Reinforcement Learning Challenges
Stella and Lukas discuss the challenges of reinforcement learning from QN feedback and the expensive nature of collecting data for training language models. Stella shares insights on building a data set called "the pile" to train the models due to the unavailability of OpenAI's training data.In this clip
From this podcast

Gradient Dissent - A Machine Learning Podcast
How EleutherAI Trains and Releases LLMs: Interview with Stella Biderman
Related Questions