pietrolesci commited on
Commit
cd4aacd
1 Parent(s): 05353c6

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +221 -0
  2. outputs/README.md +9 -0
  3. outputs/ablations/README.md +7 -0
  4. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/.early_stopping.jsonl +2 -0
  5. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/.hydra/config.yaml +99 -0
  6. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/.hydra/hydra.yaml +210 -0
  7. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/.hydra/overrides.yaml +9 -0
  8. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/active_train.log +132 -0
  9. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/hparams.yaml +99 -0
  10. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/logs/labelled_dataset.parquet +3 -0
  11. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/logs/subpool_ids.jsonl +0 -0
  12. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/tb_logs/version_0/events.out.tfevents.1712190719.gpu-q-19.3169922.0 +3 -0
  13. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/tensorboard_logs.parquet +3 -0
  14. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/.early_stopping.jsonl +4 -0
  15. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/.hydra/config.yaml +99 -0
  16. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/.hydra/hydra.yaml +210 -0
  17. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/.hydra/overrides.yaml +9 -0
  18. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/active_train.log +132 -0
  19. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/hparams.yaml +99 -0
  20. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/logs/labelled_dataset.parquet +3 -0
  21. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/logs/subpool_ids.jsonl +0 -0
  22. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/tb_logs/version_0/events.out.tfevents.1712312687.gpu-q-14.3012723.0 +3 -0
  23. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/tensorboard_logs.parquet +3 -0
  24. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/.early_stopping.jsonl +4 -0
  25. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/.hydra/config.yaml +99 -0
  26. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/.hydra/hydra.yaml +210 -0
  27. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/.hydra/overrides.yaml +9 -0
  28. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/active_train.log +132 -0
  29. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/hparams.yaml +99 -0
  30. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/logs/labelled_dataset.parquet +3 -0
  31. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/logs/subpool_ids.jsonl +0 -0
  32. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/tb_logs/version_0/events.out.tfevents.1712312687.gpu-q-14.3012722.0 +3 -0
  33. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/tensorboard_logs.parquet +3 -0
  34. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/.early_stopping.jsonl +3 -0
  35. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/.hydra/config.yaml +99 -0
  36. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/.hydra/hydra.yaml +210 -0
  37. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/.hydra/overrides.yaml +9 -0
  38. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/active_train.log +132 -0
  39. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/hparams.yaml +99 -0
  40. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/logs/labelled_dataset.parquet +3 -0
  41. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/logs/subpool_ids.jsonl +0 -0
  42. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/tb_logs/version_0/events.out.tfevents.1712349958.gpu-q-11.665611.0 +3 -0
  43. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/tensorboard_logs.parquet +3 -0
  44. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/.early_stopping.jsonl +2 -0
  45. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/.hydra/config.yaml +99 -0
  46. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/.hydra/hydra.yaml +210 -0
  47. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/.hydra/overrides.yaml +9 -0
  48. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/active_train.log +132 -0
  49. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/hparams.yaml +99 -0
  50. outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/logs/labelled_dataset.parquet +3 -0
.gitattributes CHANGED
@@ -53,3 +53,224 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
53
  *.jpg filter=lfs diff=lfs merge=lfs -text
54
  *.jpeg filter=lfs diff=lfs merge=lfs -text
55
  *.webp filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
53
  *.jpg filter=lfs diff=lfs merge=lfs -text
54
  *.jpeg filter=lfs diff=lfs merge=lfs -text
55
  *.webp filter=lfs diff=lfs merge=lfs -text
56
+ outputs/ablations/superextended_budget/bert-base_seals_badge_2023-10-09T18-56-11_active_train_0/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
57
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_badge_2024-03-25T14-54-50_48771791_32/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
58
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_badge_2024-03-25T14-54-50_48771791_33/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
59
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_badge_2024-03-25T14-54-50_48771791_34/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
60
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_badge_2024-03-25T14-54-50_48771791_35/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
61
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_badge_2024-03-25T14-54-50_48771791_36/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
62
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_badge_2024-03-25T14-54-50_48771791_37/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
63
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_badge_2024-03-25T14-54-50_48771791_38/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
64
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_badge_2024-03-25T14-54-50_48771791_39/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
65
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_entropy_2023-12-08T08-48-39_36902592_33/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
66
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_entropy_2023-12-08T08-48-39_36902592_35/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
67
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_entropy_2023-12-08T08-48-39_36902592_37/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
68
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_entropy_2023-12-08T08-48-39_36902592_39/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
69
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_entropy_2023-12-08T08-48-39_36902592_41/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
70
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_entropy_2023-12-08T08-48-39_36902592_43/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
71
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_entropy_2023-12-08T08-48-39_36902592_45/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
72
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_entropy_2023-12-08T08-48-39_36902592_47/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
73
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_ftbertkm_2024-03-25T14-54-50_48771791_40/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
74
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_ftbertkm_2024-03-25T14-54-50_48771791_41/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
75
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_ftbertkm_2024-03-25T14-54-50_48771791_42/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
76
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_ftbertkm_2024-03-25T14-54-50_48771791_43/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
77
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_ftbertkm_2024-03-25T14-54-50_48771791_44/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
78
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_ftbertkm_2024-03-25T14-54-50_48771791_45/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
79
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_ftbertkm_2024-03-25T14-54-50_48771791_46/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
80
+ outputs/main/agnews-business-.01/bert-base-uncased_randomsubset_ftbertkm_2024-03-25T14-54-50_48771791_47/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
81
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_badge_2024-03-25T14-54-50_48771791_16/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
82
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_badge_2024-03-25T14-54-50_48771791_17/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
83
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_badge_2024-03-25T14-54-50_48771791_18/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
84
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_badge_2024-03-25T14-54-50_48771791_19/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
85
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_badge_2024-03-25T14-54-50_48771791_20/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
86
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_badge_2024-03-25T14-54-50_48771791_21/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
87
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_badge_2024-03-25T14-54-50_48771791_22/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
88
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_badge_2024-03-25T14-54-50_48771791_23/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
89
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_entropy_2023-12-08T08-48-39_36902592_17/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
90
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_entropy_2023-12-08T08-48-39_36902592_19/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
91
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_entropy_2023-12-08T08-48-39_36902592_21/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
92
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_entropy_2023-12-08T08-48-39_36902592_23/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
93
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_entropy_2023-12-08T08-48-39_36902592_25/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
94
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_entropy_2023-12-08T08-48-39_36902592_27/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
95
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_entropy_2023-12-08T08-48-39_36902592_29/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
96
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_entropy_2023-12-08T08-48-39_36902592_31/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
97
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_ftbertkm_2024-03-25T14-54-50_48771791_24/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
98
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_ftbertkm_2024-03-25T14-54-50_48771791_25/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
99
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_ftbertkm_2024-03-25T14-54-50_48771791_26/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
100
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_ftbertkm_2024-03-25T14-54-50_48771791_27/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
101
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_ftbertkm_2024-03-25T14-54-50_48771791_28/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
102
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_ftbertkm_2024-03-25T14-54-50_48771791_29/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
103
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_ftbertkm_2024-03-25T14-54-50_48771791_30/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
104
+ outputs/main/agnews-business-.01/bert-base-uncased_seals_ftbertkm_2024-03-25T14-54-50_48771791_31/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
105
+ outputs/main/amazon-agri/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_10/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
106
+ outputs/main/amazon-agri/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_11/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
107
+ outputs/main/amazon-agri/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_12/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
108
+ outputs/main/amazon-agri/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_13/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
109
+ outputs/main/amazon-agri/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_14/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
110
+ outputs/main/amazon-agri/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_15/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
111
+ outputs/main/amazon-agri/bert-base-uncased_seals_badge_2023-10-11T11-26-25_29414693_1/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
112
+ outputs/main/amazon-agri/bert-base-uncased_seals_entropy_2023-10-02T17-40-02_28786438_10/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
113
+ outputs/main/amazon-agri/bert-base-uncased_seals_entropy_2023-10-02T17-40-02_28786438_11/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
114
+ outputs/main/amazon-agri/bert-base-uncased_seals_entropy_2023-10-02T17-40-02_28786438_12/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
115
+ outputs/main/amazon-agri/bert-base-uncased_seals_entropy_2023-10-02T17-40-02_28786438_13/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
116
+ outputs/main/amazon-agri/bert-base-uncased_seals_entropy_2023-10-02T17-40-02_28786438_14/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
117
+ outputs/main/amazon-agri/bert-base-uncased_seals_entropy_2023-10-02T17-40-02_28786438_15/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
118
