YX-S-Z
commited on
Commit
·
e06a12f
1
Parent(s):
7d20fc1
sae upload
Browse files- 0.pt +3 -0
- 0_cfg.json +1 -0
- 1.pt +3 -0
- 1_cfg.json +1 -0
- 2.pt +3 -0
- 2_cfg.json +1 -0
- 3.pt +3 -0
- 3_cfg.json +1 -0
- 4.pt +3 -0
- 4_cfg.json +1 -0
- 5.pt +3 -0
- 5_cfg.json +1 -0
- 6.pt +3 -0
- 6_cfg.json +1 -0
0.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:eee3666e7055ab3ce031a7814a1e28dad17e08f7c6f78c80ca00a3236361f19a
|
3 |
+
size 67177787
|
0_cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0001, "num_tokens": 2000000000, "l1_coeff": 0.0003, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 128, "enc_dtype": "fp32", "remove_rare_dir": false, "model_name": "crate-1l", "site": "post", "layer": 0, "device": "cuda:0", "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 12288, "act_name": "blocks.0.mlp.hook_post", "act_size": 512, "dict_size": 16384, "name": "crate-1l_0_16384_post"}
|
1.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:141ef47684a86cd885e0bd7dc0957d7ea1fe6b5f8eb01b239983e27966f991a1
|
3 |
+
size 67177787
|
1_cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0001, "num_tokens": 2000000000, "l1_coeff": 0.0003, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 128, "enc_dtype": "fp32", "remove_rare_dir": false, "model_name": "crate-1l", "site": "post", "layer": 0, "device": "cuda:0", "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 12288, "act_name": "blocks.0.mlp.hook_post", "act_size": 512, "dict_size": 16384, "name": "crate-1l_0_16384_post"}
|
2.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:28f8543951fb8447efa7cddbd16ceb6b6fb201bd6d3c0d35ae2f3d2cc0758acc
|
3 |
+
size 67177787
|
2_cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0001, "num_tokens": 2000000000, "l1_coeff": 0.0003, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 128, "enc_dtype": "fp32", "remove_rare_dir": false, "model_name": "crate-1l", "site": "post", "layer": 0, "device": "cuda:0", "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 12288, "act_name": "blocks.0.mlp.hook_post", "act_size": 512, "dict_size": 16384, "name": "crate-1l_0_16384_post"}
|
3.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b6c54a79f70c4418c377085cfb4a701b1bb00142a8253a76a207614dacaa4fe4
|
3 |
+
size 67177787
|
3_cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0001, "num_tokens": 2000000000, "l1_coeff": 0.0003, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 128, "enc_dtype": "fp32", "remove_rare_dir": false, "model_name": "crate-1l", "site": "post", "layer": 0, "device": "cuda:0", "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 12288, "act_name": "blocks.0.mlp.hook_post", "act_size": 512, "dict_size": 16384, "name": "crate-1l_0_16384_post"}
|
4.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e3924e93516a41a8e13fd72307eb43680c879bad9a2a557cb8c011fbfb4c4066
|
3 |
+
size 67177787
|
4_cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0001, "num_tokens": 2000000000, "l1_coeff": 0.0003, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 128, "enc_dtype": "fp32", "remove_rare_dir": false, "model_name": "crate-1l", "site": "post", "layer": 0, "device": "cuda:0", "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 12288, "act_name": "blocks.0.mlp.hook_post", "act_size": 512, "dict_size": 16384, "name": "crate-1l_0_16384_post"}
|
5.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:65d9f32085239773f0ae4bea463156ccd16d2758b4c3d2a1ac5e9dcbef77b204
|
3 |
+
size 67177787
|
5_cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0001, "num_tokens": 2000000000, "l1_coeff": 0.0003, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 128, "enc_dtype": "fp32", "remove_rare_dir": false, "model_name": "crate-1l", "site": "post", "layer": 0, "device": "cuda:0", "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 12288, "act_name": "blocks.0.mlp.hook_post", "act_size": 512, "dict_size": 16384, "name": "crate-1l_0_16384_post"}
|
6.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e293535decc5ec6f908a3e2ac504304a6a3a29111ab375c4ea4560ffb10ccb6c
|
3 |
+
size 67177787
|
6_cfg.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"seed": 49, "batch_size": 4096, "buffer_mult": 384, "lr": 0.0001, "num_tokens": 2000000000, "l1_coeff": 0.0003, "beta1": 0.9, "beta2": 0.99, "dict_mult": 32, "seq_len": 128, "enc_dtype": "fp32", "remove_rare_dir": false, "model_name": "crate-1l", "site": "post", "layer": 0, "device": "cuda:0", "model_batch_size": 512, "buffer_size": 1572864, "buffer_batches": 12288, "act_name": "blocks.0.mlp.hook_post", "act_size": 512, "dict_size": 16384, "name": "crate-1l_0_16384_post"}
|