POPVOX Foundation Cosponsors Large Language Model x Law: Hackathon
On June 25-27, POPVOX Foundation will cosponsor the “Large Language Model x Law Hackathon” in New York with the open source research community working to build Large Legal Language Models (LLLMs) that better understand (and obey) law and regulations.
The hackathon invites participants to envision and write code that leverages open source LLLMs with a focus on the following areas:
Applications: contract generation, regulatory Q&A, etc.
LLM-powered LLCs: LLMs powering the operation of companies
Data Preparation: data scraping, processing and curation for pretraining data for the next gen of larger LLLM training runs
Synthetic Data Generation: synthetic data gen for supervised fine-tuning, reinforcement learning, and evaluation of LLLMs
Model Capability Evaluations: tools for comprehensive evals of LLLMs, expanding / leveraging test sets such as LegalBench
LLM-as-Agent Simulations: tools for simulating multi-agent LLM behaviors and testing their resistance to taking illegal actions
The event is organized by the open source research community that emerged from the “Legal Informatics Approach to AI Alignment” concept developed by John Nay as a Stanford CodeX fellow. Organizers hope that the LLLMs and concepts emerging from the hackathon can serve as Foundation Model starting points for governments, AI researchers, companies, and citizens.
Cash prizes for best projects are sponsored by Modal.
Additional cosponsors include: Thrive Capital, Modal, Amplify, LexisNexis, LangChain, Nomic, Meridian Street, Hearth
The primary venue is South Park Commons, with a dinner sponsored by Amplify Partners. Remote participation options will be available.
