RAG / data_cutter.py

Commit History

feat: Switch to French-specific models and refine prompt engineering for improved French RAG.
fc8e15c

Kakarot21 commited on

feat: Update app to use Flan-T5-base model with error handling, fix Chroma embedding parameter, and add pre-deployment checklist.
38854c4

Kakarot21 commited on

fix: handle None values in extracted table cells when converting to markdown and improve empty table detection.
1d88fd5

Kakarot21 commited on

refactor: update Document import from `langchain.schema` to `langchain_core.documents` and adjust dependencies.
3e09ea4

Kakarot21 commited on

refactor: Migrate HuggingFaceEmbeddings import to langchain_huggingface package.
f516652

Kakarot21 commited on

refactor: update HuggingFaceEmbeddings import path from `langchain_huggingface` to `langchain.embeddings`.
0840d5f

Kakarot21 commited on

feat: Update LLM to MPT-7B, switch embedding model to all-MiniLM-L6-v2, refactor PDF parsing with pdfplumber, and simplify LLM prompting.
6e52b29

Kakarot21 commited on

Update data_cutter.py
dfe7d61
verified

Sof850 commited on

feat: Update embedding model to Jina, add RAG token to .env, and introduce a Gradio client script.
06c53c9

Kakarot21 commited on

feat: Implement a basic RAG chatbot application using local ChromaDB, HuggingFace embeddings and model, and a Gradio interface.
0472254

Kakarot21 commited on