Datasets:

MInAlA
/

crosscoder-multilayer-split-activations

Tags:

mechanistic-interpretability

sparse-autoencoder

License:

Dataset card Files Files and versions

crosscoder-multilayer-split-activations / v2-differentlayer /llama32-3b-grpo /singlelayer_separate /base_L10__aligned_L12 /plots

1.66 MB

1 contributor

History: 1 commit

ishangarg183's picture

Copy assembled_activations to v2-differentlayer (batch 1/12)

a167bc7 verified 13 days ago

cf_distribution_per_class.png

169 kB
xet

Copy assembled_activations to v2-differentlayer (batch 1/12) 13 days ago
cf_shift_per_class.png

101 kB
xet

Copy assembled_activations to v2-differentlayer (batch 1/12) 13 days ago
class_distribution.png

158 kB
xet

Copy assembled_activations to v2-differentlayer (batch 1/12) 13 days ago
loss_curves.png

336 kB
xet

Copy assembled_activations to v2-differentlayer (batch 1/12) 13 days ago
rho_histogram.png

83.4 kB
xet

Copy assembled_activations to v2-differentlayer (batch 1/12) 13 days ago
rho_theta_scatter.png

246 kB
xet

Copy assembled_activations to v2-differentlayer (batch 1/12) 13 days ago
shared_geometry_histograms.png

134 kB
xet

Copy assembled_activations to v2-differentlayer (batch 1/12) 13 days ago
shared_geometry_theta_vs_norm_ratio.png

263 kB
xet

Copy assembled_activations to v2-differentlayer (batch 1/12) 13 days ago
superposition_analysis.png

171 kB
xet

Copy assembled_activations to v2-differentlayer (batch 1/12) 13 days ago