Hugging Face's logo

Datasets:
ncgc
/
influence_results

Dataset card Files
xet
 Community
ncgc commited on
Commit
d9a9bae
·
verified ·
1 Parent(s): a7228fd

Add files using upload-large-folder tool

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/activation_covariance.safetensors +3 -0
  2. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/activation_eigenvalues.safetensors +3 -0
  3. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/activation_eigenvectors.safetensors +3 -0
  4. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/covariance_dataset_metadata.json +5 -0
  5. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/factor_arguments.json +20 -0
  6. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/gradient_covariance.safetensors +3 -0
  7. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/gradient_eigenvalues.safetensors +3 -0
  8. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/gradient_eigenvectors.safetensors +3 -0
  9. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/lambda_dataset_metadata.json +5 -0
  10. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/lambda_matrix.safetensors +3 -0
  11. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/num_activation_covariance_processed.safetensors +3 -0
  12. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/num_gradient_covariance_processed.safetensors +3 -0
  13. bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/num_lambda_processed.safetensors +3 -0
  14. bias_statichh-Qwen-3B-sft-bf16/scores_ekfac_half/factor_arguments.json +20 -0
  15. bias_statichh-Qwen-3B-sft-bf16/scores_ekfac_half/query_dataset_metadata.json +5 -0
  16. bias_statichh-Qwen-3B-sft-bf16/scores_ekfac_half/score_arguments.json +19 -0
  17. bias_statichh-Qwen-3B-sft-bf16/scores_ekfac_half/train_dataset_metadata.json +5 -0
  18. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/activation_covariance.safetensors +3 -0
  19. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/activation_eigenvalues.safetensors +3 -0
  20. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/activation_eigenvectors.safetensors +3 -0
  21. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/covariance_dataset_metadata.json +5 -0
  22. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/factor_arguments.json +20 -0
  23. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/gradient_covariance.safetensors +3 -0
  24. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/gradient_eigenvalues.safetensors +3 -0
  25. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/lambda_dataset_metadata.json +5 -0
  26. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/num_activation_covariance_processed.safetensors +3 -0
  27. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/num_gradient_covariance_processed.safetensors +3 -0
  28. bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/num_lambda_processed.safetensors +3 -0
  29. bias_statichh-pythia-2.8b-sft-bf16/query_grads_pure_ekfac_bottom100/aggregated_query_grads.safetensors +3 -0
  30. bias_statichh-pythia-2.8b-sft-bf16/query_grads_pure_ekfac_bottom100/timing.json +8 -0
  31. bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/factor_arguments.json +20 -0
  32. bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/pairwise_scores.safetensors +3 -0
  33. bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/query_dataset_metadata.json +5 -0
  34. bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/score_arguments.json +19 -0
  35. bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/train_dataset_metadata.json +5 -0
  36. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/activation_covariance.safetensors +3 -0
  37. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/activation_eigenvalues.safetensors +3 -0
  38. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/activation_eigenvectors.safetensors +3 -0
  39. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/covariance_dataset_metadata.json +5 -0
  40. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/factor_arguments.json +20 -0
  41. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/gradient_covariance.safetensors +3 -0
  42. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/gradient_eigenvalues.safetensors +3 -0
  43. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/gradient_eigenvectors.safetensors +3 -0
  44. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/lambda_dataset_metadata.json +5 -0
  45. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/lambda_matrix.safetensors +3 -0
  46. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/num_activation_covariance_processed.safetensors +3 -0
  47. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/num_gradient_covariance_processed.safetensors +3 -0
  48. ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/num_lambda_processed.safetensors +3 -0
  49. ethics_statichh-pythia-1.4b-sft-bf16/query_grads_pure_ekfac_bottom100/aggregated_query_grads.safetensors +3 -0
  50. ethics_statichh-pythia-1.4b-sft-bf16/scores_ekfac_half/factor_arguments.json +20 -0
