Stella Biderman: How EleutherAI Trains and Releases LLMs

On this episode of Gradient Dissent, we’re joined by Stella Biderman, Lead Scientist at Booz Allen Hamilton and Executive Director at ElutherAI. Stella and Lukas discuss EleutherAI’s origin and future, LLM similarities and differences, choosing models, reinforcement learning, pre-training/fine-tuning, GPU selection, differences from other LLM companies, interpretability, memorization importance, and public access.



Source link