Eval bug: Qwen3-Coder template parsing error when tools are enabled #19004
Description
Name and Version
❯ ./bin/llama-cli --version
ggml_cuda_init: found 1 ROCm devices:
Device 0: AMD Radeon 780M Graphics, gfx1100 (0x1100), VMM: no, Wave Size: 32
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = AMD Radeon 780M Graphics (RADV PHOENIX) (radv) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 64 | shared memory: 65536 | int dot: 1 | matrix cores: KHR_coopmat
version: 74 (94242a6)
built with GNU 15.2.1 for Linux x86_64
Operating systems
Linux
GGML backends
HIP
Hardware
AMD Ryzen 8840HS + 780M (as gfx1100 & gfx1103)
Models
Unsloth's Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf
Problem description & steps to reproduce
When tools are supplied to the Qwen3-Coder jinja template, it results in this error as reported below in the log output.
When no tools are supplied, template parsing works. When tools are supplied to the API, it results in a FilterExpression error.
First Bad Commit
Haven't done a bisect, but I suspect a recent jinja parser problem
Relevant log output
Logs
[41847] srv log_server_r: response: {"error":{"code":500,"message":"\n------------\nWhile executing FilterExpression at line 57, column 53 in source:\n...{%- for json_key in param_fields.keys() | reject(\"in\", handled_keys) %}↵ ...\n ^\nError: selectattr: unknown test 'in'","type":"server_error"}}
srv operator(): http: streamed chunk: {"error":{"code":500,"message":"\n------------\nWhile executing FilterExpression at line 57, column 53 in source:\n...{%- for json_key in param_fields.keys() | reject(\"in\", handled_keys) %}↵ ...\n ^\nError: selectattr: unknown test 'in'","type":"server_error"}}
srv operator(): http: stream ended
Example error response
{
"error": {
"name": "AI_APICallError",
"url": "http://127.0.0.1:8081/v1/chat/completions",
"requestBodyValues": {
"model": "Qwen3-Coder-30B-A3B-Instruct-Q4_K_M",
"max_tokens": 32000,
"top_p": 1,
"messages": [
{
"role": "system",
"content": "<snip>"
},
{
"role": "user",
"content": "example"
}
],
"tools": [
{
"type": "function",
"function": {
"name": "todoread",
"description": "Use this tool to read your todo list",
"parameters": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"type": "object",
"properties": {},
"additionalProperties": false
}
}
}
],
"tool_choice": "auto",
"stream": true,
"stream_options": {
"include_usage": true
}
},
"statusCode": 500,
"responseHeaders": {
"connection": "close, close",
"content-type": "application/json; charset=utf-8",
"server": "llama.cpp",
"transfer-encoding": "chunked"
},
"responseBody": "{\"error\":{\"code\":500,\"message\":\"\\n------------\\nWhile executing FilterExpression at line 57, column 53 in source:\\n...{%- for json_key in param_fields.keys() | reject(\\\"in\\\", handled_keys) %}↵ ...\\n ^\\nError: selectattr: unknown test 'in'\",\"type\":\"server_error\"}}",
"isRetryable": true,
"data": {
"error": {
"message": "\n------------\nWhile executing FilterExpression at line 57, column 53 in source:\n...{%- for json_key in param_fields.keys() | reject(\"in\", handled_keys) %}↵ ...\n ^\nError: selectattr: unknown test 'in'",
"type": "server_error",
"code": 500
}
}
}
}