Hugging Face's logo

canrager
/
saebench_gemma-2-2b_width-2pow16_date-1109

Model card Files Community
canrager commited on
Commit
2f49cbf
·
verified ·
1 Parent(s): 9f0d5d9

Delete gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_0/ae.pt +0 -3
  2. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_0/config.json +0 -26
  3. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_308/ae.pt +0 -3
  4. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_308/config.json +0 -26
  5. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_3088/ae.pt +0 -3
  6. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_3088/config.json +0 -26
  7. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_30881/ae.pt +0 -3
  8. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_30881/config.json +0 -26
  9. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_97/ae.pt +0 -3
  10. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_97/config.json +0 -26
  11. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_976/ae.pt +0 -3
  12. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_976/config.json +0 -26
  13. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_9765/ae.pt +0 -3
  14. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_9765/config.json +0 -26
  15. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_0/ae.pt +0 -3
  16. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_0/config.json +0 -26
  17. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_308/ae.pt +0 -3
  18. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_308/config.json +0 -26
  19. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_3088/ae.pt +0 -3
  20. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_3088/config.json +0 -26
  21. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_30881/ae.pt +0 -3
  22. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_30881/config.json +0 -26
  23. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_97/ae.pt +0 -3
  24. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_97/config.json +0 -26
  25. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_976/ae.pt +0 -3
  26. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_976/config.json +0 -26
  27. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_9765/ae.pt +0 -3
  28. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_9765/config.json +0 -26
  29. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_0/ae.pt +0 -3
  30. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_0/config.json +0 -26
  31. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_308/ae.pt +0 -3
  32. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_308/config.json +0 -26
  33. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_3088/ae.pt +0 -3
  34. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_3088/config.json +0 -26
  35. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_30881/ae.pt +0 -3
  36. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_30881/config.json +0 -26
  37. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_97/ae.pt +0 -3
  38. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_97/config.json +0 -26
  39. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_976/ae.pt +0 -3
  40. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_976/config.json +0 -26
  41. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_9765/ae.pt +0 -3
  42. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_9765/config.json +0 -26
  43. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_0/ae.pt +0 -3
  44. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_0/config.json +0 -26
  45. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_308/ae.pt +0 -3
  46. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_308/config.json +0 -26
  47. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_3088/ae.pt +0 -3
  48. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_3088/config.json +0 -26
  49. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_30881/ae.pt +0 -3
  50. gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_30881/config.json +0 -26
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_0/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c2fa6020d346a61f1248b174c40ae3a225c988426c118ed376952b74cc14a391
3
- size 1208232760
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_0/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "0",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 20,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_308/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:2936474d9c934443b3d5369b94bdc45053a7f4b2b5ba89cb53fc2651dccce915
3
- size 1208232776
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_308/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "308",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 20,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_3088/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:3e1e3d667af479cce5692ac6aab166258cacf1ed4d480fbd80eeadb32127315d
3
- size 1208232848
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_3088/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "3088",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 20,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_30881/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:1ae2e3b8be3f90444c4496646fb6aa604f0c2c7809b62a0d8608c47d1e697a2c
3
- size 1208233048
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_30881/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "30881",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 20,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_97/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:98396469af5a09409acad191e2fb8637e647029ad7fe6764ea11b847b1471e6d
3
- size 1208232768
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_97/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "97",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 20,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_976/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:72247aad0086f6772c263e029965661f81a79ea11caae94727f16d95c07d313c
3
- size 1208232776
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_976/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "976",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 20,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_9765/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:8de72270cab9c46aa2cdb3c51a7e986bab6f43b2e5512a1d77f4c43613e20e4e
3
- size 1208232848
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_9765/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "9765",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 20,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_0/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c2fa6020d346a61f1248b174c40ae3a225c988426c118ed376952b74cc14a391
3
- size 1208232760
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_0/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "0",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 40,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_308/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:610e1d15ff4fcb05fe0db164879b316f1205732e25850e145091c9cb48bfa09c
3
- size 1208232776
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_308/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "308",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 40,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_3088/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:9d9d3b8aef0f52860267de45ac12977fe86826f35cdb6b87a3829673a37c36cb
3
- size 1208232848
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_3088/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "3088",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 40,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_30881/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:8190d85699dad6aaeef16cad0731e96b63676f479e8f4fe151062e2eece8e4aa
3
- size 1208233048
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_30881/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "30881",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 40,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_97/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:8e5458365646730ae06d8e514b5c728bd501dab0ba227af64729218b2c76dd9d
3
- size 1208232768
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_97/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "97",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 40,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_976/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:e4a2260a96e754b304bdbfda0175f9dde53527e2c3cc76023e1f21eadc6889fd
3
- size 1208232776
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_976/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "976",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 40,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_9765/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c4fafb152c01023057a05eb8d0e2f2b50263ce78eb201cb1269ce0dbbc5c7585
3
- size 1208232848
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_9765/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "9765",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 40,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_0/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c2fa6020d346a61f1248b174c40ae3a225c988426c118ed376952b74cc14a391
3
- size 1208232760
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_0/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "0",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 80,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_308/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:819dec7bf756472ec9ad3ce387cd15ff50e18491d98df4775c70db889f84745e
3
- size 1208232776
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_308/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "308",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 80,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_3088/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:4956f6f43595e7dc95488bf8f4eac64e76b2b22f50d0a488bcc5088a229747c8
3
- size 1208232848
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_3088/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "3088",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 80,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_30881/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:2b8ca264d34fc9c813271d4740f8354d920adc3a8bb2575e010e191f6d652af8
3
- size 1208233048
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_30881/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "30881",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 80,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_97/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:435d42cfbfe4fb8260ba563ecedb62184ff808afd8f7070a8c23b2b1f503838e
3
- size 1208232768
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_97/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "97",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 80,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_976/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:f8af3df52dba4e0b8ae0f464cf3c7c7c2fc55b12e7e1ebd5e325e49f5d487b69
3
- size 1208232776
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_976/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "976",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 80,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_9765/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:025b7d7aa2e0085f3c34e1cff16b900a16cb18bf6e8d8e7984dfead442d22ca2
3
- size 1208232848
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_9765/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "9765",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 80,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_0/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c2fa6020d346a61f1248b174c40ae3a225c988426c118ed376952b74cc14a391
3
- size 1208232760
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_0/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "0",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 160,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_308/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:f1172e3e921375af16476b723059354adca19d37dadbd0e5c022ba9cc633bec2
3
- size 1208232776
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_308/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "308",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 160,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_3088/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:12b637cb1114124801da785a1af70fd65bce57142a6d7bc824308e49caba8a64
3
- size 1208232848
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_3088/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "3088",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 160,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_30881/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:821b99c90698176c46a4ad714484fbe8f816ed7deb4cd1b0fde7b2791d5f0a6e
3
- size 1208233048
 
 
 
 
gemma-2-2b_topk_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_30881/config.json DELETED
@@ -1,26 +0,0 @@
1
- {
2
- "trainer": {
3
- "trainer_class": "TrainerTopK",
4
- "dict_class": "AutoEncoderTopK",
5
- "lr": 0.0001,
6
- "steps": "30881",
7
- "seed": 0,
8
- "activation_dim": 2304,
9
- "dict_size": 65536,
10
- "k": 160,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "TopKTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5"
16
- },
17
- "buffer": {
18
- "d_submodule": 2304,
19
- "io": "out",
20
- "n_ctxs": 2048,
21
- "ctx_len": 128,
22
- "refresh_batch_size": 24,
23
- "out_batch_size": 4096,
24
- "device": "cuda:0"
25
- }
26
- }