github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 08:30:05 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #12113] ollama panic when run gpt-oss:120b with OLLAMA_FLASH_ATTENTION=1 on V100 #33810

New Issue
Closed
opened 2026-04-22 16:50:33 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-22 16:50:33 -05:00
Owner
Copy Link

Originally created by @wufei1234 on GitHub (Aug 29, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12113

Originally assigned to: @jessegross on GitHub.

What is the issue?

ollama can't run if we set OLLAMA_FLASH_ATTENTION=1 (panic: failed to sample token: sample: logits sum to NaN, check model output). if we set OLLAMA_FLASH_ATTENTION=0, ollama can run successfully without any issue.

Relevant log output

time=2025-08-29T16:21:56.594+08:00 level=INFO source=server.go:1270 msg="waiting for server to become available" status="llm server loading model"
time=2025-08-29T16:22:00.363+08:00 level=INFO source=server.go:1274 msg="llama runner started in 7.18 seconds"
[GIN] 2025/08/29 - 16:22:00 | 200 | 11.073563287s |       127.0.0.1 | POST     "/api/generate"
panic: failed to sample token: sample: logits sum to NaN, check model output

goroutine 89 [running]:
github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc0002d30e0, {0x557e99eacb20, 0xc0006a79f0})
        github.com/ollama/ollama/runner/ollamarunner/runner.go:375 +0x6a
created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1
        github.com/ollama/ollama/runner/ollamarunner/runner.go:1019 +0x4c9

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.11.7

Originally created by @wufei1234 on GitHub (Aug 29, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12113 Originally assigned to: @jessegross on GitHub. ### What is the issue? ollama can't run if we set OLLAMA_FLASH_ATTENTION=1 (panic: failed to sample token: sample: logits sum to NaN, check model output). if we set OLLAMA_FLASH_ATTENTION=0, ollama can run successfully without any issue. ### Relevant log output ```shell time=2025-08-29T16:21:56.594+08:00 level=INFO source=server.go:1270 msg="waiting for server to become available" status="llm server loading model" time=2025-08-29T16:22:00.363+08:00 level=INFO source=server.go:1274 msg="llama runner started in 7.18 seconds" [GIN] 2025/08/29 - 16:22:00 | 200 | 11.073563287s | 127.0.0.1 | POST "/api/generate" panic: failed to sample token: sample: logits sum to NaN, check model output goroutine 89 [running]: github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc0002d30e0, {0x557e99eacb20, 0xc0006a79f0}) github.com/ollama/ollama/runner/ollamarunner/runner.go:375 +0x6a created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1 github.com/ollama/ollama/runner/ollamarunner/runner.go:1019 +0x4c9 ``` ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.11.7
GiteaMirror added the bug label 2026-04-22 16:50:33 -05:00
GiteaMirror commented 2026-04-22 16:50:35 -05:00
Author
Owner
Copy Link

@jessegross commented on GitHub (Aug 29, 2025):

Can you please post the full log?

<!-- gh-comment-id:3237809808 --> @jessegross commented on GitHub (Aug 29, 2025): Can you please post the full log?
GiteaMirror commented 2026-04-22 16:50:35 -05:00
Author
Owner
Copy Link

@wufei1234 commented on GitHub (Aug 29, 2025):

