github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #14273] glm-4.7-flash emitting gibberish after a while #9294

New Issue
Open
opened 2026-04-12 22:09:39 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-04-12 22:09:39 -05:00
Owner
Copy Link

Originally created by @MitchTalmadge on GitHub (Feb 16, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/14273

What is the issue?

I have just installed Ollama on my Windows desktop and chose the glm-4.7-flash model. While having it write a poem, it started outputting complete gibberish with no end in sight. I eventually stopped it. Then, all prompts, even after running /clear, would output this gibberish. After running /bye and starting it again, it works fine again.

  • CPU: Ryzen 9 5950x
  • GPU: RTX 4070Ti
  • RAM: 128GB DDR4

Here is when it started; see the line after "Stanza 5: Conclusion"

>>> /clear
Cleared session context
>>> Write a poem about chocolate
Thinking...
1.  **Analyze the Request:**
    *   **Topic:** Chocolate.
    *   **Format:** Poem.
    *   **Tone:** Open, but usually implies appreciation, indulgence, or sensory description.

2.  **Brainstorming Imagery & Themes:**
    *   *Visual:* Dark, glossy, liquid, swirling, melting, dust of cocoa, velvety.
    *   *Tactile:* Smooth, silky, melting on the tongue, cool in the hand, warm in the mouth.
    *   *Taste:* Bitter, sweet, rich, earthy, decadent, forbidden, comforting.
    *   *Emotional:* Joy, warmth, escapism, love, comfort, ritual.
    *   *Associations:* Cacao beans, melting pot, gift, afternoon treat, dessert.

3.  **Drafting - Stanza by Stanza:**

    *   *Stanza 1: Introduction/The Origin (Earthiness).*
        *   Start with the raw material. Beans from the tropics.
        *   *Drafting ideas:* It comes from the sun, dark beans. Earthy soul.
        *   *Refining:* "The soul of the earth in a dusty pod," "Forged in the heat."

    *   *Stanza 2: The Transformation (Processing).*
        *   Roasting and grinding.
        *   *Drafting ideas:* Roasted to perfection, crushed into dust, liquid gold.
        *   *Refining:* "Turned to liquid gold," "The roast and the crush."

    *   *Stanza 3: The Sensory Experience (Taste/Texture).*
        *   Eating it. The contrast of sweet and bitter.
        *   *Drafting ideas:* Melts on the tongue, velvety coat, sweet rush.
        *   *Refining:* "Velvet on the tongue," "The sweet that bites back," "Decadence."

    *   *Stanza 4: The Emotional Connection (Comfort).*
        *   Why we eat it. The feeling of warmth.
        *   *Drafting ideas:* A hug for the soul, rainy day comfort, little piece of joy.
        *   *Refining:* "A hug for the soul," "A little piece of the sun."

    *   *Stanza 5: Conclusion.*
        *   Summary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

>>> Hello
Thinking...
=E<+;4*A3C@CD*..*5F0"(2$@.*"6A(@,#;>52F65;:632E"/9;B;2*F31?#'2;>2+E/?"DE+;0'$25.:C;)D&42#)<8B=&651EC)1%=F97)3,7'#-$'@%,406C)*67>%4@

>>> /clear
Cleared session context
>>> Hello
Thinking...
7=>81*<6FE@33<D7,5%7E4;,'0@+&(.8,3$0

>>> Send a message (/? for help)

Relevant log output

server.log:

