Open Source LLM
Open-source large language models (LLMs) aim to democratize access to and research on powerful language AI by making models, training data, and code publicly available. Current research focuses on improving these models' performance across various tasks, including code generation, multilingual capabilities, and factual accuracy, often employing techniques like reinforcement learning, knowledge distillation, and prompt engineering to enhance capabilities and address issues like bias and hallucination. The availability of open-source LLMs fosters collaboration, reproducibility, and innovation within the scientific community while also enabling broader access to powerful language technologies for diverse practical applications.
Papers
Exploring the Readiness of Prominent Small Language Models for the Democratization of Financial Literacy
Tagore Rao Kosireddy, Jeffrey D. Wall, Evan Lucas
LLM Self-Correction with DeCRIM: Decompose, Critique, and Refine for Enhanced Following of Instructions with Multiple Constraints
Thomas Palmeira Ferraz, Kartik Mehta, Yu-Hsiang Lin, Haw-Shiuan Chang, Shereen Oraby, Sijia Liu, Vivek Subramanian, Tagyoung Chung, Mohit Bansal, Nanyun Peng