time=2025-08-29T16:16:42.727+08:00 level=INFO source=routes.go:1331 msg="server config" env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:65536 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:1 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/data/models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:true OLLAMA_NEW_ESTIMATES:true OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE: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_SCHED_SPREAD:true ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]"
time=2025-08-29T16:16:42.730+08:00 level=INFO source=images.go:477 msg="total blobs: 7"
time=2025-08-29T16:16:42.730+08:00 level=INFO source=images.go:484 msg="total unused blobs removed: 0"
time=2025-08-29T16:16:42.730+08:00 level=INFO source=routes.go:1384 msg="Listening on [::]:11434 (version 0.11.8)"
time=2025-08-29T16:16:42.731+08:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs"
time=2025-08-29T16:16:44.472+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB"
time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB"
time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB"
time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB"
time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-d316bc72-0e0c-9b06-4544-616ecffae084 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB"
time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB"
time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-67524c8d-6955-5c13-102d-9090717dc14b library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB"
time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB"
[GIN] 2025/08/29 - 16:16:50 | 200 | 186.875µs | 127.0.0.1 | HEAD "/"
[GIN] 2025/08/29 - 16:16:50 | 200 | 386.637276ms | 127.0.0.1 | POST "/api/show"
time=2025-08-29T16:16:52.978+08:00 level=INFO source=server.go:166 msg="enabling new memory estimates"
time=2025-08-29T16:16:54.407+08:00 level=INFO source=server.go:199 msg="model wants flash attention"
time=2025-08-29T16:16:54.407+08:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-08-29T16:16:54.407+08:00 level=WARN source=server.go:224 msg="kv cache type not supported by model" type=""
time=2025-08-29T16:16:54.408+08:00 level=INFO source=server.go:388 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --model /data/models/blobs/sha256-90a618fe6ff21b09ca968df959104eb650658b0bef0faef785c18c2795d993e3 --port 43169"
time=2025-08-29T16:16:54.409+08:00 level=INFO source=server.go:661 msg="loading model" "model layers"=37 requested=-1
time=2025-08-29T16:16:54.438+08:00 level=INFO source=runner.go:1006 msg="starting ollama engine"
time=2025-08-29T16:16:54.438+08:00 level=INFO source=runner.go:1043 msg="Server listening on 127.0.0.1:43169"
time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:667 msg="system memory" total="376.5 GiB" free="367.7 GiB" free_swap="0 B"
time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-d316bc72-0e0c-9b06-4544-616ecffae084 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-67524c8d-6955-5c13-102d-9090717dc14b available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-08-29T16:16:55.869+08:00 level=INFO source=runner.go:925 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:65536 KvCacheType: NumThreads:40 GPULayers:37[ID:GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 Layers:37(0..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-08-29T16:16:56.069+08:00 level=INFO source=ggml.go:130 msg="" architecture=gptoss file_type=MXFP4 name="" description="" num_tensors=471 num_key_values=30
ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 8 CUDA devices:
Device 0: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5
Device 1: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833
Device 2: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4
Device 3: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e
Device 4: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-d316bc72-0e0c-9b06-4544-616ecffae084
Device 5: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b
Device 6: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-67524c8d-6955-5c13-102d-9090717dc14b
Device 7: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9
load_backend: loaded CUDA backend from /usr/lib/ollama/libggml-cuda.so
load_backend: loaded CPU backend from /usr/lib/ollama/libggml-cpu-skylakex.so
time=2025-08-29T16:16:56.206+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 CUDA.1.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.1.USE_GRAPHS=1 CUDA.1.PEER_MAX_BATCH_SIZE=128 CUDA.2.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.2.USE_GRAPHS=1 CUDA.2.PEER_MAX_BATCH_SIZE=128 CUDA.3.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.3.USE_GRAPHS=1 CUDA.3.PEER_MAX_BATCH_SIZE=128 CUDA.4.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.4.USE_GRAPHS=1 CUDA.4.PEER_MAX_BATCH_SIZE=128 CUDA.5.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.5.USE_GRAPHS=1 CUDA.5.PEER_MAX_BATCH_SIZE=128 CUDA.6.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.6.USE_GRAPHS=1 CUDA.6.PEER_MAX_BATCH_SIZE=128 CUDA.7.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.7.USE_GRAPHS=1 CUDA.7.PEER_MAX_BATCH_SIZE=128 compiler=cgo(gcc)
