How Goodfire used Ai2’s open post-training stack to trace unwanted model behavior

Goodfire used Ai2's open post-training stack to predict LLM behavioral changes, trace unwanted model behavior to individual training examples, and test targeted fixes without losing broader capability gains.

Why it matters: It shows that an open post-training stack can be used to trace and address unwanted model behavior while preserving other capability gains.

Written by AI from Ai2 Blog's published text. Read the original for full details.