Downloads · 30 days
0
jakelipner/michael-layer23-sae
michael-layer23-sae is a machine learning model from jakelipner. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Current / use this: saemichaell05e-04lr1e-03bw1e-01x16layer235k.pt Selected via hyperparameter sweep + Tyler's scoring criteria (hardl0 < 300). explvar = 0.83, hardl0 = 210, deadfrac = 0.86.
Downloads · 30 days
0
Access
Public
Updated Jul 2, 2026
Repo size
206 MB
Likes
0
Public
Click a slice to open those files.
.pt206 MB · 100%
From the Hugging Face model README
Current / use this: sae_michael_l0_5e-04_lr_1e-03_bw_1e-01_x16_layer23_5k.pt
Selected via hyperparameter sweep + Tyler's scoring criteria (hard_l0 < 300).
expl_var = 0.83, hard_l0 = 210, dead_frac = 0.86.
Superseded, do not use: sae_michael_layer23_l0_1e-03_lr_1e-03_x16_5k.pt
Early run, l0=1e-3, expl_var only 0.70. Kept for reference only.