Hugging Face's logo

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

Model card Files Community
canrager commited on
Commit
79bf48e
·
verified ·
1 Parent(s): afffc5b

Delete gemma-2-2b_vanilla_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_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_0/ae.pt +0 -3
  2. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_0/config.json +0 -27
  3. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_195/ae.pt +0 -3
  4. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_195/config.json +0 -27
  5. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_1953/ae.pt +0 -3
  6. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_1953/config.json +0 -27
  7. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_19531/ae.pt +0 -3
  8. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_19531/config.json +0 -27
  9. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_617/ae.pt +0 -3
  10. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_617/config.json +0 -27
  11. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_6176/ae.pt +0 -3
  12. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_6176/config.json +0 -27
  13. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_61763/ae.pt +0 -3
  14. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_61763/config.json +0 -27
  15. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_0/ae.pt +0 -3
  16. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_0/config.json +0 -27
  17. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_195/ae.pt +0 -3
  18. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_195/config.json +0 -27
  19. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_1953/ae.pt +0 -3
  20. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_1953/config.json +0 -27
  21. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_19531/ae.pt +0 -3
  22. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_19531/config.json +0 -27
  23. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_617/ae.pt +0 -3
  24. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_617/config.json +0 -27
  25. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_6176/ae.pt +0 -3
  26. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_6176/config.json +0 -27
  27. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_61763/ae.pt +0 -3
  28. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_61763/config.json +0 -27
  29. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_0/ae.pt +0 -3
  30. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_0/config.json +0 -27
  31. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_195/ae.pt +0 -3
  32. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_195/config.json +0 -27
  33. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_1953/ae.pt +0 -3
  34. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_1953/config.json +0 -27
  35. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_19531/ae.pt +0 -3
  36. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_19531/config.json +0 -27
  37. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_617/ae.pt +0 -3
  38. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_617/config.json +0 -27
  39. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_6176/ae.pt +0 -3
  40. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_6176/config.json +0 -27
  41. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_61763/ae.pt +0 -3
  42. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_61763/config.json +0 -27
  43. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_0/ae.pt +0 -3
  44. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_0/config.json +0 -27
  45. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_195/ae.pt +0 -3
  46. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_195/config.json +0 -27
  47. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_1953/ae.pt +0 -3
  48. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_1953/config.json +0 -27
  49. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_19531/ae.pt +0 -3
  50. gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_19531/config.json +0 -27
gemma-2-2b_vanilla_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 sha256:128cb0fe4243dbb2ec45b6e104d765dae7fba4e6ed08cdb5e9136bac1d1d8c22
3
- size 1208232760
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_0/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.025,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "0"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_195/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:e2719944aa66b4b8a560631fcad5c75b33841c9d7f912952964f5d9cb99b5fec
3
- size 1208232776
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_195/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.025,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "195"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_1953/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:3fc8256b3a3256b6e5418b9d91b5734339cefeaae85e42c82d4b52008a38b27f
3
- size 1208232848
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_1953/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.025,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "1953"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_19531/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:03dee9a5170fbb2615fb26cd612921da7f0418ceb4ba8bdbac70a2ae5d0e9860
3
- size 1208233048
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_19531/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.025,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "19531"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_617/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:43df85ceb189027f81fe27c41aa57fb75d22bc81dbafd20e64147daf97035cdf
3
- size 1208232776
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_617/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.025,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "617"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_6176/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:5c732f580ab880879cdde780baee7d26987f9867b7d7f23bca02e8907769fa58
3
- size 1208232848
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_6176/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.025,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "6176"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_61763/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b4ae0849c49f1767bb915700aae9995dc9bcff76ff3c63dd820df7ef966c28a6
3
- size 1208233048
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_0_step_61763/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.025,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "61763"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_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 sha256:128cb0fe4243dbb2ec45b6e104d765dae7fba4e6ed08cdb5e9136bac1d1d8c22
3
- size 1208232760
