github-starred/ollama-ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-04-27 16:36:54 -05:00
Code Issues 2.2k Packages Projects Releases 100 Wiki Activity

qwen3-coder:30b (qwen3moe) loses tool capability by creating a new model #8408

New Issue
Closed
opened 2025-11-12 14:41:08 -06:00 by GiteaMirror · 4 comments
GiteaMirror commented 2025-11-12 14:41:08 -06:00
Owner
Copy Link

Originally created by @dan-and on GitHub (Oct 16, 2025).

What is the issue?

Regardless if I create a simple Modelfile like:
Modelfile
FROM qwen3-coder:30b
PARAMETER num_ctx 262144

or do the same in the ollama cli, the created model is missing the tools capability. I don't have the same issue with other models (like gptoss, devstral etc.)

$ ollama show qwen3-coder:30b
Model
architecture qwen3moe
parameters 30.5B
context length 262144
embedding length 2048
quantization Q4_K_M

Capabilities
completion
tools

Parameters
stop "<|im_start|>"
stop "<|im_end|>"
stop "<|endoftext|>"
temperature 0.7
top_k 20
top_p 0.8
repeat_penalty 1.05

License
Apache License
Version 2.0, January 2004
...

$ ollama run qwen3-coder:30b

/set parameter num_ctx 262144
Set parameter 'num_ctx' to '262144'
/save qwen3-coder:30b_with_260k_context
Created new model 'qwen3-coder:30b_with_260k_context'

$ ollama show qwen3-coder:30b_with_260k_context
Model
architecture qwen3moe
parameters 30.5B
context length 262144
embedding length 2048
quantization Q4_K_M

Capabilities
completion

Parameters
num_ctx 262144
repeat_penalty 1.05
stop "<|im_start|>"
stop "<|im_end|>"
stop "<|endoftext|>"
temperature 0.7
top_k 20
top_p 0.8

License
Apache License
Version 2.0, January 2004
...

Relevant log output

$ cat create_model.log

Oct 16 14:19:18 gpu ollama[17824]: [GIN] 2025/10/16 - 14:19:18 | 200 | 50.627146765s |       127.0.0.1 | POST     "/api/generate"
Oct 16 14:19:18 gpu ollama[17824]: time=2025-10-16T14:19:18.638Z level=DEBUG source=sched.go:502 msg="context for request finished"
Oct 16 14:19:18 gpu ollama[17824]: time=2025-10-16T14:19:18.638Z level=DEBUG source=sched.go:294 msg="runner with non-zero duration has gone idle, adding timer" runner.name=registry.ollama.ai/library/qwen3-coder:30b runner.inference="[{ID:GPU-bd86bdbc-e249-0a40-fc1f-6284c27af2cf Library:CUDA} {ID:GPU-c863cd65-236e-6ee5-cafc-02039c50b3b5 Library:CUDA}]" runner.size="18.3 GiB" runner.vram="18.3 GiB" runner.parallel=2 runner.pid=34673 runner.model=/usr/share/ollama/.ollama/models/blobs/sha256-1194192cf2a187eb02722edcc3f77b11d21f537048ce04b67ccf8ba78863006a runner.num_ctx=4096 duration=1h0m0s
Oct 16 14:19:18 gpu ollama[17824]: time=2025-10-16T14:19:18.638Z level=DEBUG source=sched.go:312 msg="after processing request finished event" runner.name=registry.ollama.ai/library/qwen3-coder:30b runner.inference="[{ID:GPU-bd86bdbc-e249-0a40-fc1f-6284c27af2cf Library:CUDA} {ID:GPU-c863cd65-236e-6ee5-cafc-02039c50b3b5 Library:CUDA}]" runner.size="18.3 GiB" runner.vram="18.3 GiB" runner.parallel=2 runner.pid=34673 runner.model=/usr/share/ollama/.ollama/models/blobs/sha256-1194192cf2a187eb02722edcc3f77b11d21f537048ce04b67ccf8ba78863006a runner.num_ctx=4096 refCount=0
Oct 16 14:19:54 gpu ollama[17824]: time=2025-10-16T14:19:54.204Z level=DEBUG source=create.go:98 msg="create model from model name" from=qwen3-coder:30b
Oct 16 14:19:54 gpu ollama[17824]: time=2025-10-16T14:19:54.224Z level=DEBUG source=ggml.go:276 msg="key with type not found" key=general.alignment default=32
Oct 16 14:19:54 gpu ollama[17824]: [GIN] 2025/10/16 - 14:19:54 | 200 |   28.196158ms |       127.0.0.1 | POST     "/api/create"
Oct 16 14:20:07 gpu ollama[17824]: [GIN] 2025/10/16 - 14:20:07 | 200 |      19.576µs |       127.0.0.1 | HEAD     "/"
Oct 16 14:20:08 gpu ollama[17824]: time=2025-10-16T14:20:08.009Z level=DEBUG source=ggml.go:276 msg="key with type not found" key=general.alignment default=32
Oct 16 14:20:08 gpu ollama[17824]: [GIN] 2025/10/16 - 14:20:08 | 200 |    41.16944ms |       127.0.0.1 | POST     "/api/show"
Oct 16 14:25:02 gpu ollama[17824]: [GIN] 2025/10/16 - 14:25:02 | 200 |      23.694µs |       127.0.0.1 | HEAD     "/"
Oct 16 14:25:02 gpu ollama[17824]: time=2025-10-16T14:25:02.830Z level=DEBUG source=create.go:98 msg="create model from model name" from=qwen3-coder:30b
Oct 16 14:25:02 gpu ollama[17824]: time=2025-10-16T14:25:02.848Z level=DEBUG source=ggml.go:276 msg="key with type not found" key=general.alignment default=32
Oct 16 14:25:02 gpu ollama[17824]: [GIN] 2025/10/16 - 14:25:02 | 200 |    27.19669ms |       127.0.0.1 | POST     "/api/create"
Oct 16 14:25:12 gpu ollama[17824]: [GIN] 2025/10/16 - 14:25:12 | 200 |      20.148µs |       127.0.0.1 | HEAD     "/"
Oct 16 14:25:12 gpu ollama[17824]: time=2025-10-16T14:25:12.134Z level=DEBUG source=ggml.go:276 msg="key with type not found" key=general.alignment default=32
Oct 16 14:25:12 gpu ollama[17824]: [GIN] 2025/10/16 - 14:25:12 | 200 |   39.086398ms |       127.0.0.1 | POST     "/api/show"

