mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-03 02:22:24 +00:00
Rebuild the InvokeModel count body in place so keys stay in request order and only messages is rewritten. Gate the thinking display coercion on the dict check before calling the normalizer. Pin the already-empty content list next to an offender, malformed message entries next to an offender, and the count body key order in the mapped tests |
||
|---|---|---|
| .. | ||
| chat | ||
| count_tokens | ||
| files | ||
| image | ||
| image_edit | ||
| invoke_agent | ||
| passthrough | ||
| realtime | ||
| rerank | ||
| vector_stores | ||
| __init__.py | ||
| test_common_utils.py | ||