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/activation_covariance.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:546170749097e540375e564623b2db1ea730cc900443825383f8732ae8e9f753
3
+ size 292709750
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/activation_eigenvalues.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:086bed2c5bec5229d49491a94f36f0174106795e75dd3de32d9a63f79d6ca933
3
+ size 47902
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/activation_eigenvectors.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd04879d869a97b1fb9b7c58d7fad515443d9728a79821bf65eece40f004b4ca
3
+ size 292709750
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/covariance_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 10000,
4
+ "indices": null
5
+ }
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/factor_arguments.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "strategy": "ekfac",
3
+ "use_empirical_fisher": false,
4
+ "amp_dtype": "torch.bfloat16",
5
+ "amp_scale": 65536.0,
6
+ "has_shared_parameters": false,
7
+ "covariance_max_examples": 100000,
8
+ "covariance_data_partitions": 1,
9
+ "covariance_module_partitions": 1,
10
+ "activation_covariance_dtype": "torch.bfloat16",
11
+ "gradient_covariance_dtype": "torch.bfloat16",
12
+ "eigendecomposition_dtype": "torch.float64",
13
+ "lambda_max_examples": 100000,
14
+ "lambda_data_partitions": 1,
15
+ "lambda_module_partitions": 1,
16
+ "use_iterative_lambda_aggregation": false,
17
+ "offload_activations_to_cpu": false,
18
+ "per_sample_gradient_dtype": "torch.bfloat16",
19
+ "lambda_dtype": "torch.bfloat16"
20
+ }
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/gradient_covariance.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2032c06cf8a0c38c6b1de7b3ea2b40c60284e6d9b5775eac0982b1a899e5e70e
3
+ size 510133616
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/gradient_eigenvalues.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:acdad784b8705d2d79b7faa10a68a51b6403d6da4f016631c929e0915c3c5ec5
3
+ size 58648
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/gradient_eigenvectors.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0b4a72c40c3d1f294dcc6682a9a08542a84a5256352f4ae9f8ff70aa78af292
3
+ size 510133616
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/lambda_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 10000,
4
+ "indices": null
5
+ }
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/lambda_matrix.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bac8deecbc931165311c1baf2cd68ced54cfe3c9839e99cf4447f04d15d2ff16
3
+ size 154147184
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/num_activation_covariance_processed.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba49c750b64cfbdb978452deee1de471531634eb797629dde1920def0fc21873
3
+ size 1296
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/num_gradient_covariance_processed.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b5bcde82521a6ea3a11e796c33e500d825f7721a995acf1cf4fd776bc95f709
3
+ size 1296
bias_statichh-Qwen-3B-sft-bf16/factors_ekfac_half/num_lambda_processed.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4a0929e93a198b6a3b75294152eb30c86d47f196c3935edf96fa2e064cabe87
3
+ size 1296
bias_statichh-Qwen-3B-sft-bf16/scores_ekfac_half/factor_arguments.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "strategy": "ekfac",
3
+ "use_empirical_fisher": false,
4
+ "amp_dtype": "torch.bfloat16",
5
+ "amp_scale": 65536.0,
6
+ "has_shared_parameters": false,
7
+ "covariance_max_examples": 100000,
8
+ "covariance_data_partitions": 1,
9
+ "covariance_module_partitions": 1,
10
+ "activation_covariance_dtype": "torch.bfloat16",
11
+ "gradient_covariance_dtype": "torch.bfloat16",
12
+ "eigendecomposition_dtype": "torch.float64",
13
+ "lambda_max_examples": 100000,
14
+ "lambda_data_partitions": 1,
15
+ "lambda_module_partitions": 1,
16
+ "use_iterative_lambda_aggregation": false,
17
+ "offload_activations_to_cpu": false,
18
+ "per_sample_gradient_dtype": "torch.bfloat16",
19
+ "lambda_dtype": "torch.bfloat16"
20
+ }
bias_statichh-Qwen-3B-sft-bf16/scores_ekfac_half/query_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 936,
4
+ "indices": null
5
+ }
bias_statichh-Qwen-3B-sft-bf16/scores_ekfac_half/score_arguments.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "damping_factor": 1e-08,
3
+ "amp_dtype": "torch.bfloat16",
4
+ "offload_activations_to_cpu": false,
5
+ "data_partitions": 1,
6
+ "module_partitions": 1,
7
+ "compute_per_module_scores": false,
8
+ "compute_per_token_scores": false,
9
+ "query_gradient_accumulation_steps": 1,
10
+ "query_gradient_low_rank": null,
11
+ "use_full_svd": false,
12
+ "aggregate_query_gradients": true,
13
+ "aggregate_train_gradients": false,
14
+ "use_measurement_for_self_influence": false,
15
+ "query_gradient_svd_dtype": "torch.float32",
16
+ "per_sample_gradient_dtype": "torch.bfloat16",
17
+ "precondition_dtype": "torch.bfloat16",
18
+ "score_dtype": "torch.bfloat16"
