Datasets:

MInAlA
/

crosscoder-multilayer-split-activations

Tags:

mechanistic-interpretability

sparse-autoencoder

License:

Dataset card Files Files and versions

crosscoder-multilayer-split-activations / v2-differentlayer /qwen3-4b-kto /singlelayer_separate /L23 /plots

2.74 MB

1 contributor

History: 1 commit

ishangarg183's picture

Copy assembled_activations to v2-differentlayer (batch 6/12)

60485e9 verified about 1 month ago

cf_distribution_per_class.png

173 kB
xet

Copy assembled_activations to v2-differentlayer (batch 6/12) about 1 month ago
cf_shift_per_class.png

132 kB
xet

Copy assembled_activations to v2-differentlayer (batch 6/12) about 1 month ago
class_distribution.png

158 kB
xet

Copy assembled_activations to v2-differentlayer (batch 6/12) about 1 month ago
loss_curves.png

327 kB
xet

Copy assembled_activations to v2-differentlayer (batch 6/12) about 1 month ago
rho_histogram.png

84.7 kB
xet

Copy assembled_activations to v2-differentlayer (batch 6/12) about 1 month ago
rho_theta_scatter.png

240 kB
xet

Copy assembled_activations to v2-differentlayer (batch 6/12) about 1 month ago
shared_geometry_histograms.png

150 kB
xet

Copy assembled_activations to v2-differentlayer (batch 6/12) about 1 month ago
shared_geometry_theta_vs_norm_ratio.png

535 kB
xet

Copy assembled_activations to v2-differentlayer (batch 6/12) about 1 month ago
superposition_analysis.png

944 kB
xet

Copy assembled_activations to v2-differentlayer (batch 6/12) about 1 month ago