+ outputs/main/amazon-agri/bert-base-uncased_seals_entropy_2023-10-02T17-40-02_28786438_8/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
119
+ outputs/main/amazon-agri/bert-base-uncased_seals_entropy_2023-10-02T17-40-02_28786438_9/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
120
+ outputs/main/amazon-agri/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_10/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
121
+ outputs/main/amazon-agri/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_11/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
122
+ outputs/main/amazon-agri/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_12/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
123
+ outputs/main/amazon-agri/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_13/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
124
+ outputs/main/amazon-agri/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_14/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
125
+ outputs/main/amazon-agri/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_15/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
126
+ outputs/main/amazon-agri/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_8/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
127
+ outputs/main/amazon-agri/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_9/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
128
+ outputs/main/amazon-multi/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_40/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
129
+ outputs/main/amazon-multi/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_41/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
130
+ outputs/main/amazon-multi/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_42/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
131
+ outputs/main/amazon-multi/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_43/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
132
+ outputs/main/amazon-multi/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_44/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
133
+ outputs/main/amazon-multi/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_45/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
134
+ outputs/main/amazon-multi/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_46/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
135
+ outputs/main/amazon-multi/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_47/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
136
+ outputs/main/amazon-multi/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_10/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
137
+ outputs/main/amazon-multi/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_11/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
138
+ outputs/main/amazon-multi/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_12/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
139
+ outputs/main/amazon-multi/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_13/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
140
+ outputs/main/amazon-multi/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_14/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
141
+ outputs/main/amazon-multi/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_15/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
142
+ outputs/main/amazon-multi/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_8/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
143
+ outputs/main/amazon-multi/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_9/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
144
+ outputs/main/amazon-multi/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_40/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
145
+ outputs/main/amazon-multi/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_41/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
146
+ outputs/main/amazon-multi/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_42/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
147
+ outputs/main/amazon-multi/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_43/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
148
+ outputs/main/amazon-multi/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_44/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
149
+ outputs/main/amazon-multi/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_46/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
150
+ outputs/main/amazon-multi/bert-base-uncased_seals_ftbertkm_2023-10-11T11-53-06_29416993_4/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
151
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_74/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
152
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_75/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
153
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_77/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
154
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_badge_2023-10-09T15-09-23_29222618_79/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
155
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_badge_2023-10-11T11-23-59_29414517_4/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
156
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_badge_2023-10-11T11-28-11_29414827_3/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
157
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_badge_2023-10-11T11-29-50_29414952_3/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
158
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_badge_2023-10-11T11-44-27_29416566_0/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
159
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_badge_2023-10-11T11-55-21_29417051_3/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
160
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_41/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
161
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_42/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
162
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_43/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
163
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_44/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
164
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_46/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
165
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_entropy_2023-10-09T15-15-18_29223088_47/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
166
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_entropy_2023-10-11T11-23-59_29414517_1/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
167
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_72/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
168
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_73/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
169
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_74/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
170
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_75/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
171
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_76/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
172
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_77/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
173
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_78/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
174
+ outputs/main/wikitoxic-.01/bert-base-uncased_seals_ftbertkm_2023-10-09T15-13-57_29223081_79/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
175
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_randomsubset_entropy_2023-12-04T18-17-02_35861566_24/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
176
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_randomsubset_entropy_2023-12-04T18-17-02_35861566_25/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
177
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_randomsubset_entropy_2023-12-04T18-17-02_35861566_26/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
178
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_randomsubset_entropy_2023-12-04T18-17-02_35861566_27/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
179
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_randomsubset_entropy_2023-12-04T18-17-02_35861566_28/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
180
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_randomsubset_entropy_2023-12-04T18-17-02_35861566_29/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
181
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_randomsubset_entropy_2023-12-04T18-17-02_35861566_30/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
182
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_randomsubset_entropy_2023-12-04T18-17-02_35861566_31/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
183
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_32/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
184
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_33/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
185
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_34/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
186
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_35/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
187
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_36/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
188
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_37/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
189
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_38/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
190
+ outputs/other_models/albert-base-v2/agnews-business-.01/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_39/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
191
+ outputs/other_models/albert-base-v2/amazon-agri/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_10/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
192
+ outputs/other_models/albert-base-v2/amazon-agri/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_11/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
193
+ outputs/other_models/albert-base-v2/amazon-agri/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_12/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
194
+ outputs/other_models/albert-base-v2/amazon-agri/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_13/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
195
+ outputs/other_models/albert-base-v2/amazon-agri/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_14/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
196
+ outputs/other_models/albert-base-v2/amazon-agri/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_15/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
197
+ outputs/other_models/albert-base-v2/amazon-agri/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_8/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
198
+ outputs/other_models/albert-base-v2/amazon-agri/albert-base-v2_seals_entropy_2023-12-04T18-17-02_35861566_9/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
199
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_randomsubset_entropy_2023-12-08T08-48-39_36902592_32/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
200
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_randomsubset_entropy_2023-12-08T08-48-39_36902592_34/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
201
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_randomsubset_entropy_2023-12-08T08-48-39_36902592_36/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
202
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_randomsubset_entropy_2023-12-08T08-48-39_36902592_38/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
203
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_randomsubset_entropy_2023-12-08T08-48-39_36902592_40/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
204
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_randomsubset_entropy_2023-12-08T08-48-39_36902592_42/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
205
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_randomsubset_entropy_2023-12-08T08-48-39_36902592_44/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
206
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_randomsubset_entropy_2023-12-08T08-48-39_36902592_46/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
207
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_seals_entropy_2023-12-08T08-48-39_36902592_16/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
208
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_seals_entropy_2023-12-08T08-48-39_36902592_18/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
209
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_seals_entropy_2023-12-08T08-48-39_36902592_20/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
210
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_seals_entropy_2023-12-08T08-48-39_36902592_22/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
211
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_seals_entropy_2023-12-08T08-48-39_36902592_24/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
212
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_seals_entropy_2023-12-08T08-48-39_36902592_26/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
213
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_seals_entropy_2023-12-08T08-48-39_36902592_28/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
214
+ outputs/other_models/bert-tiny/agnews-business-.01/bert-tiny_seals_entropy_2023-12-08T08-48-39_36902592_30/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
215
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_randomsubset_entropy_2023-10-04T10-38-41_28870852_16/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
216