19
+ }
bias_statichh-Qwen-3B-sft-bf16/scores_ekfac_half/train_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 96256,
4
+ "indices": null
5
+ }
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/activation_covariance.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7e01b282209dff7b55eb2aec3f95ed16be2dc4451e6f0e34efb208105d5d7e1
3
+ size 249109584
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/activation_eigenvalues.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4a9a392c1127ea1f94f671c3d39d66c80eddf66c73fc74bdc2bf846d09a281b9
3
+ size 36896
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/activation_eigenvectors.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b7c227a7dafc8c79f5d936d536fee4531f0632239f78cb2e855dc90a8daaa79
3
+ size 249109584
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/covariance_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 10000,
4
+ "indices": null
5
+ }
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/factor_arguments.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "strategy": "ekfac",
3
+ "use_empirical_fisher": false,
4
+ "amp_dtype": "torch.bfloat16",
5
+ "amp_scale": 65536.0,
6
+ "has_shared_parameters": false,
7
+ "covariance_max_examples": 100000,
8
+ "covariance_data_partitions": 1,
9
+ "covariance_module_partitions": 1,
10
+ "activation_covariance_dtype": "torch.bfloat16",
11
+ "gradient_covariance_dtype": "torch.bfloat16",
12
+ "eigendecomposition_dtype": "torch.float64",
13
+ "lambda_max_examples": 100000,
14
+ "lambda_data_partitions": 1,
15
+ "lambda_module_partitions": 1,
16
+ "use_iterative_lambda_aggregation": false,
17
+ "offload_activations_to_cpu": false,
18
+ "per_sample_gradient_dtype": "torch.bfloat16",
19
+ "lambda_dtype": "torch.bfloat16"
20
+ }
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/gradient_covariance.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf7f4ff261428d8d7b294ad23af00ae7e81d4ebe3fb57fba165166035c60cce6
3
+ size 353895496
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/gradient_eigenvalues.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:19b867efe0ce6f05ebe3e58f03a632659af307830c2f1411e14fbeb0333a4ed7
3
+ size 47128
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/lambda_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 10000,
4
+ "indices": null
5
+ }
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/num_activation_covariance_processed.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ffaff847dee7fe7476044f6da0d4fba529f8ea7d6d7f7c1523598f8801739eac
3
+ size 1040
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/num_gradient_covariance_processed.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d61e9bc6f22b585bede28a52ad3997c5f0b5078216eebaa546a724683633a6d
3
+ size 1040
bias_statichh-pythia-2.8b-sft-bf16/factors_ekfac_half/num_lambda_processed.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:39d78ac9e8994afa010721b4fa88b68a36fce52afbcbe44aa04388563e00ce9f
3
+ size 1040
bias_statichh-pythia-2.8b-sft-bf16/query_grads_pure_ekfac_bottom100/aggregated_query_grads.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9a728472cc44d65413d02869efb01e2f830602dd21123a937e09df0be7d97f71
3
+ size 314665424
bias_statichh-pythia-2.8b-sft-bf16/query_grads_pure_ekfac_bottom100/timing.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_name": "ncgc/statichh-pythia-2.8b-sft-bf16",
3
+ "analysis_name": "bias_statichh-pythia-2.8b-sft-bf16",
4
+ "scores_type": "pure_ekfac",
5
+ "bottom_k": 100,
6
+ "num_indices": 100,
7
+ "elapsed_seconds": 11.82435154914856
8
+ }
bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/factor_arguments.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "strategy": "ekfac",
3
+ "use_empirical_fisher": false,
4
+ "amp_dtype": "torch.bfloat16",
5
+ "amp_scale": 65536.0,
6
+ "has_shared_parameters": false,
7
+ "covariance_max_examples": 100000,
8
+ "covariance_data_partitions": 1,
9
+ "covariance_module_partitions": 1,
10
+ "activation_covariance_dtype": "torch.bfloat16",
11
+ "gradient_covariance_dtype": "torch.bfloat16",
12
+ "eigendecomposition_dtype": "torch.float64",
13
+ "lambda_max_examples": 100000,
14
+ "lambda_data_partitions": 1,
15
+ "lambda_module_partitions": 1,
16
+ "use_iterative_lambda_aggregation": false,
17
+ "offload_activations_to_cpu": false,
18
+ "per_sample_gradient_dtype": "torch.bfloat16",
19
+ "lambda_dtype": "torch.bfloat16"
20
+ }
bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/pairwise_scores.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61ff5e3139286951014d64b65a75e2dcbd4d7c0ca4752f0e9322fa6e095c18c2
3
+ size 193192
bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/query_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 936,
4
+ "indices": null
5
+ }
bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/score_arguments.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "damping_factor": 1e-08,
3
+ "amp_dtype": "torch.bfloat16",
4
+ "offload_activations_to_cpu": false,
5
+ "data_partitions": 1,
6
+ "module_partitions": 1,
