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": 128, "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_128_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": 128, "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 sha256:74d5bd801bdfcd989d6e1284b7c4ace2ec862b23d38d644196e8dfcd6a08e448
|
3 |
+
size 268699984
|
gpt_neox.layers.1/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 128, "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 sha256:145d04773548ece5cf01c4a2f0ba020357d3cad16dd28274d69c8a26aaf23a0e
|
3 |
+
size 268699984
|
gpt_neox.layers.2/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 128, "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:f2a6b9fdef1e8238bb2c604a5e4728205c85780ca31e7e10f94181ea2ca86b3a
|
3 |
+
size 268699984
|
gpt_neox.layers.3/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 128, "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 sha256:69faa647c443e9b6ee26ef33752d46b935783a7acf644d521cb2fb81376c4fa7
|
3 |
+
size 268699984
|
gpt_neox.layers.4/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 128, "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:c965f08167faff0f2fce31d9d2da4f62d3fc1f355c4f9b9017469d9756e274c4
|
3 |
+
size 268699984
|
gpt_neox.layers.5/cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"expansion_factor": 128, "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 sha256:439b0b75396624cb3e253be246d3f6685006e1e12a75a574936d0f484b83eb74
|
3 |
+
size 268699984
|