OS

Linux

GPU

Nvidia

CPU

AMD

Ollama version

0.12.5

Originally created by @dan-and on GitHub (Oct 16, 2025). ### What is the issue? Regardless if I create a simple Modelfile like: Modelfile FROM qwen3-coder:30b PARAMETER num_ctx 262144 or do the same in the ollama cli, the created model is missing the tools capability. I don't have the same issue with other models (like gptoss, devstral etc.) $ ollama show qwen3-coder:30b Model architecture qwen3moe parameters 30.5B context length 262144 embedding length 2048 quantization Q4_K_M **Capabilities completion tools** Parameters stop "<|im_start|>" stop "<|im_end|>" stop "<|endoftext|>" temperature 0.7 top_k 20 top_p 0.8 repeat_penalty 1.05 License Apache License Version 2.0, January 2004 ... $ ollama run qwen3-coder:30b >>> /set parameter num_ctx 262144 Set parameter 'num_ctx' to '262144' >>> /save qwen3-coder:30b_with_260k_context Created new model 'qwen3-coder:30b_with_260k_context' >>> $ ollama show qwen3-coder:30b_with_260k_context Model architecture qwen3moe parameters 30.5B context length 262144 embedding length 2048 quantization Q4_K_M **Capabilities completion** Parameters num_ctx 262144 repeat_penalty 1.05 stop "<|im_start|>" stop "<|im_end|>" stop "<|endoftext|>" temperature 0.7 top_k 20 top_p 0.8 License Apache License Version 2.0, January 2004 ... ### Relevant log output ```shell $ cat create_model.log Oct 16 14:19:18 gpu ollama[17824]: [GIN] 2025/10/16 - 14:19:18 | 200 | 50.627146765s | 127.0.0.1 | POST "/api/generate" Oct 16 14:19:18 gpu ollama[17824]: time=2025-10-16T14:19:18.638Z level=DEBUG source=sched.go:502 msg="context for request finished" Oct 16 14:19:18 gpu ollama[17824]: time=2025-10-16T14:19:18.638Z level=DEBUG source=sched.go:294 msg="runner with non-zero duration has gone idle, adding timer" runner.name=registry.ollama.ai/library/qwen3-coder:30b runner.inference="[{ID:GPU-bd86bdbc-e249-0a40-fc1f-6284c27af2cf Library:CUDA} {ID:GPU-c863cd65-236e-6ee5-cafc-02039c50b3b5 Library:CUDA}]" runner.size="18.3 GiB" runner.vram="18.3 GiB" runner.parallel=2 runner.pid=34673 runner.model=/usr/share/ollama/.ollama/models/blobs/sha256-1194192cf2a187eb02722edcc3f77b11d21f537048ce04b67ccf8ba78863006a runner.num_ctx=4096 duration=1h0m0s Oct 16 14:19:18 gpu ollama[17824]: time=2025-10-16T14:19:18.638Z level=DEBUG source=sched.go:312 msg="after processing request finished event" runner.name=registry.ollama.ai/library/qwen3-coder:30b runner.inference="[{ID:GPU-bd86bdbc-e249-0a40-fc1f-6284c27af2cf Library:CUDA} {ID:GPU-c863cd65-236e-6ee5-cafc-02039c50b3b5 Library:CUDA}]" runner.size="18.3 GiB" runner.vram="18.3 GiB" runner.parallel=2 runner.pid=34673 runner.model=/usr/share/ollama/.ollama/models/blobs/sha256-1194192cf2a187eb02722edcc3f77b11d21f537048ce04b67ccf8ba78863006a runner.num_ctx=4096 refCount=0 Oct 16 14:19:54 gpu ollama[17824]: time=2025-10-16T14:19:54.204Z level=DEBUG source=create.go:98 msg="create model from model name" from=qwen3-coder:30b Oct 16 14:19:54 gpu ollama[17824]: time=2025-10-16T14:19:54.224Z level=DEBUG source=ggml.go:276 msg="key with type not found" key=general.alignment default=32 Oct 16 14:19:54 gpu ollama[17824]: [GIN] 2025/10/16 - 14:19:54 | 200 | 28.196158ms | 127.0.0.1 | POST "/api/create" Oct 16 14:20:07 gpu ollama[17824]: [GIN] 2025/10/16 - 14:20:07 | 200 | 19.576µs | 127.0.0.1 | HEAD "/" Oct 16 14:20:08 gpu ollama[17824]: time=2025-10-16T14:20:08.009Z level=DEBUG source=ggml.go:276 msg="key with type not found" key=general.alignment default=32 Oct 16 14:20:08 gpu ollama[17824]: [GIN] 2025/10/16 - 14:20:08 | 200 | 41.16944ms | 127.0.0.1 | POST "/api/show" Oct 16 14:25:02 gpu ollama[17824]: [GIN] 2025/10/16 - 14:25:02 | 200 | 23.694µs | 127.0.0.1 | HEAD "/" Oct 16 14:25:02 gpu ollama[17824]: time=2025-10-16T14:25:02.830Z level=DEBUG source=create.go:98 msg="create model from model name" from=qwen3-coder:30b Oct 16 14:25:02 gpu ollama[17824]: time=2025-10-16T14:25:02.848Z level=DEBUG source=ggml.go:276 msg="key with type not found" key=general.alignment default=32 Oct 16 14:25:02 gpu ollama[17824]: [GIN] 2025/10/16 - 14:25:02 | 200 | 27.19669ms | 127.0.0.1 | POST "/api/create" Oct 16 14:25:12 gpu ollama[17824]: [GIN] 2025/10/16 - 14:25:12 | 200 | 20.148µs | 127.0.0.1 | HEAD "/" Oct 16 14:25:12 gpu ollama[17824]: time=2025-10-16T14:25:12.134Z level=DEBUG source=ggml.go:276 msg="key with type not found" key=general.alignment default=32 Oct 16 14:25:12 gpu ollama[17824]: [GIN] 2025/10/16 - 14:25:12 | 200 | 39.086398ms | 127.0.0.1 | POST "/api/show" ``` ### OS Linux ### GPU Nvidia ### CPU AMD ### Ollama version 0.12.5
GiteaMirror added the bug label 2025-11-12 14:41:08 -06:00
GiteaMirror commented 2025-11-12 14:41:10 -06:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 16, 2025):