time=2025-08-29T16:16:56.388+08:00 level=INFO source=runner.go:925 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:65536 KvCacheType: NumThreads:40 GPULayers:37[ID:GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 Layers:2(0..1) ID:GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 Layers:5(2..6) ID:GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 Layers:5(7..11) ID:GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e Layers:5(12..16) ID:GPU-d316bc72-0e0c-9b06-4544-616ecffae084 Layers:5(17..21) ID:GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b Layers:5(22..26) ID:GPU-67524c8d-6955-5c13-102d-9090717dc14b Layers:5(27..31) ID:GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 Layers:5(32..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-08-29T16:16:57.601+08:00 level=INFO source=runner.go:925 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:65536 KvCacheType: NumThreads:40 GPULayers:37[ID:GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 Layers:2(0..1) ID:GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 Layers:5(2..6) ID:GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 Layers:5(7..11) ID:GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e Layers:5(12..16) ID:GPU-d316bc72-0e0c-9b06-4544-616ecffae084 Layers:5(17..21) ID:GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b Layers:5(22..26) ID:GPU-67524c8d-6955-5c13-102d-9090717dc14b Layers:5(27..31) ID:GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 Layers:5(32..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-08-29T16:16:57.845+08:00 level=INFO source=runner.go:925 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:65536 KvCacheType: NumThreads:40 GPULayers:37[ID:GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 Layers:2(0..1) ID:GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 Layers:5(2..6) ID:GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 Layers:5(7..11) ID:GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e Layers:5(12..16) ID:GPU-d316bc72-0e0c-9b06-4544-616ecffae084 Layers:5(17..21) ID:GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b Layers:5(22..26) ID:GPU-67524c8d-6955-5c13-102d-9090717dc14b Layers:5(27..31) ID:GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 Layers:5(32..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-08-29T16:16:57.845+08:00 level=INFO source=ggml.go:486 msg="offloading 36 repeating layers to GPU"
time=2025-08-29T16:16:57.845+08:00 level=INFO source=ggml.go:492 msg="offloading output layer to GPU"
time=2025-08-29T16:16:57.845+08:00 level=INFO source=ggml.go:497 msg="offloaded 37/37 layers to GPU"
time=2025-08-29T16:16:57.849+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA0 size="3.3 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA1 size="8.2 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA2 size="8.2 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA3 size="8.2 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA4 size="8.2 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA5 size="8.2 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA6 size="8.2 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA7 size="7.6 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:315 msg="model weights" device=CPU size="1.1 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA0 size="137.0 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA1 size="283.0 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA2 size="402.0 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA3 size="283.0 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA4 size="402.0 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA5 size="283.0 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA6 size="402.0 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA7 size="274.0 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA0 size="177.8 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA1 size="170.3 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA2 size="170.3 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA3 size="170.3 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA4 size="170.3 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA5 size="170.3 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA6 size="170.3 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA7 size="170.3 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:337 msg="compute graph" device=CPU size="5.6 MiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:342 msg="total memory" size="64.6 GiB"
time=2025-08-29T16:16:57.850+08:00 level=INFO source=sched.go:473 msg="loaded runners" count=1
time=2025-08-29T16:16:57.850+08:00 level=INFO source=server.go:1236 msg="waiting for llama runner to start responding"
time=2025-08-29T16:16:57.851+08:00 level=INFO source=server.go:1270 msg="waiting for server to become available" status="llm server loading model"
time=2025-08-29T16:17:17.434+08:00 level=INFO source=server.go:1274 msg="llama runner started in 23.03 seconds"
[GIN] 2025/08/29 - 16:17:17 | 200 | 26.918377909s | 127.0.0.1 | POST "/api/generate"
panic: failed to sample token: sample: logits sum to NaN, check model output