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_randomsubset_entropy_2023-10-04T10-38-41_28870852_17/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
217
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_randomsubset_entropy_2023-10-04T10-38-41_28870852_18/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
218
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_randomsubset_entropy_2023-10-04T10-38-41_28870852_19/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
219
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_randomsubset_entropy_2023-10-04T10-38-41_28870852_20/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
220
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_randomsubset_entropy_2023-10-04T10-38-41_28870852_22/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
221
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_randomsubset_entropy_2023-10-04T10-38-41_28870852_23/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
222
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_seals_entropy_2023-10-04T10-38-41_28870852_10/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
223
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_seals_entropy_2023-10-04T10-38-41_28870852_11/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
224
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_seals_entropy_2023-10-04T10-38-41_28870852_12/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
225
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_seals_entropy_2023-10-04T10-38-41_28870852_13/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
226
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_seals_entropy_2023-10-04T10-38-41_28870852_14/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
227
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_seals_entropy_2023-10-04T10-38-41_28870852_15/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
228
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_seals_entropy_2023-10-04T10-38-41_28870852_8/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
229
+ outputs/other_models/bert-tiny/amazon-agri/bert-tiny_seals_entropy_2023-10-04T10-38-41_28870852_9/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
230
+ outputs/other_models/deberta-v3-base/agnews-business-.01/deberta-v3-base_seals_entropy_2023-12-06T10-52-24_36657787_10/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
231
+ outputs/other_models/deberta-v3-base/agnews-business-.01/deberta-v3-base_seals_entropy_2023-12-06T10-52-24_36657787_11/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
232
+ outputs/other_models/deberta-v3-base/agnews-business-.01/deberta-v3-base_seals_entropy_2023-12-06T10-52-24_36657787_12/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
233
+ outputs/other_models/deberta-v3-base/agnews-business-.01/deberta-v3-base_seals_entropy_2023-12-06T10-52-24_36657787_13/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
234
+ outputs/other_models/deberta-v3-base/agnews-business-.01/deberta-v3-base_seals_entropy_2023-12-06T10-52-24_36657787_14/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
235
+ outputs/other_models/deberta-v3-base/agnews-business-.01/deberta-v3-base_seals_entropy_2023-12-06T10-52-24_36657787_15/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
236
+ outputs/other_models/deberta-v3-base/agnews-business-.01/deberta-v3-base_seals_entropy_2023-12-06T10-52-24_36657787_9/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
237
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_randomsubset_entropy_2023-12-08T18-52-14_36955715_40/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
238
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_randomsubset_entropy_2023-12-08T18-52-14_36955715_41/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
239
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_randomsubset_entropy_2023-12-08T18-52-14_36955715_42/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
240
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_randomsubset_entropy_2023-12-08T18-52-14_36955715_43/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
241
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_randomsubset_entropy_2023-12-08T18-52-14_36955715_44/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
242
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_randomsubset_entropy_2023-12-08T18-52-14_36955715_45/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
243
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_randomsubset_entropy_2023-12-08T18-52-14_36955715_46/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
244
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_randomsubset_entropy_2023-12-08T18-52-14_36955715_47/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
245
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_32/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
246
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_33/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
247
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_34/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
248
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_35/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
249
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_36/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
250
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_37/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
251
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_38/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
252
+ outputs/other_models/gpt2/agnews-business-.01/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_39/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
253
+ outputs/other_models/gpt2/amazon-agri/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_10/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
254
+ outputs/other_models/gpt2/amazon-agri/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_11/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
255
+ outputs/other_models/gpt2/amazon-agri/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_12/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
256
+ outputs/other_models/gpt2/amazon-agri/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_13/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
257
+ outputs/other_models/gpt2/amazon-agri/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_14/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
258
+ outputs/other_models/gpt2/amazon-agri/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_15/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
259
+ outputs/other_models/gpt2/amazon-agri/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_8/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
260
+ outputs/other_models/gpt2/amazon-agri/gpt2_seals_entropy_2023-12-08T18-52-14_36955715_9/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
261
+ outputs/other_models/t5-base/agnews-business-.01/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_32/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
262
+ outputs/other_models/t5-base/agnews-business-.01/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_33/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
263
+ outputs/other_models/t5-base/agnews-business-.01/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_34/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
264
+ outputs/other_models/t5-base/agnews-business-.01/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_35/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
265
+ outputs/other_models/t5-base/agnews-business-.01/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_36/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
266
+ outputs/other_models/t5-base/agnews-business-.01/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_37/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
267
+ outputs/other_models/t5-base/agnews-business-.01/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_38/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
268
+ outputs/other_models/t5-base/agnews-business-.01/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_39/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
269
+ outputs/other_models/t5-base/amazon-agri/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_10/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
270
+ outputs/other_models/t5-base/amazon-agri/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_11/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
271
+ outputs/other_models/t5-base/amazon-agri/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_12/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
272
+ outputs/other_models/t5-base/amazon-agri/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_13/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
273
+ outputs/other_models/t5-base/amazon-agri/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_14/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
274
+ outputs/other_models/t5-base/amazon-agri/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_15/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
275
+ outputs/other_models/t5-base/amazon-agri/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_8/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
276
+ outputs/other_models/t5-base/amazon-agri/t5-base_seals_entropy_2023-12-08T18-54-17_36955942_9/logs/subpool_ids.jsonl filter=lfs diff=lfs merge=lfs -text
outputs/README.md ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ This repo contains the artifacts produced from the experiments reported in the paper "Computationally Efficient Active Learning for Large Imbalanced Datasets" (Lesci and Vlachos, 2024).
2
+
3
+ Each subfolder contains a README.md file describing its contents. In summary:
4
+
5
+ - `main/` contains the main results for all datasets considered created using `bert-base-uncased`
6
+
7
+ - `other_models/` contains the results for two datasets using five additional models
8
+
9
+ - `ablations/` contains the outputs of the ablations for the proposed AnchorAL method
outputs/ablations/README.md ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ These are the ablations for the `anchoral` pool filtering method
2
+
3
+ - datasets: `amazon_agri`
4
+
5
+ - active learning trategies: `entropy`
6
+
7
+ - models: `bert-base-uncased`
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/.early_stopping.jsonl ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ {"best_score":1.0,"best_step":11,"stage":"train","interval":"epoch","stopping_step":21,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
2
+ {"best_score":1.0,"best_step":34,"stage":"train","interval":"epoch","stopping_step":44,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/.hydra/config.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 654321
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 654321
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: ${log_interval}
62
+ enable_progress_bar: ${enable_progress_bar}
63
+ limit_train_batches: ${limit_batches}
64
+ limit_validation_batches: ${limit_batches}
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: ${limit_batches}
70
+ limit_test_batches: ${limit_batches}
71
+ test:
72
+ log_interval: ${log_interval}
73
+ enable_progress_bar: ${enable_progress_bar}
74
+ limit_batches: ${limit_batches}
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 654321
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: ${data_path}/prepared/amazoncat-agri
89
+ processed_path: ${data_path}/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: ${dataset.name}/${model.name}_${strategy.name}_${now:%Y-%m-%d}T${now:%H-%M-%S}
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/.hydra/hydra.yaml ADDED
@@ -0,0 +1,210 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ hydra:
2
+ run:
3
+ dir: ./outputs/${experiment_group}/${run_name}
4
+ sweep:
5
+ dir: ./outputs/multirun/${experiment_group}
6
+ subdir: ${run_name}_${hydra.job.id}
7
+ launcher:
8
+ submitit_folder: ${hydra.sweep.dir}/.submitit/%j
9
+ timeout_min: 360
10
+ cpus_per_task: null
11
+ gpus_per_node: null
12
+ tasks_per_node: 1
13
+ mem_gb: null
14
+ nodes: 1
15
+ name: ${experiment_group}
16
+ stderr_to_stdout: false
17
+ _target_: hydra_plugins.hydra_submitit_launcher.submitit_launcher.SlurmLauncher
18
+ partition: ampere
19
+ qos: null
20
+ comment: null
21
+ constraint: null
22
+ exclude: null
23
+ gres: gpu:1
24
+ cpus_per_gpu: null
25
+ gpus_per_task: null
26
+ mem_per_gpu: null
27
+ mem_per_cpu: null
28
+ account: VLACHOS-SL3-GPU
29
+ signal_delay_s: 120
30
+ max_num_timeout: 0
31
+ additional_parameters: {}
32
+ array_parallelism: 256
33
+ setup:
34
+ - . /etc/profile.d/modules.sh
35
+ - module list
36
+ - echo -e 'loading other modules'
37
+ - module load rhel8/default-amp
38
+ - module load cuda/12.1
39
+ - module load cudnn/8.9_cuda-12.1
40
+ - module list
41
+ sweeper:
42
+ _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
43
+ max_batch_size: null
44
+ params: null
45
+ help:
46
+ app_name: ${hydra.job.name}
47
+ header: '${hydra.help.app_name} is powered by Hydra.
48
+
49
+ '
50
+ footer: 'Powered by Hydra (https://hydra.cc)
51
+
52
+ Use --hydra-help to view Hydra specific help
53
+
54
+ '
55
+ template: '${hydra.help.header}
56
+
57
+ == Configuration groups ==
58
+
59
+ Compose your configuration from those groups (group=option)
60
+
61
+
62
+ $APP_CONFIG_GROUPS
63
+
64
+
65
+ == Config ==
66
+
67
+ Override anything in the config (foo.bar=value)
68
+
69
+
70
+ $CONFIG
71
+
72
+
73
+ ${hydra.help.footer}
74
+
75
+ '
76
+ hydra_help:
77
+ template: 'Hydra (${hydra.runtime.version})
78
+
79
+ See https://hydra.cc for more info.
80
+
81
+
82
+ == Flags ==
83
+
84
+ $FLAGS_HELP
85
+
86
+
87
+ == Configuration groups ==
88
+
89
+ Compose your configuration from those groups (For example, append hydra/job_logging=disabled
90
+ to command line)
91
+
92
+
93
+ $HYDRA_CONFIG_GROUPS
94
+
95
+
96
+ Use ''--cfg hydra'' to Show the Hydra config.
97
+
98
+ '
99
+ hydra_help: ???