time=2026-02-15T19:55:36.330-07:00 level=INFO source=routes.go:1636 msg="server config" env="map[CUDA_VISIBLE_DEVICES: GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_EDITOR: OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 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_MODELS:C:\\Users\\mitcht\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false 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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES:]"
time=2026-02-15T19:55:36.332-07:00 level=INFO source=images.go:473 msg="total blobs: 0"
time=2026-02-15T19:55:36.332-07:00 level=INFO source=images.go:480 msg="total unused blobs removed: 0"
time=2026-02-15T19:55:36.332-07:00 level=INFO source=routes.go:1689 msg="Listening on 127.0.0.1:11434 (version 0.16.1)"
time=2026-02-15T19:55:36.334-07:00 level=INFO source=runner.go:67 msg="discovering available GPUs..."
time=2026-02-15T19:55:36.363-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64278"
time=2026-02-15T19:55:36.991-07:00 level=INFO source=runner.go:106 msg="experimental Vulkan support disabled.  To enable, set OLLAMA_VULKAN=1"
time=2026-02-15T19:55:36.992-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64285"
time=2026-02-15T19:55:45.615-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64297"
time=2026-02-15T19:55:45.918-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64303"
time=2026-02-15T19:55:45.918-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64302"
time=2026-02-15T19:55:46.104-07:00 level=INFO source=types.go:42 msg="inference compute" id=GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d filter_id="" library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Ti" libdirs=ollama,cuda_v13 driver=13.1 pci_id=0000:54:00.0 type=discrete total="12.0 GiB" available="10.6 GiB"
time=2026-02-15T19:55:46.104-07:00 level=INFO source=routes.go:1739 msg="vram-based default context" total_vram="12.0 GiB" default_num_ctx=4096
[GIN] 2026/02/15 - 19:55:46 | 200 |       526.9µs |       127.0.0.1 | HEAD     "/"
[GIN] 2026/02/15 - 19:55:46 | 200 |       526.9µs |       127.0.0.1 | GET      "/api/version"
[GIN] 2026/02/15 - 19:55:46 | 200 |       524.8µs |       127.0.0.1 | GET      "/api/tags"
[GIN] 2026/02/15 - 19:56:10 | 200 |            0s |       127.0.0.1 | GET      "/api/tags"
time=2026-02-15T20:04:22.276-07:00 level=INFO source=download.go:179 msg="downloading 9eba2761cf0b in 20 1 GB part(s)"
time=2026-02-15T20:07:27.157-07:00 level=INFO source=download.go:179 msg="downloading b1bca6ec8117 in 1 1.1 KB part(s)"
time=2026-02-15T20:07:28.497-07:00 level=INFO source=download.go:179 msg="downloading d8ba2f9a17b3 in 1 18 B part(s)"
time=2026-02-15T20:07:29.814-07:00 level=INFO source=download.go:179 msg="downloading 49d4bd6d5a04 in 1 486 B part(s)"
[GIN] 2026/02/15 - 20:07:51 | 200 |         3m30s |       127.0.0.1 | POST     "/api/pull"
[GIN] 2026/02/15 - 20:07:52 | 200 |    201.2473ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2026/02/15 - 20:07:52 | 200 |     204.184ms |       127.0.0.1 | POST     "/api/show"
time=2026-02-15T20:07:52.562-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 6655"
time=2026-02-15T20:07:52.747-07:00 level=INFO source=cpu_windows.go:148 msg=packages count=1
time=2026-02-15T20:07:52.747-07:00 level=INFO source=cpu_windows.go:195 msg="" package=0 cores=16 efficiency=0 threads=32
time=2026-02-15T20:07:52.848-07:00 level=INFO source=server.go:247 msg="enabling flash attention"
time=2026-02-15T20:07:52.849-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\mitcht\\.ollama\\models\\blobs\\sha256-9eba2761cf0b88b8bc11a065a7b5b47f1b13ce820e8e492cb1010b450f9ec950 --port 6661"
time=2026-02-15T20:07:52.852-07:00 level=INFO source=sched.go:463 msg="system memory" total="127.9 GiB" free="101.9 GiB" free_swap="107.6 GiB"
time=2026-02-15T20:07:52.852-07:00 level=INFO source=sched.go:470 msg="gpu memory" id=GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d library=CUDA available="9.9 GiB" free="10.3 GiB" minimum="457.0 MiB" overhead="0 B"
time=2026-02-15T20:07:52.852-07:00 level=INFO source=server.go:757 msg="loading model" "model layers"=48 requested=-1
time=2026-02-15T20:07:52.889-07:00 level=INFO source=runner.go:1411 msg="starting ollama engine"
time=2026-02-15T20:07:52.890-07:00 level=INFO source=runner.go:1446 msg="Server listening on 127.0.0.1:6661"
time=2026-02-15T20:07:52.896-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:48[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:48(0..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2026-02-15T20:07:52.931-07:00 level=INFO source=ggml.go:136 msg="" architecture=glm4moelite file_type=Q4_K_M name="" description="" num_tensors=844 num_key_values=39
load_backend: loaded CPU backend from C:\Users\mitcht\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-haswell.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Ti, compute capability 8.9, VMM: yes, ID: GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d
load_backend: loaded CUDA backend from C:\Users\mitcht\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2026-02-15T20:07:53.023-07:00 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.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1030,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2026-02-15T20:08:43.805-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2026-02-15T20:08:43.863-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2026-02-15T20:08:44.394-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=ggml.go:482 msg="offloading 24 repeating layers to GPU"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=ggml.go:486 msg="offloading output layer to CPU"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=ggml.go:494 msg="offloaded 24/48 layers to GPU"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:240 msg="model weights" device=CUDA0 size="9.3 GiB"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:245 msg="model weights" device=CPU size="8.4 GiB"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:251 msg="kv cache" device=CUDA0 size="204.0 MiB"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:256 msg="kv cache" device=CPU size="195.5 MiB"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:262 msg="compute graph" device=CUDA0 size="272.0 MiB"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:267 msg="compute graph" device=CPU size="4.0 MiB"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:272 msg="total memory" size="18.4 GiB"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=sched.go:537 msg="loaded runners" count=1