goroutine 91 [running]:
github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc0002d50e0, {0x55c49e8dfb20, 0xc0003b2960})
github.com/ollama/ollama/runner/ollamarunner/runner.go:375 +0x6a
created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1
github.com/ollama/ollama/runner/ollamarunner/runner.go:1019 +0x4c9
time=2025-08-29T16:17:35.190+08:00 level=ERROR source=server.go:1444 msg="post predict" error="Post "http://127.0.0.1:43169/completion": EOF"
[GIN] 2025/08/29 - 16:17:35 | 200 | 1.545018903s | 127.0.0.1 | POST "/api/chat"

<!-- gh-comment-id:3237827927 --> @wufei1234 commented on GitHub (Aug 29, 2025): time=2025-08-29T16:16:42.727+08:00 level=INFO source=routes.go:1331 msg="server config" env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:65536 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:1 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/data/models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:true OLLAMA_NEW_ESTIMATES:true OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE: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_SCHED_SPREAD:true ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]" time=2025-08-29T16:16:42.730+08:00 level=INFO source=images.go:477 msg="total blobs: 7" time=2025-08-29T16:16:42.730+08:00 level=INFO source=images.go:484 msg="total unused blobs removed: 0" time=2025-08-29T16:16:42.730+08:00 level=INFO source=routes.go:1384 msg="Listening on [::]:11434 (version 0.11.8)" time=2025-08-29T16:16:42.731+08:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs" time=2025-08-29T16:16:44.472+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB" time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB" time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB" time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB" time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-d316bc72-0e0c-9b06-4544-616ecffae084 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB" time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB" time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-67524c8d-6955-5c13-102d-9090717dc14b library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB" time=2025-08-29T16:16:44.473+08:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 library=cuda variant=v12 compute=7.0 driver=12.4 name="Tesla V100-SXM2-32GB" total="31.7 GiB" available="31.4 GiB" [GIN] 2025/08/29 - 16:16:50 | 200 | 186.875µs | 127.0.0.1 | HEAD "/" [GIN] 2025/08/29 - 16:16:50 | 200 | 386.637276ms | 127.0.0.1 | POST "/api/show" time=2025-08-29T16:16:52.978+08:00 level=INFO source=server.go:166 msg="enabling new memory estimates" time=2025-08-29T16:16:54.407+08:00 level=INFO source=server.go:199 msg="model wants flash attention" time=2025-08-29T16:16:54.407+08:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-08-29T16:16:54.407+08:00 level=WARN source=server.go:224 msg="kv cache type not supported by model" type="" time=2025-08-29T16:16:54.408+08:00 level=INFO source=server.go:388 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --model /data/models/blobs/sha256-90a618fe6ff21b09ca968df959104eb650658b0bef0faef785c18c2795d993e3 --port 43169" time=2025-08-29T16:16:54.409+08:00 level=INFO source=server.go:661 msg="loading model" "model layers"=37 requested=-1 time=2025-08-29T16:16:54.438+08:00 level=INFO source=runner.go:1006 msg="starting ollama engine" time=2025-08-29T16:16:54.438+08:00 level=INFO source=runner.go:1043 msg="Server listening on 127.0.0.1:43169" time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:667 msg="system memory" total="376.5 GiB" free="367.7 GiB" free_swap="0 B" time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-d316bc72-0e0c-9b06-4544-616ecffae084 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-67524c8d-6955-5c13-102d-9090717dc14b available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-08-29T16:16:55.865+08:00 level=INFO source=server.go:671 msg="gpu memory" id=GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 available="31.0 GiB" free="31.4 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-08-29T16:16:55.869+08:00 level=INFO source=runner.go:925 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:65536 KvCacheType: NumThreads:40 GPULayers:37[ID:GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 Layers:37(0..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-08-29T16:16:56.069+08:00 level=INFO source=ggml.go:130 msg="" architecture=gptoss file_type=MXFP4 name="" description="" num_tensors=471 num_key_values=30 ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 8 CUDA devices: Device 0: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 Device 1: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 Device 2: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 Device 3: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e Device 4: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-d316bc72-0e0c-9b06-4544-616ecffae084 Device 5: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b Device 6: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-67524c8d-6955-5c13-102d-9090717dc14b Device 7: Tesla V100-SXM2-32GB, compute capability 7.0, VMM: yes, ID: GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 load_backend: loaded CUDA backend from /usr/lib/ollama/libggml-cuda.so load_backend: loaded CPU backend from /usr/lib/ollama/libggml-cpu-skylakex.so time=2025-08-29T16:16:56.206+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 CUDA.1.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.1.USE_GRAPHS=1 CUDA.1.PEER_MAX_BATCH_SIZE=128 CUDA.2.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.2.USE_GRAPHS=1 CUDA.2.PEER_MAX_BATCH_SIZE=128 CUDA.3.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.3.USE_GRAPHS=1 CUDA.3.PEER_MAX_BATCH_SIZE=128 CUDA.4.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.4.USE_GRAPHS=1 CUDA.4.PEER_MAX_BATCH_SIZE=128 CUDA.5.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.5.USE_GRAPHS=1 CUDA.5.PEER_MAX_BATCH_SIZE=128 CUDA.6.