100
+ hydra_logging:
101
+ version: 1
102
+ formatters:
103
+ colorlog:
104
+ (): colorlog.ColoredFormatter
105
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
106
+ handlers:
107
+ console:
108
+ class: logging.StreamHandler
109
+ formatter: colorlog
110
+ stream: ext://sys.stdout
111
+ root:
112
+ level: INFO
113
+ handlers:
114
+ - console
115
+ disable_existing_loggers: false
116
+ job_logging:
117
+ version: 1
118
+ formatters:
119
+ simple:
120
+ format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
121
+ colorlog:
122
+ (): colorlog.ColoredFormatter
123
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
124
+ - %(message)s'
125
+ log_colors:
126
+ DEBUG: purple
127
+ INFO: green
128
+ WARNING: yellow
129
+ ERROR: red
130
+ CRITICAL: bold_red
131
+ handlers:
132
+ console:
133
+ class: logging.StreamHandler
134
+ formatter: colorlog
135
+ stream: ext://sys.stdout
136
+ file:
137
+ class: logging.FileHandler
138
+ formatter: simple
139
+ filename: ${hydra.job.name}.log
140
+ root:
141
+ level: INFO
142
+ handlers:
143
+ - console
144
+ - file
145
+ disable_existing_loggers: false
146
+ env: {}
147
+ mode: MULTIRUN
148
+ searchpath: []
149
+ callbacks: {}
150
+ output_subdir: .hydra
151
+ overrides:
152
+ hydra:
153
+ - hydra.launcher.timeout_min=360
154
+ - hydra.mode=MULTIRUN
155
+ task:
156
+ - experiment_group=additional_randomsubset_1000
157
+ - dataset=amazon_agri
158
+ - strategy=randomsubset_entropy
159
+ - data.seed=654321
160
+ - model.seed=654321
161
+ - active_data.seed=654321
162
+ - model.name=bert-base-uncased
163
+ - +launcher=slurm
164
+ - strategy.args.subpool_size=1000
165
+ job:
166
+ name: active_train
167
+ chdir: true
168
+ override_dirname: +launcher=slurm,active_data.seed=654321,data.seed=654321,dataset=amazon_agri,experiment_group=additional_randomsubset_1000,model.name=bert-base-uncased,model.seed=654321,strategy.args.subpool_size=1000,strategy=randomsubset_entropy
169
+ id: '49537867_0'
170
+ num: 0
171
+ config_name: conf
172
+ env_set: {}
173
+ env_copy: []
174
+ config:
175
+ override_dirname:
176
+ kv_sep: '='
177
+ item_sep: ','
178
+ exclude_keys: []
179
+ runtime:
180
+ version: 1.3.2
181
+ version_base: '1.3'
182
+ cwd: /rds/user/pl487/hpc-work/anchoral
183
+ config_sources:
184
+ - path: hydra.conf
185
+ schema: pkg
186
+ provider: hydra
187
+ - path: /rds/user/pl487/hpc-work/anchoral/conf
188
+ schema: file
189
+ provider: main
190
+ - path: hydra_plugins.hydra_colorlog.conf
191
+ schema: pkg
192
+ provider: hydra-colorlog
193
+ - path: ''
194
+ schema: structured
195
+ provider: schema
196
+ output_dir: /rds/user/pl487/hpc-work/anchoral/outputs/multirun/additional_randomsubset_1000/amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0
197
+ choices:
198
+ launcher: slurm
199
+ dataset: amazon_agri
200
+ strategy: randomsubset_entropy
201
+ hydra/env: default
202
+ hydra/callbacks: null
203
+ hydra/job_logging: colorlog
204
+ hydra/hydra_logging: colorlog
205
+ hydra/hydra_help: default
206
+ hydra/help: default
207
+ hydra/sweeper: basic
208
+ hydra/launcher: submitit_slurm
209
+ hydra/output: default
210
+ verbose: false
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/.hydra/overrides.yaml ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ - experiment_group=additional_randomsubset_1000
2
+ - dataset=amazon_agri
3
+ - strategy=randomsubset_entropy
4
+ - data.seed=654321
5
+ - model.seed=654321
6
+ - active_data.seed=654321
7
+ - model.name=bert-base-uncased
8
+ - +launcher=slurm
9
+ - strategy.args.subpool_size=1000
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/active_train.log ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [2024-04-04 01:31:45,918][hydra][INFO] -
2
+ estimator:
3
+ accelerator: gpu
4
+ precision: 32
5
+ deterministic: true
6
+ tf32_mode: high
7
+ callbacks:
8
+ timer:
9
+ _target_: energizer.active_learning.callbacks.Timer
10
+ save_outputs:
11
+ _target_: src.callbacks.SaveOutputs
12
+ dirpath: ./logs/
13
+ instance_level: false
14
+ batch_level: false
15
+ epoch_level: false
16
+ early_stopping:
17
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
18
+ monitor: train/avg_f1_minclass
19
+ stage: train
20
+ interval: epoch
21
+ mode: max
22
+ min_delta: 1.0e-05
23
+ patience: 10
24
+ stopping_threshold: null
25
+ divergence_threshold: null
26
+ verbose: true
27
+ model_checkpoint:
28
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
29
+ dirpath: .checkpoints
30
+ monitor: train/avg_f1_minclass
31
+ stage: train
32
+ mode: max
33
+ save_last: false
34
+ save_top_k: 1
35
+ verbose: true
36
+ loggers:
37
+ tensorboard:
38
+ _target_: energizer.loggers.TensorBoardLogger
39
+ root_dir: ./
40
+ name: tb_logs
41
+ version: null
42
+ data:
43
+ batch_size: 32
44
+ eval_batch_size: 256
45
+ num_workers: 32
46
+ pin_memory: true
47
+ drop_last: false
48
+ persistent_workers: true
49
+ shuffle: true
50
+ seed: 654321
51
+ replacement: false
52
+ max_length: 512
53
+ active_data:
54
+ budget: 100
55
+ positive_budget: 5
56
+ seed: 654321
57
+ fit:
58
+ min_steps: 100
59
+ max_epochs: 10
60
+ learning_rate: 4.0e-05
61
+ optimizer: adamw
62
+ log_interval: 1
63
+ enable_progress_bar: false
64
+ limit_train_batches: null
65
+ limit_validation_batches: null
66
+ active_fit:
67
+ max_budget: 5000
68
+ query_size: 25
69
+ reinit_model: true
70
+ limit_pool_batches: null
71
+ limit_test_batches: null
72
+ test:
73
+ log_interval: 1
74
+ enable_progress_bar: false
75
+ limit_batches: null
76
+ strategy:
77
+ name: randomsubset_entropy
78
+ args:
79
+ seed: 42
80
+ subpool_size: 1000
81
+ model:
82
+ name: bert-base-uncased
83
+ seed: 654321
84
+ dataset:
85
+ name: amazon-agri
86
+ text_column: text
87
+ label_column: labels
88
+ uid_column: uid
89
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
90
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
91
+ minority_classes:
92
+ - 1
93
+ index_metric: all-mpnet-base-v2_cosine
94
+ log_interval: 1
95
+ enable_progress_bar: false
96
+ limit_batches: null
97
+ seed: 42
98
+ experiment_group: additional_randomsubset_1000
99
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
100
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
101
+
102
+ ======================================================================
103
+ [2024-04-04 01:31:45,934][hydra][INFO] - Running active learning with strategy {'name': 'randomsubset_entropy', 'args': {'seed': 42, 'subpool_size': 1000}}
104
+ [2024-04-04 01:31:45,995][hydra][INFO] - Seed enabled: 42
105
+ [2024-04-04 01:31:54,137][hydra][INFO] - Labelled size: 100 Pool size: 1186139 Test size: 5285
106
+ Label distribution:
107
+ | | labels | count | perc |
108
+ |---:|:---------|--------:|-------:|
109
+ | 0 | Negative | 95 | 0.95 |
110
+ | 1 | Positive | 5 | 0.05 |
111
+ [2024-04-04 01:31:54,238][hydra][INFO] - Batch:
112
+ {<InputKeys.INPUT_IDS: 'input_ids'>: tensor([[ 101, 3521, 13903, 102]]), <InputKeys.ATT_MASK: 'attention_mask'>: tensor([[1, 1, 1, 1]]), <InputKeys.LABELS: 'labels'>: tensor([0]), <InputKeys.ON_CPU: 'on_cpu'>: {<SpecialKeys.ID: 'uid'>: [1462254]}}
113
+ [2024-04-04 01:31:59,115][hydra][INFO] - Loggers: {'tensorboard': <energizer.loggers.tensorboard.TensorBoardLogger object at 0x1465666bc310>}
114
+ [2024-04-04 01:31:59,115][hydra][INFO] - Callbacks: {'timer': <energizer.active_learning.callbacks.Timer object at 0x1462bd3c2370>, 'save_outputs': <src.callbacks.SaveOutputs object at 0x1462bd3c2340>, 'early_stopping': <energizer.callbacks.early_stopping.EarlyStopping object at 0x1462bd3c2e80>, 'model_checkpoint': <energizer.callbacks.model_checkpoint.ModelCheckpoint object at 0x146139ebd7f0>}
115
+ [2024-04-04 01:31:59,138][hydra][INFO] -
116
+ | Name | Type | Params
117
+ -----------------------------------------
118
+ 0 | bert | BertModel | 109 M
119
+ 1 | dropout | Dropout | 0
120
+ 2 | classifier | Linear | 1.5 K
121
+ -----------------------------------------
122
+ 109 M Trainable params
123
+ 0 Non-trainable params
124
+ 109 M Total params
125
+ 437.935 Total estimated model params size (MB)
126
+ 0.00 GB CUDA Memory used
127
+ [2024-04-04 07:28:59,149][submitit][INFO] - Job has timed out. Ran 357 minutes out of requested 360 minutes.
