use preciz SDK from pip package instead of local path
syntaxhacker
af4937c
- Import _call_llm, _parse_tool_calls, execute_tool, _strip_tool_tags
from pip-installed preciz (summarizer.llm_client)
- Remove sys.path.insert hack for local redisum
- Register rag_qa handler in TOOL_HANDLERS at module level
- Add scripts/utils_shim.py + Dockerfile step to work around
missing utils module in preciz pip package
- Update requirements.txt with preciz git dependency
- Fix system prompt to match parser's expected XML format
(name wrapped in <longcat_arg_key>)
fix XML tool call parser for malformed key-value pairs
syntaxhacker
87285c1
handle XML-style tool calls for models without native function calling
syntaxhacker
8630c76
switch to openrouter/owl-alpha model, move model name to env var
syntaxhacker
431af70
fix: use onnxruntime directly without torch dependency
syntaxhacker
2925df7
feat: switch to ONNX runtime for embeddings
syntaxhacker
3f3aedf
refactor: remove Google AI, switch to OpenRouter, improve service runner
syntaxhacker
59b7a59
add promtps to docker path
rohit
f9322c0
add fallback
rohit
01ec3c1
seperate prompts
rohit
b924bc1
remove logs
rohit
8946f02
improved tool calling , prompt enhance
rohit
b5e33d3
add tests
rohit
3e7266f
Integrate OpenRouter GLM model with RAG tool calling
rohit
0a1d4cf
- Replace Google Gemini with OpenRouter GLM-4.5-air model
- Add intelligent tool calling for RAG pipeline access
- Implement new /chat endpoint with multi-turn conversation support
- Remove Google API key dependency
- Add comprehensive .gitignore for Python projects
- Maintain backward compatibility with existing /answer endpoint
Features:
- GLM model automatically decides when to use RAG vs general chat
- Tool calling architecture for extensible functionality
- Background dataset loading for faster startup
- Support for developer-portfolio and other datasets
Add /questions endpoint
rohit
6de22e1
Remove bolding from prompt examples
rohit
239a856
Add few-shot examples to prompt
rohit
89b80a1
Enhance prompt for better readability
rohit
8bf73b1
Refine prompt for developer portfolio
rohit
2e540d6
Deploy RAG Pipeline with FastAPI and ML capabilities