time=2026-02-15T20:08:44.395-07:00 level=INFO source=server.go:1350 msg="waiting for llama runner to start responding"
time=2026-02-15T20:08:44.395-07:00 level=INFO source=server.go:1384 msg="waiting for server to become available" status="llm server loading model"
time=2026-02-15T20:08:47.404-07:00 level=INFO source=server.go:1388 msg="llama runner started in 54.55 seconds"
[GIN] 2026/02/15 - 20:08:47 | 200 |   55.0765122s |       127.0.0.1 | POST     "/api/generate"
ggml_backend_cuda_device_get_memory device GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d utilizing NVML memory reporting free: 312586240 total: 12878610432
time=2026-02-15T20:13:47.792-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 54417"
time=2026-02-15T20:17:20.986-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 9549"
time=2026-02-15T20:17:21.145-07:00 level=INFO source=cpu_windows.go:148 msg=packages count=1
time=2026-02-15T20:17:21.145-07:00 level=INFO source=cpu_windows.go:195 msg="" package=0 cores=16 efficiency=0 threads=32
time=2026-02-15T20:17:21.232-07:00 level=INFO source=server.go:247 msg="enabling flash attention"
time=2026-02-15T20:17:21.233-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\mitcht\\.ollama\\models\\blobs\\sha256-9eba2761cf0b88b8bc11a065a7b5b47f1b13ce820e8e492cb1010b450f9ec950 --port 9554"
time=2026-02-15T20:17:21.237-07:00 level=INFO source=sched.go:463 msg="system memory" total="127.9 GiB" free="101.2 GiB" free_swap="106.8 GiB"
time=2026-02-15T20:17:21.237-07:00 level=INFO source=sched.go:470 msg="gpu memory" id=GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d library=CUDA available="9.9 GiB" free="10.3 GiB" minimum="457.0 MiB" overhead="0 B"
time=2026-02-15T20:17:21.237-07:00 level=INFO source=server.go:757 msg="loading model" "model layers"=48 requested=-1
time=2026-02-15T20:17:21.271-07:00 level=INFO source=runner.go:1411 msg="starting ollama engine"
time=2026-02-15T20:17:21.272-07:00 level=INFO source=runner.go:1446 msg="Server listening on 127.0.0.1:9554"
time=2026-02-15T20:17:21.280-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:48[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:48(0..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2026-02-15T20:17:21.314-07:00 level=INFO source=ggml.go:136 msg="" architecture=glm4moelite file_type=Q4_K_M name="" description="" num_tensors=844 num_key_values=39
load_backend: loaded CPU backend from C:\Users\mitcht\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-haswell.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Ti, compute capability 8.9, VMM: yes, ID: GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d
load_backend: loaded CUDA backend from C:\Users\mitcht\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2026-02-15T20:17:21.401-07:00 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.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1030,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2026-02-15T20:17:21.681-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2026-02-15T20:17:21.733-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=ggml.go:482 msg="offloading 24 repeating layers to GPU"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=ggml.go:486 msg="offloading output layer to CPU"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=ggml.go:494 msg="offloaded 24/48 layers to GPU"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:240 msg="model weights" device=CUDA0 size="9.3 GiB"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:245 msg="model weights" device=CPU size="8.4 GiB"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:251 msg="kv cache" device=CUDA0 size="204.0 MiB"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:256 msg="kv cache" device=CPU size="195.5 MiB"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:262 msg="compute graph" device=CUDA0 size="272.0 MiB"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:267 msg="compute graph" device=CPU size="4.0 MiB"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:272 msg="total memory" size="18.4 GiB"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=sched.go:537 msg="loaded runners" count=1
time=2026-02-15T20:17:21.994-07:00 level=INFO source=server.go:1350 msg="waiting for llama runner to start responding"
time=2026-02-15T20:17:21.994-07:00 level=INFO source=server.go:1384 msg="waiting for server to become available" status="llm server loading model"
time=2026-02-15T20:17:24.752-07:00 level=INFO source=server.go:1388 msg="llama runner started in 3.52 seconds"
[GIN] 2026/02/15 - 20:17:44 | 200 |   23.4476178s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2026/02/15 - 20:19:38 | 200 |         1m25s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2026/02/15 - 20:22:40 | 200 |         1m52s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2026/02/15 - 20:26:43 | 200 |         1m46s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2026/02/15 - 20:32:41 | 200 |         1m42s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2026/02/15 - 20:33:11 | 500 |    2.6082429s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2026/02/15 - 20:42:08 | 200 |         8m43s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2026/02/15 - 20:42:29 | 200 |   13.4512325s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2026/02/15 - 20:42:41 | 200 |    2.3295111s |       127.0.0.1 | POST     "/api/chat"

