How knowrohit/alfred_local_gpt works
Project type and purpose
• This is a standalone, full-stack Python CLI (with optional HTTP API hooks) that allows users to:
• Load and vectorize documents (load_docs.py)
• Store/retrieve embeddings in a local Chroma vector database (db_vector/)
• Serve queries to a local LLM binary (ggml-gpt4all-j-v1.3-groovy.bin or alternative) using SentenceTransformers for context
• Return generated answers to the user
File structure
• Top-level files:
• alfredGPT.py — entrypoint for CLI/possibly API server
• load_docs.py — ingestion & indexing pipeline
• constants.py — shared configuration
• requirements.txt / .env (implied) — dependency and environment configuration
• Virtual environment folder (alfred_gpt/) contains all installed packages and CLI scripts (langchain, uvicorn, huggingface-cli, etc.)
• db_vector/ — local Chroma database storage: embeddings, parquet collections, index binaries & metadata
• .github & CI configs — project governance, not part of runtime
README for additional insights
• Instructs user to download a local LLM binary and choose a SentenceTransformers model via .env
• No external cloud services are required—everything runs locally