Upload folder using huggingface_hub
Browse files- config.json +54 -0
- n_images_1300012.pt +3 -0
- n_images_1300012_log_feature_sparsity.pt +3 -0
- n_images_650016.pt +3 -0
- n_images_650016_log_feature_sparsity.pt +3 -0
config.json
ADDED
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"model_class_name": "HookedViT",
|
3 |
+
"model_name": "vit_base_patch16_224",
|
4 |
+
"vit_model_cfg": null,
|
5 |
+
"model_path": null,
|
6 |
+
"hook_point_layer": 11,
|
7 |
+
"layer_subtype": "hook_resid_post",
|
8 |
+
"hook_point_head_index": null,
|
9 |
+
"context_size": 197,
|
10 |
+
"use_cached_activations": false,
|
11 |
+
"use_patches_only": false,
|
12 |
+
"cached_activations_path": "activations/_datasets_imagenet1k/vit_base_patch16_224/blocks.11.hook_resid_post",
|
13 |
+
"image_size": 224,
|
14 |
+
"architecture": "standard",
|
15 |
+
"b_dec_init_method": "geometric_median",
|
16 |
+
"expansion_factor": 64,
|
17 |
+
"from_pretrained_path": null,
|
18 |
+
"seed": 42,
|
19 |
+
"d_in": 768,
|
20 |
+
"activation_fn_str": "topk",
|
21 |
+
"activation_fn_kwargs": {
|
22 |
+
"k": 64
|
23 |
+
},
|
24 |
+
"cls_token_only": false,
|
25 |
+
"max_grad_norm": 1.0,
|
26 |
+
"initialization_method": "encoder_transpose_decoder",
|
27 |
+
"normalize_activations": null,
|
28 |
+
"n_batches_in_buffer": 20,
|
29 |
+
"store_batch_size": 256,
|
30 |
+
"num_workers": 32,
|
31 |
+
"num_epochs": 1,
|
32 |
+
"verbose": false,
|
33 |
+
"l1_coefficient": 3e-05,
|
34 |
+
"lp_norm": 1,
|
35 |
+
"lr": 0.0002,
|
36 |
+
"lr_scheduler_name": "cosineannealingwarmup",
|
37 |
+
"lr_warm_up_steps": 200,
|
38 |
+
"train_batch_size": 4096,
|
39 |
+
"dataset_name": "imagenet1k",
|
40 |
+
"dataset_path": "/datasets/imagenet1k",
|
41 |
+
"dataset_train_path": "/datasets/imagenet1k/ILSVRC/Data/CLS-LOC/train",
|
42 |
+
"dataset_val_path": "/datasets/imagenet1k/ILSVRC/Data/CLS-LOC/val",
|
43 |
+
"use_ghost_grads": true,
|
44 |
+
"feature_sampling_window": 1000,
|
45 |
+
"dead_feature_window": 5000,
|
46 |
+
"dead_feature_threshold": 1e-08,
|
47 |
+
"log_to_wandb": true,
|
48 |
+
"wandb_project": "vit-sae-eval",
|
49 |
+
"wandb_entity": "hhixai",
|
50 |
+
"wandb_log_frequency": 100,
|
51 |
+
"n_validation_runs": 4,
|
52 |
+
"n_checkpoints": 2,
|
53 |
+
"checkpoint_path": "models/sae/topk/e538d3eb-vit-sae-eval"
|
54 |
+
}
|
n_images_1300012.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e95465644d6cbbfcbe2a67ead7ecf81303601e4177d9b1f99e1f9070b8991e59
|
3 |
+
size 302193752
|
n_images_1300012_log_feature_sparsity.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:885f88a790a3556f23cc87bce4305f6afe6cb712cdbe3462e1434ae9da508a7a
|
3 |
+
size 198002
|
n_images_650016.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8f9b37baa728836faa7513c997e164877a3f5b2a2773b3c300a63e50a6c35e1b
|
3 |
+
size 302193744
|
n_images_650016_log_feature_sparsity.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f67935b8dc40c40811a256e9f43e2d549f73550fbbb0d8e4b0849745e876481d
|
3 |
+
size 197997
|