Activations, linear+MLP refusal probes, WildGuard-judged generations & eval data to fully reproduce amir/RESULTS.md (PhillipHoward/high-temp-refusal).
Abdullah PRO
amirali1985
AI & ML interests
Mechanistic interpretability, high dimensional geometry, persona role playing.
Organizations
models 15
amirali1985/interpreting_reward_models
Updated
amirali1985/gpt-neo-125m_hh_reward
Text Generation • 0.1B • Updated • 28
amirali1985/gpt-neo-125m_utility_reward
Reinforcement Learning • Updated • 6
amirali1985/pythia-70m_sentiment_reward
Reinforcement Learning • Updated • 15
amirali1985/pythia-160m_sentiment_reward
Reinforcement Learning • Updated • 9
amirali1985/gpt-neo-125m_sentiment_reward
Reinforcement Learning • Updated • 14
amirali1985/pythia-160m_utility_reward
Reinforcement Learning • Updated • 12
amirali1985/pythia-70m_utility_reward
Reinforcement Learning • 70.4M • Updated • 18
amirali1985/gpt-j-6b-sharded-bf16_sentiment_reward
Reinforcement Learning • Updated
amirali1985/pythia-410m_utility_reward
Reinforcement Learning • Updated • 7
datasets 27
amirali1985/high-temp-refusal-mech-interp
Updated • 130
amirali1985/high-temp-refusal-probe-artifacts
Viewer • Updated • 1 • 15
amirali1985/convsersations_sadness_llama3.1-8B-it_large
Viewer • Updated • 9.78k • 10
amirali1985/convsersations_excitement_llama3.1-8B-it_large
Viewer • Updated • 8k • 36
amirali1985/convsersations_rude_llama3.1-8B-it_large
Viewer • Updated • 15.9k • 17
amirali1985/convsersations_humor_llama3.1-8B-it_large
Viewer • Updated • 9.63k • 21
amirali1985/convsersations_corrigible_more_llama3.1-8B-it_large
Viewer • Updated • 8.27k • 13
amirali1985/convsersations_power_seeking_llama3.1-8B-it_large
Viewer • Updated • 8.27k • 32
amirali1985/convsersations_wealth_seeking_llama3.1-8B-it_large
Viewer • Updated • 11.4k • 24
amirali1985/convsersations_self_awareness_general_llama3.1-8B-it_large
Viewer • Updated • 13.4k • 26