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_0/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.035,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "0"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_195/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:94b2d3c50422f04a443b32a439c897098fbbc36aa205194d42848cd016a3b8f7
3
- size 1208232776
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_195/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.035,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "195"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_1953/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:96c629f309743528eae9100f7ad4de436f74372ec3ad4b9df2b8b42766e73928
3
- size 1208232848
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_1953/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.035,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "1953"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_19531/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:1d80023e2d3c4825e43bdf2863afe3be88422263ffbca3e22cf6f1c861197be1
3
- size 1208233048
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_19531/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.035,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "19531"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_617/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c2154ad3037938d4584f8f6b38773f58fe3a28a1d95ca64122baa8b7ca6fe681
3
- size 1208232776
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_617/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.035,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "617"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_6176/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0fc3a384661f113926ac9709e5a0279eb5818f7c95d360459eea20e7783c1929
3
- size 1208232848
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_6176/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.035,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "6176"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_61763/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:f2f5f81d086fdbb0d0c970b1a8592d474224cc8a10771351045fae65e5a11406
3
- size 1208233048
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_1_step_61763/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.035,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "61763"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_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 sha256:128cb0fe4243dbb2ec45b6e104d765dae7fba4e6ed08cdb5e9136bac1d1d8c22
3
- size 1208232760
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_0/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.04,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "0"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_195/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:5c82ad7f17773d76485798e5bbfd41d1854d5d28cdce8015e1b537cf671197d1
3
- size 1208232776
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_195/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.04,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "195"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_1953/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:6cfe58ba4ea5e4aa094a4d7f41bb1d067965ba57ed9c342c81335f62b8863ba8
3
- size 1208232848
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_1953/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.04,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "1953"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_19531/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:64f7c9c06f1e41050ad18309eaf59b11e5d60b9cad6cf323353f67cbf94c6893
3
- size 1208233048
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_19531/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.04,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "19531"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_617/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:375a73377c3ab24b5a2a3ea113530d00b47d856c236430df5841022e8680e79d
3
- size 1208232776
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_617/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.04,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "617"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_6176/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:466756eef1a6aff1f13301308339b0e57502a85914d51a874f95531ef442020c
3
- size 1208232848
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_6176/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.04,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "6176"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_61763/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:a75d62beba010099b105df801d0277ea97ee73b78aa42606ad714a21f2c8fa08
3
- size 1208233048
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_2_step_61763/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.04,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "61763"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_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 sha256:128cb0fe4243dbb2ec45b6e104d765dae7fba4e6ed08cdb5e9136bac1d1d8c22
3
- size 1208232760
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_0/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.05,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "0"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_195/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:7fd81969d804d3b87a344c72d08d26dcb8a65c25dc87d818f875d4e2ac09f347
3
- size 1208232776
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_195/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.05,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "195"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_1953/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:7dd1e2ddd31bf59234b04214e8231f913688d48efb36667e84dc5bcee4e39285
3
- size 1208232848
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_1953/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.05,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "1953"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_19531/ae.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:350beb4c71c62ff10017f1024b8888cffb47e03b4a9773357d800de783ae719e
3
- size 1208233048
 
 
 
 
gemma-2-2b_vanilla_width-2pow16_date-1109/resid_post_layer_5_checkpoints/trainer_3_step_19531/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "trainer": {
3
- "dict_class": "AutoEncoder",
4
- "trainer_class": "StandardTrainer",
5
- "activation_dim": 2304,
6
- "dict_size": 65536,
7
- "lr": 0.0003,
8
- "l1_penalty": 0.05,
9
- "warmup_steps": 1000,
10
- "resample_steps": null,
11
- "device": "cuda:0",
12
- "layer": 5,
13
- "lm_name": "google/gemma-2-2b",
14
- "wandb_name": "StandardTrainer-google/gemma-2-2b-resid_post_layer_5",
15
- "submodule_name": "resid_post_layer_5",
16
- "steps": "19531"
17
- },
18
- "buffer": {
19
- "d_submodule": 2304,
20
- "io": "out",
21
- "n_ctxs": 2048,
22
- "ctx_len": 128,
23
- "refresh_batch_size": 32,
24
- "out_batch_size": 2048,
25
- "device": "cuda:0"
26
- }
27
- }