一方的赛博真君 8c309c00e1 fix(ai_model): support reasoning field for Ollama/LMStudio thinking models ...

Add compatibility for the `reasoning` field in addition to `reasoning_content`.
Ollama/LMStudio follows GPT-OSS official recommendation to use `reasoning`
field for thinking content, while DeepSeek uses `reasoning_content`.

Priority: reasoning_content > reasoning (fallback)

2026-01-22 14:26:57 +08:00

..

openai

fix(ai_model): support reasoning field for Ollama/LMStudio thinking models

2026-01-22 14:26:57 +08:00

__init__.py

feat: generate sql with history

2025-05-30 17:41:45 +08:00

embedding.py

feat: disable TOKENIZERS_PARALLELISM warning

2025-08-27 16:21:37 +08:00

llm.py

feat: generate sql with history

2025-05-30 17:41:45 +08:00

model_factory.py

perf: Encryption of sensitive data in ai_model table #9

2025-08-27 17:44:34 +08:00