128
+ [2024-04-04 07:28:59,196][submitit][WARNING] - Caught signal SIGUSR2 on gpu-q-19: this job is timed-out.
129
+ [2024-04-04 07:28:59,216][submitit][INFO] - Calling checkpoint method.
130
+ [2024-04-04 07:28:59,286][submitit][INFO] - Job not requeued because: timed-out too many times.
131
+ [2024-04-04 07:28:59,286][submitit][WARNING] - Bypassing signal SIGCONT
132
+ [2024-04-04 07:28:59,303][submitit][INFO] - Job completed successfully
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/hparams.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 654321
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 654321
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: 1
62
+ enable_progress_bar: false
63
+ limit_train_batches: null
64
+ limit_validation_batches: null
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: null
70
+ limit_test_batches: null
71
+ test:
72
+ log_interval: 1
73
+ enable_progress_bar: false
74
+ limit_batches: null
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 654321
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
89
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/logs/labelled_dataset.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1722c0291bb5a4876a365982b5e7cdb13d0ccb67775b4b06041803403a46474c
3
+ size 6285837
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/logs/subpool_ids.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/tb_logs/version_0/events.out.tfevents.1712190719.gpu-q-19.3169922.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60547cc7a054714358acf8b5514df314746814bcf264769cb28344e8f2fd15cf
3
+ size 7818912
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_0/tensorboard_logs.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3540628c61a741b2b78235a0fe74ff1b96944f835a6bd83b64cd23f0da4d5d53
3
+ size 1369944
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/.early_stopping.jsonl ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {"best_score":0.1875,"best_step":0,"stage":"train","interval":"epoch","stopping_step":10,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
2
+ {"best_score":1.0,"best_step":23,"stage":"train","interval":"epoch","stopping_step":33,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
3
+ {"best_score":0.0,"best_step":86,"stage":"train","interval":"epoch","stopping_step":96,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
4
+ {"best_score":0.08,"best_step":97,"stage":"train","interval":"epoch","stopping_step":107,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/.hydra/config.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 654321
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 123456
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: ${log_interval}
62
+ enable_progress_bar: ${enable_progress_bar}
63
+ limit_train_batches: ${limit_batches}
64
+ limit_validation_batches: ${limit_batches}
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: ${limit_batches}
70
+ limit_test_batches: ${limit_batches}
71
+ test:
72
+ log_interval: ${log_interval}
73
+ enable_progress_bar: ${enable_progress_bar}
74
+ limit_batches: ${limit_batches}
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 654321
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: ${data_path}/prepared/amazoncat-agri
89
+ processed_path: ${data_path}/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: ${dataset.name}/${model.name}_${strategy.name}_${now:%Y-%m-%d}T${now:%H-%M-%S}
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/.hydra/hydra.yaml ADDED
@@ -0,0 +1,210 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ hydra:
2
+ run:
3
+ dir: ./outputs/${experiment_group}/${run_name}
4
+ sweep:
5
+ dir: ./outputs/multirun/${experiment_group}
6
+ subdir: ${run_name}_${hydra.job.id}
7
+ launcher:
8
+ submitit_folder: ${hydra.sweep.dir}/.submitit/%j
9
+ timeout_min: 360
10
+ cpus_per_task: null
11
+ gpus_per_node: null
12
+ tasks_per_node: 1
13
+ mem_gb: null
14
+ nodes: 1
15
+ name: ${experiment_group}
16
+ stderr_to_stdout: false
17
+ _target_: hydra_plugins.hydra_submitit_launcher.submitit_launcher.SlurmLauncher
18
+ partition: ampere
19
+ qos: null
20
+ comment: null
21
+ constraint: null
22
+ exclude: null
23
+ gres: gpu:1
24
+ cpus_per_gpu: null
25
+ gpus_per_task: null
26
+ mem_per_gpu: null
27
+ mem_per_cpu: null
28
+ account: VLACHOS-SL3-GPU
29
+ signal_delay_s: 120
30
+ max_num_timeout: 0
31
+ additional_parameters: {}
32
+ array_parallelism: 256
33
+ setup:
34
+ - . /etc/profile.d/modules.sh
35
+ - module list
36
+ - echo -e 'loading other modules'
37
+ - module load rhel8/default-amp
38
+ - module load cuda/12.1
39
+ - module load cudnn/8.9_cuda-12.1
40
+ - module list
41
+ sweeper:
42
+ _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
43
+ max_batch_size: null
44
+ params: null
45
+ help:
46
+ app_name: ${hydra.job.name}
47
+ header: '${hydra.help.app_name} is powered by Hydra.
48
+
49
+ '
50
+ footer: 'Powered by Hydra (https://hydra.cc)
51
+
52
+ Use --hydra-help to view Hydra specific help
53
+
54
+ '
55
+ template: '${hydra.help.header}
56
+
57
+ == Configuration groups ==
58
+
59
+ Compose your configuration from those groups (group=option)
60
+
61
+
62
+ $APP_CONFIG_GROUPS
63
+
64
+
65
+ == Config ==
66
+
67
+ Override anything in the config (foo.bar=value)
68
+
69
+
70
+ $CONFIG
71
+
72
+
73
+ ${hydra.help.footer}
74
+
75
+ '
76
+ hydra_help:
77
+ template: 'Hydra (${hydra.runtime.version})
78
+
79
+ See https://hydra.cc for more info.
80
+
81
+
82
+ == Flags ==
83
+
84
+ $FLAGS_HELP
85
+
86
+
87
+ == Configuration groups ==
88
+
89
+ Compose your configuration from those groups (For example, append hydra/job_logging=disabled
90
+ to command line)
91
+
92
+
93
+ $HYDRA_CONFIG_GROUPS
94
+
95
+
96
+ Use ''--cfg hydra'' to Show the Hydra config.
97
+
98
+ '
99
+ hydra_help: ???
