Attention weight distributions across transformer architectures.
-
angerami/pythia-70m-deduped_weight_evolution
Viewer • Updated • 22.2k • 9 -
angerami/pythia-160m-deduped_weight_evolution
Viewer • Updated • 66.5k • 6 -
angerami/pythia-410m-deduped_weight_evolution
Viewer • Updated • 177k • 8 -
angerami/pythia-1b-deduped_weight_evolution
Viewer • Updated • 59.1k • 7