On-policy Qwen3-8B training mixture reproducing 'Building Better Activation Oracles' (arXiv:2606.02609). Start with the recipe card.
Francesco Ortu
francescortu
AI & ML interests
NLP, Interpretabilty
Recent Activity
updated a collection 3 days ago
CoT Activation Oracle — On-Policy Qwen3-8B Recipe updated a collection 3 days ago
CoT Activation Oracle — On-Policy Qwen3-8B Recipe updated a collection 3 days ago
CoT Activation Oracle — On-Policy Qwen3-8B Recipe