OS

Windows

GPU

Nvidia

CPU

AMD

Ollama version

0.16.1

Originally created by @MitchTalmadge on GitHub (Feb 16, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/14273 ### What is the issue? I have just installed Ollama on my Windows desktop and chose the [glm-4.7-flash](https://ollama.com/library/glm-4.7-flash) model. While having it write a poem, it started outputting complete gibberish with no end in sight. I eventually stopped it. Then, all prompts, even after running `/clear`, would output this gibberish. After running `/bye` and starting it again, it works fine again. - CPU: Ryzen 9 5950x - GPU: RTX 4070Ti - RAM: 128GB DDR4 Here is when it started; see the line after "Stanza 5: Conclusion" ``` >>> /clear Cleared session context >>> Write a poem about chocolate Thinking... 1. **Analyze the Request:** * **Topic:** Chocolate. * **Format:** Poem. * **Tone:** Open, but usually implies appreciation, indulgence, or sensory description. 2. **Brainstorming Imagery & Themes:** * *Visual:* Dark, glossy, liquid, swirling, melting, dust of cocoa, velvety. * *Tactile:* Smooth, silky, melting on the tongue, cool in the hand, warm in the mouth. * *Taste:* Bitter, sweet, rich, earthy, decadent, forbidden, comforting. * *Emotional:* Joy, warmth, escapism, love, comfort, ritual. * *Associations:* Cacao beans, melting pot, gift, afternoon treat, dessert. 3. **Drafting - Stanza by Stanza:** * *Stanza 1: Introduction/The Origin (Earthiness).* * Start with the raw material. Beans from the tropics. * *Drafting ideas:* It comes from the sun, dark beans. Earthy soul. * *Refining:* "The soul of the earth in a dusty pod," "Forged in the heat." * *Stanza 2: The Transformation (Processing).* * Roasting and grinding. * *Drafting ideas:* Roasted to perfection, crushed into dust, liquid gold. * *Refining:* "Turned to liquid gold," "The roast and the crush." * *Stanza 3: The Sensory Experience (Taste/Texture).* * Eating it. The contrast of sweet and bitter. * *Drafting ideas:* Melts on the tongue, velvety coat, sweet rush. * *Refining:* "Velvet on the tongue," "The sweet that bites back," "Decadence." * *Stanza 4: The Emotional Connection (Comfort).* * Why we eat it. The feeling of warmth. * *Drafting ideas:* A hug for the soul, rainy day comfort, little piece of joy. * *Refining:* "A hug for the soul," "A little piece of the sun." * *Stanza 5: Conclusion.* * Summary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ello Thinking... =E<+;4*A3C@CD*..*5F0"(2$@.*"6A(@,#;>52F65;:632E"/9;B;2*F31?#'2;>2+E/?"DE+;0'$25.:C;)D&42#)<8B=&651EC)1%=F97)3,7'#-$'@%,406C)*67>%4@ >>> /clear Cleared session context >>> Hello Thinking... 7=>81*<6FE@33<D7,5%7E4;,'0@+&(.8,3$0 >>> Send a message (/? for help) ``` ### Relevant log output ```shell server.log: time=2026-02-15T19:55:36.330-07:00 level=INFO source=routes.go:1636 msg="server config" env="map[CUDA_VISIBLE_DEVICES: GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_EDITOR: OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 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_MODELS:C:\\Users\\mitcht\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false 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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES:]" time=2026-02-15T19:55:36.332-07:00 level=INFO source=images.go:473 msg="total blobs: 0" time=2026-02-15T19:55:36.332-07:00 level=INFO source=images.go:480 msg="total unused blobs removed: 0" time=2026-02-15T19:55:36.332-07:00 level=INFO source=routes.go:1689 msg="Listening on 127.0.0.1:11434 (version 0.16.1)" time=2026-02-15T19:55:36.334-07:00 level=INFO source=runner.go:67 msg="discovering available GPUs..." time=2026-02-15T19:55:36.363-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64278" time=2026-02-15T19:55:36.991-07:00 level=INFO