Hugging Face's logo

JackBAI
/
crate-1l-l0-sae-1x

Model card Files Community
YX-S-Z commited on
Commit
6fbe99d
·
1 Parent(s): 908783a

all ckpts

Browse files
Files changed (30) hide show
  1. 0.pt +3 -0
  2. 0_cfg.json +1 -0
  3. 1.pt +3 -0
  4. 10.pt +3 -0
  5. 10_cfg.json +1 -0
  6. 11.pt +3 -0
  7. 11_cfg.json +1 -0
  8. 12.pt +3 -0
  9. 12_cfg.json +1 -0
  10. 13.pt +3 -0
  11. 13_cfg.json +1 -0
  12. 14.pt +3 -0
  13. 14_cfg.json +1 -0
  14. 1_cfg.json +1 -0
  15. 2.pt +3 -0
  16. 2_cfg.json +1 -0
  17. 3.pt +3 -0
  18. 3_cfg.json +1 -0
  19. 4.pt +3 -0
  20. 4_cfg.json +1 -0
  21. 5.pt +3 -0
  22. 5_cfg.json +1 -0
  23. 6.pt +3 -0
  24. 6_cfg.json +1 -0
  25. 7.pt +3 -0
  26. 7_cfg.json +1 -0
  27. 8.pt +3 -0
  28. 8_cfg.json +1 -0
  29. 9.pt +3 -0
  30. 9_cfg.json +1 -0
0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:009b603ac778702a5e18a626f55e4955baa55b19307218bd8bd2c6de67e3a78e
3
+ size 2102587
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": 1, "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": 512, "name": "crate-1l_0_512_post"}
1.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:960cca000885e927d6fb485efe3dc23bffcf2373b2dd7ede0b80dba319c24d5a
3
+ size 2102587
10.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e957b5c78dfa37b207759d019f59944bf7055f09d62c2cc81b5c31fc111b2171
3
+ size 2102593
10_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": 1, "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": 512, "name": "crate-1l_0_512_post"}
11.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1008b5232ffad32d8ef2b9a6a1032b24cf6d5dafe8617c7868cc7ee8f111baed
3
+ size 2102593
11_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": 1, "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": 512, "name": "crate-1l_0_512_post"}
12.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f07bd22fbce3d8cc480dd5983519d1e49a4762222b6968456e34a39c6fadf0fc
3
+ size 2102593
12_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": 1, "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": 512, "name": "crate-1l_0_512_post"}
13.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f49d676de5ffc222daf6f973da55ef0e87dbdab0a9c67c895133aca9f923ed57
3
+ size 2102593
13_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": 1, "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": 512, "name": "crate-1l_0_512_post"}
14.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2bfebb835f58cb991d02f11683bbb5e6c741d7c767b0a3f7dd5aea1a28543aef
3
+ size 2102593
14_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": 1, "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": 512, "name": "crate-1l_0_512_post"}
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": 1, "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": 512, "name": "crate-1l_0_512_post"}
2.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:020ee826b2f906a3da242aff29346bc0560b6cf7c48220754d53b91b54fe62b0
3
+ size 2102587
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": 1, "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": 512, "name": "crate-1l_0_512_post"}
3.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05d92e7ec9e6e0b3e95f4b90b57d5ef0d5562af575934ab242a2d1514ae45cd1
3
+ size 2102587
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": 1, "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": 512, "name": "crate-1l_0_512_post"}
4.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:baa26d7f937935f4f7ebf7ed8f024e17f36003085830469c9cd4a4780bda5a75
3
+ size 2102587
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": 1, "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": 512, "name": "crate-1l_0_512_post"}
5.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81640d87a8b6543a1da842974abc63916d22bd33cfd65253bddc93346661e01c
3
+ size 2102587
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": 1, "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": 512, "name": "crate-1l_0_512_post"}
6.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a569aafa1e62dc4823e967a53707e3f6e147d471f1996757c04d65edc0ff77e5
3
+ size 2102587
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": 1, "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": 512, "name": "crate-1l_0_512_post"}
7.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7c7ec720f5c13da42fdf5d61ff2c770bc04592cd4ea150b466f001250b7fdd0b
3
+ size 2102587
7_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": 1, "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": 512, "name": "crate-1l_0_512_post"}
8.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:21cdfba8334dd7686c85c31f7566150c46195b51ba26cc911d729748c1db2149
3
+ size 2102587
8_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": 1, "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": 512, "name": "crate-1l_0_512_post"}
9.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a2371bd8972e714df46e739c6be79df505af3edc48d083a2942023f2a5d1b96
3
+ size 2102587
9_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": 1, "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": 512, "name": "crate-1l_0_512_post"}