# ============================================================================= # RAG Chatbot with Vector Database - Requirements # ============================================================================= # Production-ready dependencies for medical document analysis and Q&A # Core ML/AI Framework torch>=2.0.0 # PyTorch for model inference transformers>=4.30.0 # Hugging Face transformers huggingface_hub>=0.20.0 # Hugging Face Hub API (for Inference API) accelerate>=0.20.0 # Model loading optimization safetensors>=0.3.0 # Safe model loading # Vector Database & Embeddings chromadb>=0.4.0 # Vector database for fast retrieval sentence-transformers>=2.2.0 # Semantic embeddings (all-MiniLM-L6-v2) # Data Processing pandas>=1.3.0 # Data manipulation and CSV handling numpy>=1.21.0 # Numerical computing scikit-learn>=1.0.0 # ML utilities and TF-IDF # Text Analysis & Readability textstat>=0.7.0 # Flesch-Kincaid Grade Level calculation nltk>=3.8.0 # Natural language processing utilities # Document Processing (Core) pypdf>=3.0.0 # PDF document parsing python-docx>=0.8.11 # DOCX document parsing # Optional Document Processing rank-bm25>=0.2.2 # BM25 retrieval algorithm (alternative to TF-IDF) # Utilities & Progress tqdm>=4.65.0 # Progress bars pathlib2>=2.3.0 # Enhanced path handling (if needed) # Web Interface gradio==4.44.1 # Gradio web interface for chatbot (updated for Spaces compatibility) # Development & Testing (Optional) pytest>=7.0.0 # Testing framework black>=22.0.0 # Code formatting flake8>=4.0.0 # Code linting # Performance Monitoring (Optional) psutil>=5.8.0 # System resource monitoring memory-profiler>=0.60.0 # Memory usage profiling # ============================================================================= # Installation Notes: # ============================================================================= # 1. Install with: pip install -r requirements.txt # 2. For Apple Silicon: PyTorch will automatically use MPS acceleration # 3. Optional packages can be installed separately if needed # 4. Model files (~6GB) will be downloaded on first run # =============================================================================