100
+ hydra_logging:
101
+ version: 1
102
+ formatters:
103
+ colorlog:
104
+ (): colorlog.ColoredFormatter
105
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
106
+ handlers:
107
+ console:
108
+ class: logging.StreamHandler
109
+ formatter: colorlog
110
+ stream: ext://sys.stdout
111
+ root:
112
+ level: INFO
113
+ handlers:
114
+ - console
115
+ disable_existing_loggers: false
116
+ job_logging:
117
+ version: 1
118
+ formatters:
119
+ simple:
120
+ format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
121
+ colorlog:
122
+ (): colorlog.ColoredFormatter
123
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
124
+ - %(message)s'
125
+ log_colors:
126
+ DEBUG: purple
127
+ INFO: green
128
+ WARNING: yellow
129
+ ERROR: red
130
+ CRITICAL: bold_red
131
+ handlers:
132
+ console:
133
+ class: logging.StreamHandler
134
+ formatter: colorlog
135
+ stream: ext://sys.stdout
136
+ file:
137
+ class: logging.FileHandler
138
+ formatter: simple
139
+ filename: ${hydra.job.name}.log
140
+ root:
141
+ level: INFO
142
+ handlers:
143
+ - console
144
+ - file
145
+ disable_existing_loggers: false
146
+ env: {}
147
+ mode: MULTIRUN
148
+ searchpath: []
149
+ callbacks: {}
150
+ output_subdir: .hydra
151
+ overrides:
152
+ hydra:
153
+ - hydra.launcher.timeout_min=360
154
+ - hydra.mode=MULTIRUN
155
+ task:
156
+ - experiment_group=additional_randomsubset_1000
157
+ - dataset=amazon_agri
158
+ - strategy=randomsubset_entropy
159
+ - data.seed=654321
160
+ - model.seed=654321
161
+ - active_data.seed=123456
162
+ - model.name=bert-base-uncased
163
+ - +launcher=slurm
164
+ - strategy.args.subpool_size=1000
165
+ job:
166
+ name: active_train
167
+ chdir: true
168
+ override_dirname: +launcher=slurm,active_data.seed=123456,data.seed=654321,dataset=amazon_agri,experiment_group=additional_randomsubset_1000,model.name=bert-base-uncased,model.seed=654321,strategy.args.subpool_size=1000,strategy=randomsubset_entropy
169
+ id: '49537867_1'
170
+ num: 1
171
+ config_name: conf
172
+ env_set: {}
173
+ env_copy: []
174
+ config:
175
+ override_dirname:
176
+ kv_sep: '='
177
+ item_sep: ','
178
+ exclude_keys: []
179
+ runtime:
180
+ version: 1.3.2
181
+ version_base: '1.3'
182
+ cwd: /rds/user/pl487/hpc-work/anchoral
183
+ config_sources:
184
+ - path: hydra.conf
185
+ schema: pkg
186
+ provider: hydra
187
+ - path: /rds/user/pl487/hpc-work/anchoral/conf
188
+ schema: file
189
+ provider: main
190
+ - path: hydra_plugins.hydra_colorlog.conf
191
+ schema: pkg
192
+ provider: hydra-colorlog
193
+ - path: ''
194
+ schema: structured
195
+ provider: schema
196
+ output_dir: /rds/user/pl487/hpc-work/anchoral/outputs/multirun/additional_randomsubset_1000/amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1
197
+ choices:
198
+ launcher: slurm
199
+ dataset: amazon_agri
200
+ strategy: randomsubset_entropy
201
+ hydra/env: default
202
+ hydra/callbacks: null
203
+ hydra/job_logging: colorlog
204
+ hydra/hydra_logging: colorlog
205
+ hydra/hydra_help: default
206
+ hydra/help: default
207
+ hydra/sweeper: basic
208
+ hydra/launcher: submitit_slurm
209
+ hydra/output: default
210
+ verbose: false
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/.hydra/overrides.yaml ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ - experiment_group=additional_randomsubset_1000
2
+ - dataset=amazon_agri
3
+ - strategy=randomsubset_entropy
4
+ - data.seed=654321
5
+ - model.seed=654321
6
+ - active_data.seed=123456
7
+ - model.name=bert-base-uncased
8
+ - +launcher=slurm
9
+ - strategy.args.subpool_size=1000
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/active_train.log ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [2024-04-05 11:24:12,955][hydra][INFO] -
2
+ estimator:
3
+ accelerator: gpu
4
+ precision: 32
5
+ deterministic: true
6
+ tf32_mode: high
7
+ callbacks:
8
+ timer:
9
+ _target_: energizer.active_learning.callbacks.Timer
10
+ save_outputs:
11
+ _target_: src.callbacks.SaveOutputs
12
+ dirpath: ./logs/
13
+ instance_level: false
14
+ batch_level: false
15
+ epoch_level: false
16
+ early_stopping:
17
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
18
+ monitor: train/avg_f1_minclass
19
+ stage: train
20
+ interval: epoch
21
+ mode: max
22
+ min_delta: 1.0e-05
23
+ patience: 10
24
+ stopping_threshold: null
25
+ divergence_threshold: null
26
+ verbose: true
27
+ model_checkpoint:
28
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
29
+ dirpath: .checkpoints
30
+ monitor: train/avg_f1_minclass
31
+ stage: train
32
+ mode: max
33
+ save_last: false
34
+ save_top_k: 1
35
+ verbose: true
36
+ loggers:
37
+ tensorboard:
38
+ _target_: energizer.loggers.TensorBoardLogger
39
+ root_dir: ./
40
+ name: tb_logs
41
+ version: null
42
+ data:
43
+ batch_size: 32
44
+ eval_batch_size: 256
45
+ num_workers: 32
46
+ pin_memory: true
47
+ drop_last: false
48
+ persistent_workers: true
49
+ shuffle: true
50
+ seed: 654321
51
+ replacement: false
52
+ max_length: 512
53
+ active_data:
54
+ budget: 100
55
+ positive_budget: 5
56
+ seed: 123456
57
+ fit:
58
+ min_steps: 100
59
+ max_epochs: 10
60
+ learning_rate: 4.0e-05
61
+ optimizer: adamw
62
+ log_interval: 1
63
+ enable_progress_bar: false
64
+ limit_train_batches: null
65
+ limit_validation_batches: null
66
+ active_fit:
67
+ max_budget: 5000
68
+ query_size: 25
69
+ reinit_model: true
70
+ limit_pool_batches: null
71
+ limit_test_batches: null
72
+ test:
73
+ log_interval: 1
74
+ enable_progress_bar: false
75
+ limit_batches: null
76
+ strategy:
77
+ name: randomsubset_entropy
78
+ args:
79
+ seed: 42
80
+ subpool_size: 1000
81
+ model:
82
+ name: bert-base-uncased
83
+ seed: 654321
84
+ dataset:
85
+ name: amazon-agri
86
+ text_column: text
87
+ label_column: labels
88
+ uid_column: uid
89
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
90
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
91
+ minority_classes:
92
+ - 1
93
+ index_metric: all-mpnet-base-v2_cosine
94
+ log_interval: 1
95
+ enable_progress_bar: false
96
+ limit_batches: null
97
+ seed: 42
98
+ experiment_group: additional_randomsubset_1000
99
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
100
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
101
+
102
+ ======================================================================
103
+ [2024-04-05 11:24:12,962][hydra][INFO] - Running active learning with strategy {'name': 'randomsubset_entropy', 'args': {'seed': 42, 'subpool_size': 1000}}
104
+ [2024-04-05 11:24:12,989][hydra][INFO] - Seed enabled: 42
105
+ [2024-04-05 11:24:40,874][hydra][INFO] - Labelled size: 100 Pool size: 1186139 Test size: 5285
106
+ Label distribution:
107
+ | | labels | count | perc |
108
+ |---:|:---------|--------:|-------:|
109
+ | 0 | Negative | 95 | 0.95 |
110
+ | 1 | Positive | 5 | 0.05 |
111
+ [2024-04-05 11:24:40,990][hydra][INFO] - Batch:
112
+ {<InputKeys.INPUT_IDS: 'input_ids'>: tensor([[ 101, 3521, 13903, 102]]), <InputKeys.ATT_MASK: 'attention_mask'>: tensor([[1, 1, 1, 1]]), <InputKeys.LABELS: 'labels'>: tensor([0]), <InputKeys.ON_CPU: 'on_cpu'>: {<SpecialKeys.ID: 'uid'>: [1462254]}}
113
+ [2024-04-05 11:24:47,413][hydra][INFO] - Loggers: {'tensorboard': <energizer.loggers.tensorboard.TensorBoardLogger object at 0x148f9578f310>}
114
+ [2024-04-05 11:24:47,413][hydra][INFO] - Callbacks: {'timer': <energizer.active_learning.callbacks.Timer object at 0x148b60e612e0>, 'save_outputs': <src.callbacks.SaveOutputs object at 0x148b68365340>, 'early_stopping': <energizer.callbacks.early_stopping.EarlyStopping object at 0x148b68365310>, 'model_checkpoint': <energizer.callbacks.model_checkpoint.ModelCheckpoint object at 0x148b683652e0>}
115
+ [2024-04-05 11:24:47,481][hydra][INFO] -
116
+ | Name | Type | Params
117
+ -----------------------------------------
118
+ 0 | bert | BertModel | 109 M
119
+ 1 | dropout | Dropout | 0
120
+ 2 | classifier | Linear | 1.5 K
121
+ -----------------------------------------
122
+ 109 M Trainable params
123
+ 0 Non-trainable params
124
+ 109 M Total params
125
+ 437.935 Total estimated model params size (MB)
126
+ 0.00 GB CUDA Memory used
127
+ [2024-04-05 17:21:24,875][submitit][INFO] - Job has timed out. Ran 357 minutes out of requested 360 minutes.
128
+ [2024-04-05 17:21:24,906][submitit][WARNING] - Caught signal SIGUSR2 on gpu-q-14: this job is timed-out.
129
+ [2024-04-05 17:21:24,920][submitit][INFO] - Calling checkpoint method.
130
+ [2024-04-05 17:21:24,943][submitit][INFO] - Job not requeued because: timed-out too many times.