source=runner.go:106 msg="experimental Vulkan support disabled. To enable, set OLLAMA_VULKAN=1" time=2026-02-15T19:55:36.992-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64285" time=2026-02-15T19:55:45.615-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64297" time=2026-02-15T19:55:45.918-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64303" time=2026-02-15T19:55:45.918-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 64302" time=2026-02-15T19:55:46.104-07:00 level=INFO source=types.go:42 msg="inference compute" id=GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d filter_id="" library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Ti" libdirs=ollama,cuda_v13 driver=13.1 pci_id=0000:54:00.0 type=discrete total="12.0 GiB" available="10.6 GiB" time=2026-02-15T19:55:46.104-07:00 level=INFO source=routes.go:1739 msg="vram-based default context" total_vram="12.0 GiB" default_num_ctx=4096 [GIN] 2026/02/15 - 19:55:46 | 200 | 526.9µs | 127.0.0.1 | HEAD "/" [GIN] 2026/02/15 - 19:55:46 | 200 | 526.9µs | 127.0.0.1 | GET "/api/version" [GIN] 2026/02/15 - 19:55:46 | 200 | 524.8µs | 127.0.0.1 | GET "/api/tags" [GIN] 2026/02/15 - 19:56:10 | 200 | 0s | 127.0.0.1 | GET "/api/tags" time=2026-02-15T20:04:22.276-07:00 level=INFO source=download.go:179 msg="downloading 9eba2761cf0b in 20 1 GB part(s)" time=2026-02-15T20:07:27.157-07:00 level=INFO source=download.go:179 msg="downloading b1bca6ec8117 in 1 1.1 KB part(s)" time=2026-02-15T20:07:28.497-07:00 level=INFO source=download.go:179 msg="downloading d8ba2f9a17b3 in 1 18 B part(s)" time=2026-02-15T20:07:29.814-07:00 level=INFO source=download.go:179 msg="downloading 49d4bd6d5a04 in 1 486 B part(s)" [GIN] 2026/02/15 - 20:07:51 | 200 | 3m30s | 127.0.0.1 | POST "/api/pull" [GIN] 2026/02/15 - 20:07:52 | 200 | 201.2473ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/02/15 - 20:07:52 | 200 | 204.184ms | 127.0.0.1 | POST "/api/show" time=2026-02-15T20:07:52.562-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 6655" time=2026-02-15T20:07:52.747-07:00 level=INFO source=cpu_windows.go:148 msg=packages count=1 time=2026-02-15T20:07:52.747-07:00 level=INFO source=cpu_windows.go:195 msg="" package=0 cores=16 efficiency=0 threads=32 time=2026-02-15T20:07:52.848-07:00 level=INFO source=server.go:247 msg="enabling flash attention" time=2026-02-15T20:07:52.849-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\mitcht\\.ollama\\models\\blobs\\sha256-9eba2761cf0b88b8bc11a065a7b5b47f1b13ce820e8e492cb1010b450f9ec950 --port 6661" time=2026-02-15T20:07:52.852-07:00 level=INFO source=sched.go:463 msg="system memory" total="127.9 GiB" free="101.9 GiB" free_swap="107.6 GiB" time=2026-02-15T20:07:52.852-07:00 level=INFO source=sched.go:470 msg="gpu memory" id=GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d library=CUDA available="9.9 GiB" free="10.3 GiB" minimum="457.0 MiB" overhead="0 B" time=2026-02-15T20:07:52.852-07:00 level=INFO source=server.go:757 msg="loading model" "model layers"=48 requested=-1 time=2026-02-15T20:07:52.889-07:00 level=INFO source=runner.go:1411 msg="starting ollama engine" time=2026-02-15T20:07:52.890-07:00 level=INFO source=runner.go:1446 msg="Server listening on 127.0.0.1:6661" time=2026-02-15T20:07:52.896-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:48[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:48(0..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2026-02-15T20:07:52.931-07:00 level=INFO source=ggml.go:136 msg="" architecture=glm4moelite file_type=Q4_K_M name="" description="" num_tensors=844 num_key_values=39 load_backend: loaded CPU backend from C:\Users\mitcht\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-haswell.