What happened?
I was trying to get qwen code running with a local instance of the small coder model. I can see the model responds with what look like appropriate tool calls to me, but the tool calls seem to not be executed. Unfortunately I don't see any errors, so I don't why they don't work as expected.
If needed, I can provide further logs of the server requests.
What did you expect to happen?
When I ask qwen code to init a git repository and the response from the locally hosted server is:
[format_partial_response_oaicompat] DEBUG: Streaming finish_reason check | tid="7532" timestamp=1754063577 generated_text="<tool_call>\n<function=run_shell_command>\n<parameter=command>\ngit init\n</parameter>\n<parameter=description>\nInitializing a new Git repository in the current directory\n</parameter>\n</function>\n</tool_call>" model_name="qwen3-coder-30b-a3b" tool_calls_count=0
qwen code should recognize the command and execute it.
Client information
Details
$ qwen /about
│ About Qwen Code │
│ │
│ CLI Version 0.0.2 │
│ Git Commit 2786f10 (local modifications) │
│ Model qwen3-coder-30b-a3b │
│ Sandbox no sandbox │
│ OS win32 │
│ Auth Method openai
Login information
Running against locally hosted server with ik_llama.cpp with these environment variables:
$Env:OPENAI_BASE_URL='http://127.0.0.1:8080/v1';
$Env:OPENAI_API_KEY='local';
$Env:OPENAI_MODEL='qwen3-coder-30b-a3b';
Anything else we need to know?
No response
What happened?
I was trying to get qwen code running with a local instance of the small coder model. I can see the model responds with what look like appropriate tool calls to me, but the tool calls seem to not be executed. Unfortunately I don't see any errors, so I don't why they don't work as expected.
If needed, I can provide further logs of the server requests.
What did you expect to happen?
When I ask qwen code to init a git repository and the response from the locally hosted server is:
[format_partial_response_oaicompat] DEBUG: Streaming finish_reason check | tid="7532" timestamp=1754063577 generated_text="<tool_call>\n<function=run_shell_command>\n<parameter=command>\ngit init\n</parameter>\n<parameter=description>\nInitializing a new Git repository in the current directory\n</parameter>\n</function>\n</tool_call>" model_name="qwen3-coder-30b-a3b" tool_calls_count=0qwen code should recognize the command and execute it.
Client information
Details
Login information
Running against locally hosted server with ik_llama.cpp with these environment variables:
Anything else we need to know?
No response