Chain of Tables Chat with Groq

Chain-of-Table Reasoning for Table Understanding

This template implements the Chain-of-Table framework for improved table reasoning with large language models. Key features:

Dynamically plans and executes a series of table operations to answer questions Supports operations like adding columns, selecting rows/columns, grouping, and sorting Stores intermediate results in transformed tables as a proxy for reasoning steps Uses Groq API to leverage large language models for planning and reasoning Includes a Gradio interface for interactive experimentation Try uploading your own tabular data and questions to see how Chain-of-Table can break down complex reasoning into interpretable steps!

Based on the ICLR 2024 paper "Chain-of-Table: Evolving Tables in the Reasoning Chain for Table Understanding". https://arxiv.org/pdf/2401.04398