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Ti, compute capability 8.9, VMM: yes, ID: GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d load_backend: loaded CUDA backend from C:\Users\mitcht\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2026-02-15T20:07:53.023-07:00 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.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1030,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2026-02-15T20:08:43.805-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2026-02-15T20:08:43.863-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2026-02-15T20:08:44.394-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2026-02-15T20:08:44.395-07:00 level=INFO source=ggml.go:482 msg="offloading 24 repeating layers to GPU" time=2026-02-15T20:08:44.395-07:00 level=INFO source=ggml.go:486 msg="offloading output layer to CPU" time=2026-02-15T20:08:44.395-07:00 level=INFO source=ggml.go:494 msg="offloaded 24/48 layers to GPU" time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:240 msg="model weights" device=CUDA0 size="9.3 GiB" time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:245 msg="model weights" device=CPU size="8.4 GiB" time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:251 msg="kv cache" device=CUDA0 size="204.0 MiB" time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:256 msg="kv cache" device=CPU size="195.5 MiB" time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:262 msg="compute graph" device=CUDA0 size="272.0 MiB" time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:267 msg="compute graph" device=CPU size="4.0 MiB" time=2026-02-15T20:08:44.395-07:00 level=INFO source=device.go:272 msg="total memory" size="18.4 GiB" time=2026-02-15T20:08:44.395-07:00 level=INFO source=sched.go:537 msg="loaded runners" count=1 time=2026-02-15T20:08:44.395-07:00 level=INFO source=server.go:1350 msg="waiting for llama runner to start responding" time=2026-02-15T20:08:44.395-07:00 level=INFO source=server.go:1384 msg="waiting for server to become available" status="llm server loading model" time=2026-02-15T20:08:47.404-07:00 level=INFO source=server.go:1388 msg="llama runner started in 54.55 seconds" [GIN] 2026/02/15 - 20:08:47 | 200 | 55.0765122s | 127.0.0.1 | POST "/api/generate" ggml_backend_cuda_device_get_memory device GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d utilizing NVML memory reporting free: 312586240 total: 12878610432 time=2026-02-15T20:13:47.792-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 54417" time=2026-02-15T20:17:20.986-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --port 9549" time=2026-02-15T20:17:21.145-07:00 level=INFO source=cpu_windows.go:148 msg=packages count=1 time=2026-02-15T20:17:21.145-07:00 level=INFO source=cpu_windows.go:195 msg="" package=0 cores=16 efficiency=0 threads=32 time=2026-02-15T20:17:21.232-07:00 level=INFO source=server.go:247 msg="enabling flash attention" time=2026-02-15T20:17:21.233-07:00 level=INFO source=server.go:431 msg="starting runner" cmd="C:\\Users\\mitcht\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\mitcht\\.ollama\\models\\blobs\\sha256-9eba2761cf0b88b8bc11a065a7b5b47f1b13ce820e8e492cb1010b450f9ec950 --port 9554" time=2026-02-15T20:17:21.237-07:00 level=INFO source=sched.go:463 msg="system memory" total="127.9 GiB" free="101.2 GiB" free_swap="106.8 GiB" time=2026-02-15T20:17:21.237-07:00 level=INFO source=sched.go:470 msg="gpu memory" id=GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d library=CUDA available="9.9 GiB" free="10.3 GiB" minimum="457.0 MiB" overhead="0 B" time=2026-02-15T20:17:21.237-07:00 level=INFO source=server.go:757 msg="loading model" "model layers"=48 requested=-1 time=2026-02-15T20:17:21.271-07:00 level=INFO source=runner.go:1411 msg="starting ollama engine" time=2026-02-15T20:17:21.272-07:00 level=INFO source=runner.go:1446 msg="Server listening on 127.0.0.1:9554" time=2026-02-15T20:17:21.280-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:48[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:48(0..