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.6.USE_GRAPHS=1 CUDA.6.PEER_MAX_BATCH_SIZE=128 CUDA.7.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.7.USE_GRAPHS=1 CUDA.7.PEER_MAX_BATCH_SIZE=128 compiler=cgo(gcc) time=2025-08-29T16:16:56.388+08:00 level=INFO source=runner.go:925 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:65536 KvCacheType: NumThreads:40 GPULayers:37[ID:GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 Layers:2(0..1) ID:GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 Layers:5(2..6) ID:GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 Layers:5(7..11) ID:GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e Layers:5(12..16) ID:GPU-d316bc72-0e0c-9b06-4544-616ecffae084 Layers:5(17..21) ID:GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b Layers:5(22..26) ID:GPU-67524c8d-6955-5c13-102d-9090717dc14b Layers:5(27..31) ID:GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 Layers:5(32..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-08-29T16:16:57.601+08:00 level=INFO source=runner.go:925 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:65536 KvCacheType: NumThreads:40 GPULayers:37[ID:GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 Layers:2(0..1) ID:GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 Layers:5(2..6) ID:GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 Layers:5(7..11) ID:GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e Layers:5(12..16) ID:GPU-d316bc72-0e0c-9b06-4544-616ecffae084 Layers:5(17..21) ID:GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b Layers:5(22..26) ID:GPU-67524c8d-6955-5c13-102d-9090717dc14b Layers:5(27..31) ID:GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 Layers:5(32..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-08-29T16:16:57.845+08:00 level=INFO source=runner.go:925 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:65536 KvCacheType: NumThreads:40 GPULayers:37[ID:GPU-720b1cbc-e07b-06e5-33b3-04f1795ed3b5 Layers:2(0..1) ID:GPU-dc20d4a3-b76b-33f3-74ff-765ecad0d833 Layers:5(2..6) ID:GPU-bbd9ab7d-a6b0-1db4-57c3-6ea3b3a9fab4 Layers:5(7..11) ID:GPU-9ba7e0f7-14b2-8c82-af0a-1e8fa8082d6e Layers:5(12..16) ID:GPU-d316bc72-0e0c-9b06-4544-616ecffae084 Layers:5(17..21) ID:GPU-6a0d29d6-ad75-c875-c285-6a04aa107f2b Layers:5(22..26) ID:GPU-67524c8d-6955-5c13-102d-9090717dc14b Layers:5(27..31) ID:GPU-7b4762f2-9a22-89c9-c605-a570b2e68bd9 Layers:5(32..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-08-29T16:16:57.845+08:00 level=INFO source=ggml.go:486 msg="offloading 36 repeating layers to GPU" time=2025-08-29T16:16:57.845+08:00 level=INFO source=ggml.go:492 msg="offloading output layer to GPU" time=2025-08-29T16:16:57.845+08:00 level=INFO source=ggml.go:497 msg="offloaded 37/37 layers to GPU" time=2025-08-29T16:16:57.849+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA0 size="3.3 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA1 size="8.2 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA2 size="8.2 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA3 size="8.2 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA4 size="8.2 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA5 size="8.2 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA6 size="8.2 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:310 msg="model weights" device=CUDA7 size="7.6 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:315 msg="model weights" device=CPU size="1.1 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA0 size="137.0 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA1 size="283.0 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA2 size="402.0 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA3 size="283.0 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA4 size="402.0 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA5 size="283.0 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA6 size="402.0 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:321 msg="kv cache" device=CUDA7 size="274.0 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA0 size="177.8 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA1 size="170.3 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA2 size="170.3 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA3 size="170.3 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA4 size="170.3 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA5 size="170.3 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA6 size="170.3 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:332 msg="compute graph" device=CUDA7 size="170.3 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:337 msg="compute graph" device=CPU size="5.6 MiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=backend.go:342 msg="total memory" size="64.6 GiB" time=2025-08-29T16:16:57.850+08:00 level=INFO source=sched.go:473 msg="loaded runners" count=1 time=2025-08-29T16:16:57.850+08:00 level=INFO source=server.go:1236 msg="waiting for llama runner to start responding" time=2025-08-29T16:16:57.851+08:00 level=INFO source=server.go:1270 msg="waiting for server to become available" status="llm server loading model" time=2025-08-29T16:17:17.434+08:00 level=INFO source=server.go:1274 msg="llama runner started in 23.03 seconds" [GIN] 2025/08/29 - 16:17:17 | 200 | 26.918377909s | 127.0.0.1 | POST "/api/generate" panic: failed to sample token: sample: logits sum to NaN, check model output goroutine 91 [running]: github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc0002d50e0, {0x55c49e8dfb20, 0xc0003b2960}) github.com/ollama/ollama/runner/ollamarunner/runner.go:375 +0x6a created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1 github.com/ollama/ollama/runner/ollamarunner/runner.go:1019 +0x4c9 time=2025-08-29T16:17:35.190+08:00 level=ERROR source=server.go:1444 msg="post predict" error="Post \"http://127.0.0.1:43169/completion\": EOF" [GIN] 2025/08/29 - 16:17:35 | 200 | 1.545018903s | 127.0.0.1 | POST "/api/chat"
GiteaMirror commented 2026-04-22 16:50:36 -05:00
Author
Owner
Copy Link