FROM qwen3-coder:30b
PARAMETER num_ctx 262144
PARSER qwen3-coder
RENDERER qwen3-coder
@rick-github commented on GitHub (Oct 16, 2025): ``` FROM qwen3-coder:30b PARAMETER num_ctx 262144 PARSER qwen3-coder RENDERER qwen3-coder ```
GiteaMirror commented 2025-11-12 14:41:10 -06:00
Author
Owner
Copy Link

@dan-and commented on GitHub (Oct 16, 2025):

Thanks @rick-github
Is that a qwen3-coder issue only?

I thought it was enough to modify one parameter and be able to store that change into a new model.

@dan-and commented on GitHub (Oct 16, 2025): Thanks @rick-github Is that a qwen3-coder issue only? I thought it was enough to modify one parameter and be able to store that change into a new model.
GiteaMirror commented 2025-11-12 14:41:10 -06:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 16, 2025):

qwen3-coder has a different tool format that is not easily captured in the usual template, so a custom renderer/parser was written to make it a better tool user. As a result the modelfile currently needs to explicitly call out the renderer/parser requirements. I believe this will be automatic in the future, but for now these items need to be added.

@rick-github commented on GitHub (Oct 16, 2025): qwen3-coder has a different tool format that is not easily captured in the usual template, so a custom renderer/parser was written to make it a better tool user. As a result the modelfile currently needs to explicitly call out the renderer/parser requirements. I believe this will be automatic in the future, but for now these items need to be added.
GiteaMirror commented 2025-11-12 14:41:10 -06:00
Author
Owner
Copy Link

@dan-and commented on GitHub (Oct 16, 2025):

Thanks

@dan-and commented on GitHub (Oct 16, 2025): Thanks
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
client2
cloud
compatibility
context-length
create
docker
documentation
embeddings
engine
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
install
integration
intel
js
linux
macos
memory
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
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-ollama#8408
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?