rufimelo commited on
Commit
584fe75
·
verified ·
1 Parent(s): 0cec6d6

Upload folder using huggingface_hub

Browse files
README.md CHANGED
@@ -8,7 +8,7 @@ tags:
8
 
9
  # Sparse Autoencoders for Unknown
10
 
11
- This repository contains 1 Sparse Autoencoder(s) (SAE) trained using [SAELens](https://github.com/jbloomAus/SAELens).
12
 
13
  ## Model Details
14
 
@@ -25,6 +25,8 @@ This repository contains 1 Sparse Autoencoder(s) (SAE) trained using [SAELens](h
25
  | Hook Point |
26
  |------------|
27
  | `blocks.0.hook_resid_post` |
 
 
28
 
29
  ## Usage
30
 
@@ -53,6 +55,12 @@ features = sae.encode(activations)
53
  - `blocks.0.hook_resid_post/cfg.json` - SAE configuration
54
  - `blocks.0.hook_resid_post/sae_weights.safetensors` - Model weights
55
  - `blocks.0.hook_resid_post/sparsity.safetensors` - Feature sparsity statistics
 
 
 
 
 
 
56
 
57
  ## Training
58
 
 
8
 
9
  # Sparse Autoencoders for Unknown
10
 
11
+ This repository contains 3 Sparse Autoencoder(s) (SAE) trained using [SAELens](https://github.com/jbloomAus/SAELens).
12
 
13
  ## Model Details
14
 
 
25
  | Hook Point |
26
  |------------|
27
  | `blocks.0.hook_resid_post` |
28
+ | `blocks.14.hook_resid_post` |
29
+ | `blocks.27.hook_resid_post` |
30
 
31
  ## Usage
32
 
 
55
  - `blocks.0.hook_resid_post/cfg.json` - SAE configuration
56
  - `blocks.0.hook_resid_post/sae_weights.safetensors` - Model weights
57
  - `blocks.0.hook_resid_post/sparsity.safetensors` - Feature sparsity statistics
58
+ - `blocks.14.hook_resid_post/cfg.json` - SAE configuration
59
+ - `blocks.14.hook_resid_post/sae_weights.safetensors` - Model weights
60
+ - `blocks.14.hook_resid_post/sparsity.safetensors` - Feature sparsity statistics
61
+ - `blocks.27.hook_resid_post/cfg.json` - SAE configuration
62
+ - `blocks.27.hook_resid_post/sae_weights.safetensors` - Model weights
63
+ - `blocks.27.hook_resid_post/sparsity.safetensors` - Feature sparsity statistics
64
 
65
  ## Training
66
 
blocks.14.hook_resid_post/cfg.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"d_in": 3584, "d_sae": 16384, "dtype": "float32", "device": "cuda", "apply_b_dec_to_input": true, "normalize_activations": "none", "reshape_activations": "none", "metadata": {"sae_lens_version": "6.26.2", "sae_lens_training_version": "6.26.2", "hook_name": "blocks.14.hook_resid_post"}, "decoder_init_norm": 0.1, "k": 64, "use_sparse_activations": false, "aux_loss_coefficient": 1.0, "rescale_acts_by_decoder_norm": true, "contrastive_weight": 0.1, "contrastive_temperature": 0.07, "contrastive_mode": "infonce", "triplet_margin": 1.0, "use_feature_contrastive": true, "architecture": "topk"}
blocks.14.hook_resid_post/sae_weights.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3f6ad2047868d8eed2ac9f74b7bcca3ab1c308ac41d043ddc7d7672f3873a62c
3
+ size 469842240
blocks.14.hook_resid_post/sparsity.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ddf17024c127d1c35907a165e8dae04c8efa34cd1cc76a592708e397e97a2274
3
+ size 65616
blocks.27.hook_resid_post/cfg.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"d_in": 3584, "d_sae": 16384, "dtype": "float32", "device": "cuda", "apply_b_dec_to_input": true, "normalize_activations": "none", "reshape_activations": "none", "metadata": {"sae_lens_version": "6.26.2", "sae_lens_training_version": "6.26.2", "hook_name": "blocks.27.hook_resid_post"}, "decoder_init_norm": 0.1, "k": 64, "use_sparse_activations": false, "aux_loss_coefficient": 1.0, "rescale_acts_by_decoder_norm": true, "contrastive_weight": 0.1, "contrastive_temperature": 0.07, "contrastive_mode": "infonce", "triplet_margin": 1.0, "use_feature_contrastive": true, "architecture": "topk"}
blocks.27.hook_resid_post/sae_weights.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38e2cb08f96d39f7c33c50399fd6bb52a73259a42a5d605b168f9e20a64c79c3
3
+ size 469842240
blocks.27.hook_resid_post/sparsity.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6709e97cc4f979df0dc76ca55f08632f2a29dab1e372fcbd6d053622be7277c8
3
+ size 65616