Refactor query and prompt generation in EndToEndRAG class to support multiple-choice questions and improve context formatting. Simplify user input handling in app.py by passing raw user text to the retrieval function.
Refactor EndToEndRAG class to support multimodal input handling, integrating CLIP for image processing and enhancing query capabilities. Update prompt generation for improved user interaction in Persian. Enhance initialization to load indices and metadata from Hugging Face Hub.
Add chat completion method to EndToEndRAG class for improved text generation. Enhance error handling to provide clearer feedback on model compatibility and response issues.
Enhance text generation in EndToEndRAG class with multiple fallback strategies: non-streaming, streaming accumulation, and chat completion. Improve output normalization and error handling for empty responses.
Refactor text generation handling in EndToEndRAG class to normalize output from different response types. Enhance error handling for StopIteration exceptions to improve robustness.
Improve compatibility for HfHubHTTPError import across different huggingface_hub versions. Enhance error handling in EndToEndRAG class to ensure robust response detail extraction.
Enhance EndToEndRAG class initialization to support model name from environment variable. Add error handling for HfHubHTTPError with detailed response information. Update class documentation for clarity.
Refactor EndToEndRAG class to focus on a text-only pipeline using Hugging Face Inference API. Remove unused imports and methods related to image processing and multimodal retrieval. Update class documentation accordingly.
Refactor EndToEndRAG class to use InferenceClient for text generation and simplify initialization. Update requirements.txt to remove unused dependencies.
