Mukul Rayana
Day 1: data pipeline, session tracker, query router, adversarial probes, Colab training notebooks
bc3ba9e | ο»Ώ# ββ PyTorch (must install separately with CUDA index β see below) ββ | |
| # torch==2.1.2+cu121 <- installed via separate command, NOT from this file | |
| # ββ HuggingFace stack ββ | |
| transformers==4.38.2 | |
| peft==0.10.0 | |
| datasets==2.18.0 | |
| accelerate==0.27.2 | |
| evaluate==0.4.1 | |
| huggingface_hub>=0.20.0 | |
| # ββ Retrieval ββ | |
| faiss-cpu==1.7.4 | |
| sentence-transformers==2.5.1 | |
| # ββ Generator ββ | |
| # llama-cpp-python installed separately with CUDA wheel β see below | |
| # ββ Explainability ββ | |
| captum==0.7.0 | |
| # ββ Evaluation ββ | |
| ragas==0.1.21 | |
| bert-score==0.3.13 | |
| # ββ Demo ββ | |
| gradio==4.21.0 | |
| # ββ Data / Utilities ββ | |
| pandas==2.2.1 | |
| numpy==1.26.4 | |
| tqdm==4.66.2 | |
| kaggle==1.6.3 | |
| scipy==1.12.0 | |
| scikit-learn==1.4.1 | |