +0x38d Mar 01 12:45:31 localhost ollama[3824]: created by github.com/ollama/ollama/runner/llamarunner.Execute in goroutine 1 Mar 01 12:45:31 localhost ollama[3824]: github.com/ollama/ollama/runner/llamarunner/runner.go:968 +0xcd5 Mar 01 12:45:31 localhost ollama[3824]: time=2025-03-01T12:45:31.149+08:00 level=ERROR source=server.go:421 msg="llama runner terminated" error="exit status 2" Mar 01 12:45:31 localhost ollama[3824]: time=2025-03-01T12:45:31.279+08:00 level=ERROR source=sched.go:456 msg="error loading llama server" error="llama runner process has terminated: error loading model: missing tensor 'output.weight'\nllama_load_model_from_file: failed to load model" Mar 01 12:45:31 localhost ollama[3824]: [GIN] 2025/03/01 - 12:45:31 | 500 | 1.351642665s | 127.0.0.1 | POST "/api/generate" Mar 01 12:45:36 localhost ollama[3824]: time=2025-03-01T12:45:36.428+08:00 level=WARN source=sched.go:647 msg="gpu VRAM usage didn't recover within timeout" seconds=5.149761061 model=/usr/share/ollama/.ollama/models/blobs/sha256-e7bb32183dad1cc57730edf523bd6ac18716005bb579384d279e029e63828f97 ``` ### OS Linux localhost 6.1.0-31-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.128-1 (2025-02-07) x86_64 GNU/Linux ### GPU NVIDIA RTX 3060 x1 ### CPU AMD Ryzen 5 5500 ### Ollama version 0.5.12
GiteaMirror added the bug label 2026-04-12 17:30:04 -05:00
GiteaMirror commented 2026-04-12 17:30:05 -05:00
Author
Owner
Copy Link

@mswcap commented on GitHub (Mar 1, 2025):

You need to download the rc version of Ollama, and then it wil run just fine. The Phi4-mini model page states it, see screenshot.

Image

<!-- gh-comment-id:2692039423 --> @mswcap commented on GitHub (Mar 1, 2025): You need to download the rc version of Ollama, and then it wil run just fine. The Phi4-mini model page states it, see screenshot. ![Image](https://github.com/user-attachments/assets/8ed9c723-cf1b-41d0-9da0-79efb0214d6e)
GiteaMirror commented 2026-04-12 17:30:05 -05:00
Author
Owner
Copy Link

@husy8 commented on GitHub (Mar 1, 2025):

You need to download the rc version of Ollama, and then it wil run just fine. The Phi4-mini model page states it, see screenshot.

Image

Thank you! I totally missed that note

<!-- gh-comment-id:2692118642 --> @husy8 commented on GitHub (Mar 1, 2025): > You need to download the rc version of Ollama, and then it wil run just fine. The Phi4-mini model page states it, see screenshot. > > ![Image](https://github.com/user-attachments/assets/8ed9c723-cf1b-41d0-9da0-79efb0214d6e) Thank you! I totally missed that note
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#6151
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?