Upload 13 files
Browse files- cfg.json +1 -0
- gpt_neox.layers.0/cfg.json +1 -0
- gpt_neox.layers.0/sae.safetensors +3 -0
- gpt_neox.layers.1/cfg.json +1 -0
- gpt_neox.layers.1/sae.safetensors +3 -0
- gpt_neox.layers.2/cfg.json +1 -0
- gpt_neox.layers.2/sae.safetensors +3 -0
- gpt_neox.layers.3/cfg.json +1 -0
- gpt_neox.layers.3/sae.safetensors +3 -0
- gpt_neox.layers.4/cfg.json +1 -0
- gpt_neox.layers.4/sae.safetensors +3 -0
- gpt_neox.layers.5/cfg.json +1 -0
- gpt_neox.layers.5/sae.safetensors +3 -0
cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"sae": {"expansion_factor": 64, "normalize_decoder": true, "num_latents": 0, "k": 32, "multi_topk": false}, "batch_size": 4, "grad_acc_steps": 8, "micro_acc_steps": 2, "lr": null, "lr_warmup_steps": 1000, "auxk_alpha": 0.0, "dead_feature_threshold": 10000000, "hookpoints": ["gpt_neox.layers.0", "gpt_neox.layers.1", "gpt_neox.layers.2", "gpt_neox.layers.3", "gpt_neox.layers.4", "gpt_neox.layers.5"], "layers": [0, 1, 2, 3, 4, 5], "layer_stride": 1, "distribute_modules": false, "save_every": 1000, "max_tokens": 1000000000, "log_to_wandb": true, "run_name": "/home/random_sae/experiments/saved_models/pythia-70m_64_k32/redpajama-data-1t-sample_plain_text_100M_trained", "wandb_log_frequency": 1}
|
gpt_neox.layers.0/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 64, "normalize_decoder": true, "num_latents": 0, "k": 32, "multi_topk": false, "d_in": 512}
|
gpt_neox.layers.0/sae.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:44f5c918185d8baa49dede4a8467fb05b5412d11adbdc9f3a6b83fe624e0565c
|
3 |
+
size 134351176
|
gpt_neox.layers.1/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 64, "normalize_decoder": true, "num_latents": 0, "k": 32, "multi_topk": false, "d_in": 512}
|
gpt_neox.layers.1/sae.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bdff8d017d371e29ba6d7a32f1d76edde1a5692968b1362220e41e707f143353
|
3 |
+
size 134351176
|
gpt_neox.layers.2/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 64, "normalize_decoder": true, "num_latents": 0, "k": 32, "multi_topk": false, "d_in": 512}
|
gpt_neox.layers.2/sae.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c0656a3d724e0761aafe738c323ea630553402e95ff729bdb70ae7a649a43569
|
3 |
+
size 134351176
|
gpt_neox.layers.3/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 64, "normalize_decoder": true, "num_latents": 0, "k": 32, "multi_topk": false, "d_in": 512}
|
gpt_neox.layers.3/sae.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bce2c10c6222e58fa50f2b5a9066990a73826f522d1e157d33b8601667e7b2a6
|
3 |
+
size 134351176
|
gpt_neox.layers.4/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 64, "normalize_decoder": true, "num_latents": 0, "k": 32, "multi_topk": false, "d_in": 512}
|
gpt_neox.layers.4/sae.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d7f6053ec4ac27760e0fc5338186080eb13e8d42815732590dfab3eaca9f9d71
|
3 |
+
size 134351176
|
gpt_neox.layers.5/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 64, "normalize_decoder": true, "num_latents": 0, "k": 32, "multi_topk": false, "d_in": 512}
|
gpt_neox.layers.5/sae.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e780f9de36e8b5ab4e9d9a5052e0c65132b4d44b711fee60bb738a18e22d400e
|
3 |
+
size 134351176
|