131
+ [2024-04-05 17:21:24,943][submitit][WARNING] - Bypassing signal SIGCONT
132
+ [2024-04-05 17:21:24,947][submitit][INFO] - Job completed successfully
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/hparams.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 654321
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 123456
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: 1
62
+ enable_progress_bar: false
63
+ limit_train_batches: null
64
+ limit_validation_batches: null
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: null
70
+ limit_test_batches: null
71
+ test:
72
+ log_interval: 1
73
+ enable_progress_bar: false
74
+ limit_batches: null
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 654321
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
89
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/logs/labelled_dataset.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:def4df8ca65dde069bb7a06988a3ff5a4e945a4d2cce8d1ae7e1b52db8ff4d19
3
+ size 8998912
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/logs/subpool_ids.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/tb_logs/version_0/events.out.tfevents.1712312687.gpu-q-14.3012723.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bef835c44464799231b9afe159f70720bee016dd9c974babd54d7b3e23d5bfc7
3
+ size 7709249
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_1/tensorboard_logs.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4330fbcf02b8ba34a3ee6a330d347dab379165856a2f22a7b9c0fdf382d2e914
3
+ size 1332545
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/.early_stopping.jsonl ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {"best_score":1.0,"best_step":10,"stage":"train","interval":"epoch","stopping_step":20,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
2
+ {"best_score":1.0,"best_step":33,"stage":"train","interval":"epoch","stopping_step":43,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
3
+ {"best_score":0.0,"best_step":44,"stage":"train","interval":"epoch","stopping_step":54,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
4
+ {"best_score":0.0,"best_step":87,"stage":"train","interval":"epoch","stopping_step":97,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/.hydra/config.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 654321
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 654321
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: ${log_interval}
62
+ enable_progress_bar: ${enable_progress_bar}
63
+ limit_train_batches: ${limit_batches}
64
+ limit_validation_batches: ${limit_batches}
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: ${limit_batches}
70
+ limit_test_batches: ${limit_batches}
71
+ test:
72
+ log_interval: ${log_interval}
73
+ enable_progress_bar: ${enable_progress_bar}
74
+ limit_batches: ${limit_batches}
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 123456
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: ${data_path}/prepared/amazoncat-agri
89
+ processed_path: ${data_path}/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: ${dataset.name}/${model.name}_${strategy.name}_${now:%Y-%m-%d}T${now:%H-%M-%S}
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/.hydra/hydra.yaml ADDED
@@ -0,0 +1,210 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ hydra:
2
+ run:
3
+ dir: ./outputs/${experiment_group}/${run_name}
4
+ sweep:
5
+ dir: ./outputs/multirun/${experiment_group}
6
+ subdir: ${run_name}_${hydra.job.id}
7
+ launcher:
8
+ submitit_folder: ${hydra.sweep.dir}/.submitit/%j
9
+ timeout_min: 360
10
+ cpus_per_task: null
11
+ gpus_per_node: null
12
+ tasks_per_node: 1
13
+ mem_gb: null
14
+ nodes: 1
15
+ name: ${experiment_group}
16
+ stderr_to_stdout: false
17
+ _target_: hydra_plugins.hydra_submitit_launcher.submitit_launcher.SlurmLauncher
18
+ partition: ampere
19
+ qos: null
20
+ comment: null
21
+ constraint: null
22
+ exclude: null
23
+ gres: gpu:1
24
+ cpus_per_gpu: null
25
+ gpus_per_task: null
26
+ mem_per_gpu: null
27
+ mem_per_cpu: null
28
+ account: VLACHOS-SL3-GPU
29
+ signal_delay_s: 120
30
+ max_num_timeout: 0
31
+ additional_parameters: {}
32
+ array_parallelism: 256
33
+ setup:
34
+ - . /etc/profile.d/modules.sh
35
+ - module list
36
+ - echo -e 'loading other modules'
37
+ - module load rhel8/default-amp
38
+ - module load cuda/12.1
39
+ - module load cudnn/8.9_cuda-12.1
40
+ - module list
41
+ sweeper:
42
+ _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
43
+ max_batch_size: null
44
+ params: null
45
+ help:
46
+ app_name: ${hydra.job.name}
47
+ header: '${hydra.help.app_name} is powered by Hydra.
48
+
49
+ '
50
+ footer: 'Powered by Hydra (https://hydra.cc)
51
+
52
+ Use --hydra-help to view Hydra specific help
53
+
54
+ '
55
+ template: '${hydra.help.header}
56
+
57
+ == Configuration groups ==
58
+
59
+ Compose your configuration from those groups (group=option)
60
+
61
+
62
+ $APP_CONFIG_GROUPS
63
+
64
+
65
+ == Config ==
66
+
67
+ Override anything in the config (foo.bar=value)
68
+
69
+
70
+ $CONFIG
71
+
72
+
73
+ ${hydra.help.footer}
74
+
75
+ '
76
+ hydra_help:
77
+ template: 'Hydra (${hydra.runtime.version})
78
+
79
+ See https://hydra.cc for more info.
80
+
81
+
82
+ == Flags ==
83
+
84
+ $FLAGS_HELP
85
+
86
+
87
+ == Configuration groups ==
88
+
89
+ Compose your configuration from those groups (For example, append hydra/job_logging=disabled
90
+ to command line)
91
+
92
+
93
+ $HYDRA_CONFIG_GROUPS
94
+
95
+
96
+ Use ''--cfg hydra'' to Show the Hydra config.
97
+
98
+ '
99
+ hydra_help: ???
100
+ hydra_logging:
101
+ version: 1
102
+ formatters:
103
+ colorlog:
104
+ (): colorlog.ColoredFormatter
105
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
106
+ handlers:
107
+ console:
108
+ class: logging.StreamHandler
109
+ formatter: colorlog
110
+ stream: ext://sys.stdout
111
+ root:
112
+ level: INFO
113
+ handlers:
114
+ - console
115
+ disable_existing_loggers: false
116
+ job_logging:
117
+ version: 1
118
+ formatters:
119
+ simple:
120
+ format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
121
+ colorlog:
122
+ (): colorlog.ColoredFormatter
123
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
124
+ - %(message)s'
125
+ log_colors:
126
+ DEBUG: purple
127
+ INFO: green
128
+ WARNING: yellow
129
+ ERROR: red
130
+ CRITICAL: bold_red
131
+ handlers:
132
+ console:
133
+ class: logging.StreamHandler
134
+ formatter: colorlog
135
+ stream: ext://sys.stdout
136
+ file:
137
+ class: logging.FileHandler
138
+ formatter: simple
139
+ filename: ${hydra.job.name}.log
140
+ root:
141
+ level: INFO
142
+ handlers:
143
+ - console
144
+ - file
145
+ disable_existing_loggers: false
146
+ env: {}
147
+ mode: MULTIRUN
148
+ searchpath: []
149
+ callbacks: {}
150
+ output_subdir: .hydra
151
+ overrides:
152
+ hydra:
153
+ - hydra.launcher.timeout_min=360
154
+ - hydra.mode=MULTIRUN
155
+ task:
156
+ - experiment_group=additional_randomsubset_1000
157
+ - dataset=amazon_agri
158
+ - strategy=randomsubset_entropy
159
+ - data.seed=654321
160
+ - model.seed=123456
161
+ - active_data.seed=654321
162
+ - model.name=bert-base-uncased
163
+ - +launcher=slurm
164
+ - strategy.args.subpool_size=1000
165
+ job:
166
+ name: active_train
167
+ chdir: true
168
+ override_dirname: +launcher=slurm,active_data.seed=654321,data.seed=654321,dataset=amazon_agri,experiment_group=additional_randomsubset_1000,model.name=bert-base-uncased,model.seed=123456,strategy.args.subpool_size=1000,strategy=randomsubset_entropy
169
+ id: '49537867_2'
170
+ num: 2
171
+ config_name: conf
172
+ env_set: {}
173
+ env_copy: []
174
+ config:
175
+ override_dirname:
176
+ kv_sep: '='
177
+ item_sep: ','
178
+ exclude_keys: []
179
+ runtime:
180
+ version: 1.3.2
181
+ version_base: '1.3'
182
+ cwd: /rds/user/pl487/hpc-work/anchoral
183
+ config_sources:
184
+ - path: hydra.conf
185
+ schema: pkg
186
+ provider: hydra
187
+ - path: /rds/user/pl487/hpc-work/anchoral/conf
188
+ schema: file
189
+ provider: main
190
+ - path: hydra_plugins.hydra_colorlog.conf
191
+ schema: pkg
192
+ provider: hydra-colorlog
193
+ - path: ''
194
+ schema: structured
195
+ provider: schema
196
+ output_dir: /rds/user/pl487/hpc-work/anchoral/outputs/multirun/additional_randomsubset_1000/amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2
197
+ choices:
198
+ launcher: slurm
199
+ dataset: amazon_agri
200
+ strategy: randomsubset_entropy
201
+ hydra/env: default
202
+ hydra/callbacks: null
203
+ hydra/job_logging: colorlog
204
+ hydra/hydra_logging: colorlog
205
+ hydra/hydra_help: default
206
+ hydra/help: default
207
+ hydra/sweeper: basic
208