@arbv commented on GitHub (Aug 30, 2025):

The same thing happens on RX 7900 XTX (partial GPU offload case, obviously):

сер 30 22:45:08 cauldron ollama[2294635]: time=2025-08-30T19:45:08.551Z level=INFO source=server.go:1274 msg="llama runner started in 11.89 seconds"
сер 30 22:45:34 cauldron ollama[2294635]: panic: failed to sample token: sample: logits sum to NaN, check model output
сер 30 22:45:34 cauldron ollama[2294635]: 
сер 30 22:45:34 cauldron ollama[2294635]: goroutine 69 [running]:
сер 30 22:45:34 cauldron ollama[2294635]: github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc0002610e0, {0x55fafd5bfb20, 0xc000130ff0})
сер 30 22:45:34 cauldron ollama[2294635]:         github.com/ollama/ollama/runner/ollamarunner/runner.go:375 +0x6a
сер 30 22:45:34 cauldron ollama[2294635]: created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1
сер 30 22:45:34 cauldron ollama[2294635]:         github.com/ollama/ollama/runner/ollamarunner/runner.go:1019 +0x4c9
сер 30 22:45:36 cauldron ollama[2294635]: time=2025-08-30T19:45:36.062Z level=ERROR source=server.go:1444 msg="post predict" error="Post \"http://127.0.0.1:43811/completion\": EOF"

The latest docker build of ollama is used.

<!-- gh-comment-id:3239506657 --> @arbv commented on GitHub (Aug 30, 2025): The same thing happens on RX 7900 XTX (partial GPU offload case, obviously): ``` сер 30 22:45:08 cauldron ollama[2294635]: time=2025-08-30T19:45:08.551Z level=INFO source=server.go:1274 msg="llama runner started in 11.89 seconds" сер 30 22:45:34 cauldron ollama[2294635]: panic: failed to sample token: sample: logits sum to NaN, check model output сер 30 22:45:34 cauldron ollama[2294635]: сер 30 22:45:34 cauldron ollama[2294635]: goroutine 69 [running]: сер 30 22:45:34 cauldron ollama[2294635]: github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc0002610e0, {0x55fafd5bfb20, 0xc000130ff0}) сер 30 22:45:34 cauldron ollama[2294635]: github.com/ollama/ollama/runner/ollamarunner/runner.go:375 +0x6a сер 30 22:45:34 cauldron ollama[2294635]: created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1 сер 30 22:45:34 cauldron ollama[2294635]: github.com/ollama/ollama/runner/ollamarunner/runner.go:1019 +0x4c9 сер 30 22:45:36 cauldron ollama[2294635]: time=2025-08-30T19:45:36.062Z level=ERROR source=server.go:1444 msg="post predict" error="Post \"http://127.0.0.1:43811/completion\": EOF" ``` The latest docker build of ollama is used.
GiteaMirror commented 2026-04-22 16:50:36 -05:00
Author
Owner
Copy Link

@jessegross commented on GitHub (Oct 3, 2025):

For those that are running into this issue, please test out one of the 0.12.4 RCs and let us know if it fixes the issue with V100s - we have updated the kernels.

<!-- gh-comment-id:3366842541 --> @jessegross commented on GitHub (Oct 3, 2025): For those that are running into this issue, please test out one of the 0.12.4 RCs and let us know if it fixes the issue with V100s - we have updated the kernels.
GiteaMirror commented 2026-04-22 16:50:37 -05:00
Author
Owner
Copy Link

@jessegross commented on GitHub (Oct 8, 2025):

Fixed by #12245

<!-- gh-comment-id:3382454063 --> @jessegross commented on GitHub (Oct 8, 2025): Fixed by #12245
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#33810
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?