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2026-02-15T20:17:21.314-07:00 level=INFO source=ggml.go:136 msg="" architecture=glm4moelite file_type=Q4_K_M name="" description="" num_tensors=844 num_key_values=39 load_backend: loaded CPU backend from C:\Users\mitcht\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-haswell.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Ti, compute capability 8.9, VMM: yes, ID: GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d load_backend: loaded CUDA backend from C:\Users\mitcht\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2026-02-15T20:17:21.401-07:00 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.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1030,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2026-02-15T20:17:21.681-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2026-02-15T20:17:21.733-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2026-02-15T20:17:21.994-07:00 level=INFO source=runner.go:1284 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:16 GPULayers:24[ID:GPU-dfbb6591-91db-65e5-ea99-49e51d733d1d Layers:24(23..46)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2026-02-15T20:17:21.994-07:00 level=INFO source=ggml.go:482 msg="offloading 24 repeating layers to GPU" time=2026-02-15T20:17:21.994-07:00 level=INFO source=ggml.go:486 msg="offloading output layer to CPU" time=2026-02-15T20:17:21.994-07:00 level=INFO source=ggml.go:494 msg="offloaded 24/48 layers to GPU" time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:240 msg="model weights" device=CUDA0 size="9.3 GiB" time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:245 msg="model weights" device=CPU size="8.4 GiB" time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:251 msg="kv cache" device=CUDA0 size="204.0 MiB" time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:256 msg="kv cache" device=CPU size="195.5 MiB" time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:262 msg="compute graph" device=CUDA0 size="272.0 MiB" time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:267 msg="compute graph" device=CPU size="4.0 MiB" time=2026-02-15T20:17:21.994-07:00 level=INFO source=device.go:272 msg="total memory" size="18.4 GiB" time=2026-02-15T20:17:21.994-07:00 level=INFO source=sched.go:537 msg="loaded runners" count=1 time=2026-02-15T20:17:21.994-07:00 level=INFO source=server.go:1350 msg="waiting for llama runner to start responding" time=2026-02-15T20:17:21.994-07:00 level=INFO source=server.go:1384 msg="waiting for server to become available" status="llm server loading model" time=2026-02-15T20:17:24.752-07:00 level=INFO source=server.go:1388 msg="llama runner started in 3.52 seconds" [GIN] 2026/02/15 - 20:17:44 | 200 | 23.4476178s | 127.0.0.1 | POST "/api/chat" [GIN] 2026/02/15 - 20:19:38 | 200 | 1m25s | 127.0.0.1 | POST "/api/chat" [GIN] 2026/02/15 - 20:22:40 | 200 | 1m52s | 127.0.0.1 | POST "/api/chat" [GIN] 2026/02/15 - 20:26:43 | 200 | 1m46s | 127.0.0.1 | POST "/api/chat" [GIN] 2026/02/15 - 20:32:41 | 200 | 1m42s | 127.0.0.1 | POST "/api/chat" [GIN] 2026/02/15 - 20:33:11 | 500 | 2.6082429s | 127.0.0.1 | POST "/api/chat" [GIN] 2026/02/15 - 20:42:08 | 200 | 8m43s | 127.0.0.1 | POST "/api/chat" [GIN] 2026/02/15 - 20:42:29 | 200 | 13.4512325s | 127.0.0.1 | POST "/api/chat" [GIN] 2026/02/15 - 20:42:41 | 200 | 2.3295111s | 127.0.0.1 | POST "/api/chat" ``` ### OS Windows ### GPU Nvidia ### CPU AMD ### Ollama version 0.16.1
GiteaMirror added the bug label 2026-04-12 22:09:39 -05:00
GiteaMirror commented 2026-04-12 22:09:40 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 16, 2026):

