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-grpo /singlelayer_separate /base_L25__aligned_L21 /plots

1 contributor

History: 1 commit

ishangarg183's picture

Copy assembled_activations to v2-differentlayer (batch 5/12)

55276cd verified 16 days ago

cf_distribution_per_class.png

215 kB
xet

Copy assembled_activations to v2-differentlayer (batch 5/12) 16 days ago
cf_shift_per_class.png

96.8 kB
xet

Copy assembled_activations to v2-differentlayer (batch 5/12) 16 days ago
class_distribution.png

160 kB
xet

Copy assembled_activations to v2-differentlayer (batch 5/12) 16 days ago
loss_curves.png

305 kB
xet

Copy assembled_activations to v2-differentlayer (batch 5/12) 16 days ago
rho_histogram.png

82.6 kB
xet

Copy assembled_activations to v2-differentlayer (batch 5/12) 16 days ago
rho_theta_scatter.png

419 kB
xet

Copy assembled_activations to v2-differentlayer (batch 5/12) 16 days ago
shared_geometry_histograms.png

125 kB
xet

Copy assembled_activations to v2-differentlayer (batch 5/12) 16 days ago
shared_geometry_theta_vs_norm_ratio.png

540 kB
xet

Copy assembled_activations to v2-differentlayer (batch 5/12) 16 days ago
superposition_analysis.png

481 kB
xet

Copy assembled_activations to v2-differentlayer (batch 5/12) 16 days ago