Enjoyed joining @apoorv03 for MS&E435 to talk about where frontier model training is headed!

The most interesting work right now sits at the boundary between evals, RL environments, post-training, and systems that learn from real production feedback. Powerful methodologies to train highly performant models are here. A lot of model progress comes down to defining the hill you want to climb.

We are very focused on pushing the frontier on post-training and continual learning at Applied Compute. If that is interesting to you, come join us!