Figure gallery
IN_to_logits
newline_feature_inputs
QK_from_url_start
chinese_feature_to_logits
french_feature_to_logits
summary
waveforms
virtual-weights
in-token-leaderboard
in-to-logits-filter
position-scatter
feat-1254-token-leaderboard
direct-attention-gate
feature-circuit-diagram
effect-help-scatter
threshold-eval
helpfulness-table
acetylcholine-pairplot
single-weight-helpfulness
fisher-v-ea
vw-help-scatter
threshold-eval-family
rope
feat-157-examples
feat-3013-examples
feat-1254-examples
weight-superposition-cartoon
vw-model-schematic
mean-helpfulness-histograms
mean-helpfulness-cdfs
helpfulness-mass-distribution
fisher-effectiveness-cdfs
fisher-mass-distribution
full-helpfulness-table