Abdullah's picture

2 3 1

Abdullah

amirabdullah19852020

·

amirabdullah19852020

AI & ML interests

Mechanistic interpretability, high dimensional geometry, persona role playing.

Recent Activity

updated a collection 26 days ago

Transferring Activation Features for model interventions

published a dataset 26 days ago

withmartian/binary_bbq

updated a model about 1 month ago

withmartian/trained_mediqa_model

View all activity

Organizations

upvoted a collection 4 months ago

Transferring Activation Features for model interventions

22 items • Updated 26 days ago • 1

upvoted a collection 7 months ago

Blog: Activations transfer for model interventions.

Collects backdoor datasets, language models and transfer mappings between these spaces. • 6 items • Updated May 10 • 3

upvoted a paper over 1 year ago

Beyond Training Objectives: Interpreting Reward Model Divergence in Large Language Models

Paper • 2310.08164 • Published Oct 12, 2023 • 4