This repository holds the LLM-based project using HuggingFace and Google open-source language model.
We used the following frameworks:
Streamlit:- Used to build the app.
🦜🔗 LangChain:- Used to build LLM Workflow.
🤗 HuggingFaceHub:- Needed to use the open source model hosted in Huggingface by Google. This model is Flan-T5, which is an alternative to GPT-3. Flan-T5 is an encoder-decoder transformer model that reframes all NLP tasks into a text-to-text format. With appropriate prompting, it can perform zero-shot NLP tasks such as text summarization, common sense reasoning, natural language inference, question answering, sentence and sentiment classification, translation, and pronoun resolution. Details of this model are given on the HuggingFace page
This model is open-source and used only for showcasing purposes, however, it is mentioned in the Risks associated with this model that it is not assessed deeply on HuggingFace's "Bias, Risks and Limitation" section.
Wikipedia:- Used to connect GPT to Wikipedia.
Link of this app is provided here:- Link