A modular, explainable Retrieval-Augmented Generation (RAG) pipeline for automated fact-checking on the FEVER dataset. Includes multi-model evaluation (Mistral, Qwen, GPT), PEFT & DPO fine-tuning, and token-level explainability with Captum, SHAP, and LIME
nlp lora fact-checking explainable-ai rag dpo llms chatgpt-api mistral-7b qwen2 peft-fine-tuning-llm fever-dataset
-
Updated
Oct 13, 2025 - Jupyter Notebook