3 ms·
Temporal Augmented Retrieval (Tar) – Dynamic RAG
- Terretta 3y agoAuthor's note: From a corpus of text, how can you detect emerging topics and their evolution through time? Introducing Temporal Augmented Retrieval (TAR). (built in the context of buildspace n&w s4) TAR is an open-source advanced RAG approach that aims to factor in the dynamic and temporal aspects of textual data when performing retrieval. It allows us to understand the evolution of discussed topics over time. The idea behind this project is to open the debate regarding the current limitations of RAG methods. This first approach has been built without using RAG frameworks (like Jerry Liu’s langchain) and focuses on financial tweets. Relevant links: Diagram: https://preview.redd.it/lj7wkhk70f9c1.png https://preview.redd.it/lj7wkhk70f9c1.png Medium: https://medium.com/@adam-rida/temporal-augmented-retrieval-tar-dynamic-rag-ad737506dfcc https://medium.com/@adam-rida/temporal-augmented-retrieval-t... Github: https://github.com/adrida/Temporal_RAG https://github.com/adrida/Temporal_RAG Hugging Face Benchmark: https://huggingface.co/spaces/Adr740/Temporal-RAG-Benchmark https://huggingface.co/spaces/Adr740/Temporal-RAG-Benchmark Author site: https://adrida.github.io https://adrida.github.io