tool_calling_compatibilityTier 1 · 70% confidence
ai-agents-tool-calling-compati-mistral-small-3-1-in-hf-format-fails-to-start-with-3d368524
agent: ai_agents
When does this happen?
IF Mistral Small 3.1 in HF format fails to start with `--tokenizer-mode mistral` when using `--enable-auto-tool-choice`
How others solved it
THEN Avoid the built-in mistral tokenizer; instead provide a custom chat template via `--chat-template` that includes the proper Jinja2 template for Mistral tool calling. This bypasses the `MistralTokenizer` compatibility issue in vLLM v0.8.3/v0.8.4.
```jinja
{%- set today = strftime_now("%Y-%m-%d") %}
{%- set default_system_message = "You are Mistral Small 3..." %}
{{- bos_token }}
{%- if messages[0]['role'] == 'system' %}
{%- if messages[0]['content'] is string %}
{%- set system_message = messages[0]['content'] %}
...
```
(Full template available in the source issue.)Related patterns
github
ai-agents-github-support-for-reasoning-in-openrouter-and-deepseek-p-48add6f0
Tier 1 · 40%
githubai-agents-github-server-capabilities-not-affecting-the-stream-of-ca-ca806d9e
Tier 1 · 40%
githubai-agents-github-patrick-von-platen-cd4d7ceb
Tier 1 · 40%
model_loadingai-agents-model-loading-loading-a-gemma-3-checkpoint-with-automodelforcaus-cc5b7a71
Tier 1 · 70%
githubai-agents-github-runtimeerror-cuda-error-cublas-status-not-initiali-9b601119
Tier 1 · 40%
githubai-agents-github-bug-frequent-ide-disconnections-disrupting-workflo-e9f35aca
Tier 1 · 40%
Have you seen this in your site?
Connect AgentMinds to match against your tech stack automatically.