Before you ran /clear, had there been a lot of input/output?

<!-- gh-comment-id:3910188108 --> @rick-github commented on GitHub (Feb 16, 2026): Before you ran `/clear`, had there been a lot of input/output?
GiteaMirror commented 2026-04-12 22:09:40 -05:00
Author
Owner
Copy Link

@MitchTalmadge commented on GitHub (Feb 17, 2026):

Not a ton. I had it generate a bit of code but it wasn't excessive. It hasn't done this again though.

<!-- gh-comment-id:3912202344 --> @MitchTalmadge commented on GitHub (Feb 17, 2026): Not a ton. I had it generate a bit of code but it wasn't excessive. It hasn't done this again though.
GiteaMirror commented 2026-04-12 22:09:41 -05:00
Author
Owner
Copy Link

@3abdelazim commented on GitHub (Feb 18, 2026):

What context size do you see when running ollama ps?

<!-- gh-comment-id:3923734529 --> @3abdelazim commented on GitHub (Feb 18, 2026): What context size do you see when running `ollama ps`?
GiteaMirror commented 2026-04-12 22:09:42 -05:00
Author
Owner
Copy Link

@MitchTalmadge commented on GitHub (Mar 18, 2026):

I'm sorry guys, I tried to reproduce this but couldn't get it to happen again. I'll close this for now, if anyone else experiences it maybe it can be reopened.

<!-- gh-comment-id:4078987974 --> @MitchTalmadge commented on GitHub (Mar 18, 2026): I'm sorry guys, I tried to reproduce this but couldn't get it to happen again. I'll close this for now, if anyone else experiences it maybe it can be reopened.
GiteaMirror commented 2026-04-12 22:09:42 -05:00
Author
Owner
Copy Link

@NoMansPC commented on GitHub (Apr 10, 2026):

I know this issue is closed, but it keeps happening to me and I don't know why. When it works, it's so great, but randomly, it just stops working.

<!-- gh-comment-id:4225353089 --> @NoMansPC commented on GitHub (Apr 10, 2026): I know this issue is closed, but it keeps happening to me and I don't know why. When it works, it's so great, but randomly, it just stops working.
GiteaMirror commented 2026-04-12 22:09:43 -05:00
Author
Owner
Copy Link

@MitchTalmadge commented on GitHub (Apr 10, 2026):

Good to see I'm not alone! I'll try using the model some more to see if it happens again

<!-- gh-comment-id:4226019726 --> @MitchTalmadge commented on GitHub (Apr 10, 2026): Good to see I'm not alone! I'll try using the model some more to see if it happens again
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#9294
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?