7
+ "compute_per_module_scores": false,
8
+ "compute_per_token_scores": false,
9
+ "query_gradient_accumulation_steps": 1,
10
+ "query_gradient_low_rank": null,
11
+ "use_full_svd": false,
12
+ "aggregate_query_gradients": true,
13
+ "aggregate_train_gradients": false,
14
+ "use_measurement_for_self_influence": false,
15
+ "query_gradient_svd_dtype": "torch.float32",
16
+ "per_sample_gradient_dtype": "torch.bfloat16",
17
+ "precondition_dtype": "torch.bfloat16",
18
+ "score_dtype": "torch.bfloat16"
19
+ }
bias_statichh-pythia-2.8b-sft-bf16/scores_ekfac_half/train_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 96256,
4
+ "indices": null
5
+ }
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/activation_covariance.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a695c1d2e265a1e4c9c623514d0c7d0314d1e312f5f7b6edfc20fa47736bbfc5
3
+ size 159442000
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/activation_eigenvalues.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6bf4dcbc0f61e988e87e34ab24bea05226941641a7a171b98f9b5ae3033f466
3
+ size 29728
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/activation_eigenvectors.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:acd81db4762328edcabec82159cb22f25a5bbfdd24f21458e8691ae9be753362
3
+ size 159442000
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/covariance_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 10000,
4
+ "indices": null
5
+ }
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/factor_arguments.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "strategy": "ekfac",
3
+ "use_empirical_fisher": false,
4
+ "amp_dtype": "torch.bfloat16",
5
+ "amp_scale": 65536.0,
6
+ "has_shared_parameters": false,
7
+ "covariance_max_examples": 100000,
8
+ "covariance_data_partitions": 1,
9
+ "covariance_module_partitions": 1,
10
+ "activation_covariance_dtype": "torch.bfloat16",
11
+ "gradient_covariance_dtype": "torch.bfloat16",
12
+ "eigendecomposition_dtype": "torch.float64",
13
+ "lambda_max_examples": 100000,
14
+ "lambda_data_partitions": 1,
15
+ "lambda_module_partitions": 1,
16
+ "use_iterative_lambda_aggregation": false,
17
+ "offload_activations_to_cpu": false,
18
+ "per_sample_gradient_dtype": "torch.bfloat16",
19
+ "lambda_dtype": "torch.bfloat16"
20
+ }
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/gradient_covariance.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a944ede178c84be3adbcbe9b9ae072619e08619b47a98ebcb77cdd0bef0876c3
3
+ size 226493504
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/gradient_eigenvalues.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f2594b58c2da457b3e4a061738ad8a9858232bcd68531052cd13b7d75443845a
3
+ size 37912
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/gradient_eigenvectors.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e46ac4f9819ade05c8d4260a12a6d2089d17360fb5d4db749df6516e21e6ca9
3
+ size 226493504
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/lambda_dataset_metadata.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "type": "Dataset",
3
+ "dataset_size": 10000,
4
+ "indices": null
5
+ }
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/lambda_matrix.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c138255dcf5215f5a4851c66703e5577f1377d50aa39b68d7a1c04925a1e2e38
3
+ size 100701248
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/num_activation_covariance_processed.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cd3cc0e8cc637b4ba9070fc6d6cee74864bf6983d22dd8212c17239716a8f611
3
+ size 1040
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/num_gradient_covariance_processed.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0fd6534be76db3d7981b1b5bb09e186d88d6a23fa879f663fce9bac3918794c9
3
+ size 1040
ethics_statichh-pythia-1.4b-sft-bf16/factors_ekfac_half/num_lambda_processed.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:553147c6f1666d90bcc7a738454cb85bb2aa8c9cded2f47a258fc0ee95dea304
3
+ size 1040
ethics_statichh-pythia-1.4b-sft-bf16/query_grads_pure_ekfac_bottom100/aggregated_query_grads.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2726eccfa754d0e0b970d614193e2966e52d280f586689eb257720d786f2209d
3
+ size 201400784
ethics_statichh-pythia-1.4b-sft-bf16/scores_ekfac_half/factor_arguments.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "strategy": "ekfac",
3
+ "use_empirical_fisher": false,
4
+ "amp_dtype": "torch.bfloat16",
5
+ "amp_scale": 65536.0,
6
+ "has_shared_parameters": false,
7
+ "covariance_max_examples": 100000,
8
+ "covariance_data_partitions": 1,
9
+ "covariance_module_partitions": 1,
10
+ "activation_covariance_dtype": "torch.bfloat16",
11
+ "gradient_covariance_dtype": "torch.bfloat16",
12
+ "eigendecomposition_dtype": "torch.float64",
13
+ "lambda_max_examples": 100000,
14
+ "lambda_data_partitions": 1,
15
+ "lambda_module_partitions": 1,
16
+ "use_iterative_lambda_aggregation": false,
17
+ "offload_activations_to_cpu": false,
18
+ "per_sample_gradient_dtype": "torch.bfloat16",
19
+ "lambda_dtype": "torch.bfloat16"
20
+ }