README.md

metadata

datasets:
  - cerebras/SlimPajama-627B
language:
  - en

This is the trained 1.3 billion parameter LLAMA-2 architecture model for the work Multi-Agent Collaborative Data Selection for Efficient LLM Pretraining