+ hydra/launcher: submitit_slurm
209
+ hydra/output: default
210
+ verbose: false
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/.hydra/overrides.yaml ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ - experiment_group=additional_randomsubset_1000
2
+ - dataset=amazon_agri
3
+ - strategy=randomsubset_entropy
4
+ - data.seed=654321
5
+ - model.seed=123456
6
+ - active_data.seed=654321
7
+ - model.name=bert-base-uncased
8
+ - +launcher=slurm
9
+ - strategy.args.subpool_size=1000
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/active_train.log ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [2024-04-05 11:24:12,956][hydra][INFO] -
2
+ estimator:
3
+ accelerator: gpu
4
+ precision: 32
5
+ deterministic: true
6
+ tf32_mode: high
7
+ callbacks:
8
+ timer:
9
+ _target_: energizer.active_learning.callbacks.Timer
10
+ save_outputs:
11
+ _target_: src.callbacks.SaveOutputs
12
+ dirpath: ./logs/
13
+ instance_level: false
14
+ batch_level: false
15
+ epoch_level: false
16
+ early_stopping:
17
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
18
+ monitor: train/avg_f1_minclass
19
+ stage: train
20
+ interval: epoch
21
+ mode: max
22
+ min_delta: 1.0e-05
23
+ patience: 10
24
+ stopping_threshold: null
25
+ divergence_threshold: null
26
+ verbose: true
27
+ model_checkpoint:
28
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
29
+ dirpath: .checkpoints
30
+ monitor: train/avg_f1_minclass
31
+ stage: train
32
+ mode: max
33
+ save_last: false
34
+ save_top_k: 1
35
+ verbose: true
36
+ loggers:
37
+ tensorboard:
38
+ _target_: energizer.loggers.TensorBoardLogger
39
+ root_dir: ./
40
+ name: tb_logs
41
+ version: null
42
+ data:
43
+ batch_size: 32
44
+ eval_batch_size: 256
45
+ num_workers: 32
46
+ pin_memory: true
47
+ drop_last: false
48
+ persistent_workers: true
49
+ shuffle: true
50
+ seed: 654321
51
+ replacement: false
52
+ max_length: 512
53
+ active_data:
54
+ budget: 100
55
+ positive_budget: 5
56
+ seed: 654321
57
+ fit:
58
+ min_steps: 100
59
+ max_epochs: 10
60
+ learning_rate: 4.0e-05
61
+ optimizer: adamw
62
+ log_interval: 1
63
+ enable_progress_bar: false
64
+ limit_train_batches: null
65
+ limit_validation_batches: null
66
+ active_fit:
67
+ max_budget: 5000
68
+ query_size: 25
69
+ reinit_model: true
70
+ limit_pool_batches: null
71
+ limit_test_batches: null
72
+ test:
73
+ log_interval: 1
74
+ enable_progress_bar: false
75
+ limit_batches: null
76
+ strategy:
77
+ name: randomsubset_entropy
78
+ args:
79
+ seed: 42
80
+ subpool_size: 1000
81
+ model:
82
+ name: bert-base-uncased
83
+ seed: 123456
84
+ dataset:
85
+ name: amazon-agri
86
+ text_column: text
87
+ label_column: labels
88
+ uid_column: uid
89
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
90
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
91
+ minority_classes:
92
+ - 1
93
+ index_metric: all-mpnet-base-v2_cosine
94
+ log_interval: 1
95
+ enable_progress_bar: false
96
+ limit_batches: null
97
+ seed: 42
98
+ experiment_group: additional_randomsubset_1000
99
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
100
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
101
+
102
+ ======================================================================
103
+ [2024-04-05 11:24:12,963][hydra][INFO] - Running active learning with strategy {'name': 'randomsubset_entropy', 'args': {'seed': 42, 'subpool_size': 1000}}
104
+ [2024-04-05 11:24:12,989][hydra][INFO] - Seed enabled: 42
105
+ [2024-04-05 11:24:40,874][hydra][INFO] - Labelled size: 100 Pool size: 1186139 Test size: 5285
106
+ Label distribution:
107
+ | | labels | count | perc |
108
+ |---:|:---------|--------:|-------:|
109
+ | 0 | Negative | 95 | 0.95 |
110
+ | 1 | Positive | 5 | 0.05 |
111
+ [2024-04-05 11:24:40,990][hydra][INFO] - Batch:
112
+ {<InputKeys.INPUT_IDS: 'input_ids'>: tensor([[ 101, 3521, 13903, 102]]), <InputKeys.ATT_MASK: 'attention_mask'>: tensor([[1, 1, 1, 1]]), <InputKeys.LABELS: 'labels'>: tensor([0]), <InputKeys.ON_CPU: 'on_cpu'>: {<SpecialKeys.ID: 'uid'>: [1462254]}}
113
+ [2024-04-05 11:24:47,413][hydra][INFO] - Loggers: {'tensorboard': <energizer.loggers.tensorboard.TensorBoardLogger object at 0x152349aec340>}
114
+ [2024-04-05 11:24:47,413][hydra][INFO] - Callbacks: {'timer': <energizer.active_learning.callbacks.Timer object at 0x151f1513af70>, 'save_outputs': <src.callbacks.SaveOutputs object at 0x151f1513a940>, 'early_stopping': <energizer.callbacks.early_stopping.EarlyStopping object at 0x151f15156580>, 'model_checkpoint': <energizer.callbacks.model_checkpoint.ModelCheckpoint object at 0x151f15156040>}
115
+ [2024-04-05 11:24:47,481][hydra][INFO] -
116
+ | Name | Type | Params
117
+ -----------------------------------------
118
+ 0 | bert | BertModel | 109 M
119
+ 1 | dropout | Dropout | 0
120
+ 2 | classifier | Linear | 1.5 K
121
+ -----------------------------------------
122
+ 109 M Trainable params
123
+ 0 Non-trainable params
124
+ 109 M Total params
125
+ 437.935 Total estimated model params size (MB)
126
+ 0.00 GB CUDA Memory used
127
+ [2024-04-05 17:21:24,817][submitit][INFO] - Job has timed out. Ran 357 minutes out of requested 360 minutes.
128
+ [2024-04-05 17:21:24,845][submitit][WARNING] - Caught signal SIGUSR2 on gpu-q-14: this job is timed-out.
129
+ [2024-04-05 17:21:24,866][submitit][INFO] - Calling checkpoint method.
130
+ [2024-04-05 17:21:24,890][submitit][INFO] - Job not requeued because: timed-out too many times.
131
+ [2024-04-05 17:21:24,890][submitit][WARNING] - Bypassing signal SIGCONT
132
+ [2024-04-05 17:21:24,894][submitit][INFO] - Job completed successfully
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/hparams.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 654321
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 654321
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: 1
62
+ enable_progress_bar: false
63
+ limit_train_batches: null
64
+ limit_validation_batches: null
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: null
70
+ limit_test_batches: null
71
+ test:
72
+ log_interval: 1
73
+ enable_progress_bar: false
74
+ limit_batches: null
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 123456
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
89
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/logs/labelled_dataset.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e2b994b3dcaca8205dbd158ffaee83491928e5015a94102a062933ab4749ccb
3
+ size 11253046
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/logs/subpool_ids.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/tb_logs/version_0/events.out.tfevents.1712312687.gpu-q-14.3012722.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b2bdad78638224bdadf483c3251865ef9af3c5a74d7a182782b1d4d372bdbae1
3
+ size 7725057
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_2/tensorboard_logs.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9a6bf3791e0b2bd9231f1596852765c3d2958b597df27a6a45aa06ba4626ab7c
3
+ size 1308044
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/.early_stopping.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {"best_score":0.5,"best_step":8,"stage":"train","interval":"epoch","stopping_step":18,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
2
+ {"best_score":1.0,"best_step":27,"stage":"train","interval":"epoch","stopping_step":37,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
3
+ {"best_score":0.0,"best_step":38,"stage":"train","interval":"epoch","stopping_step":48,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/.hydra/config.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 654321
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 123456
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: ${log_interval}
62
+ enable_progress_bar: ${enable_progress_bar}
63
+ limit_train_batches: ${limit_batches}
64
+ limit_validation_batches: ${limit_batches}
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: ${limit_batches}
70
+ limit_test_batches: ${limit_batches}
71
+ test:
72
+ log_interval: ${log_interval}
73
+ enable_progress_bar: ${enable_progress_bar}
74
+ limit_batches: ${limit_batches}
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 123456
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: ${data_path}/prepared/amazoncat-agri
89
+ processed_path: ${data_path}/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: ${dataset.name}/${model.name}_${strategy.name}_${now:%Y-%m-%d}T${now:%H-%M-%S}
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/.hydra/hydra.yaml ADDED
@@ -0,0 +1,210 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ hydra:
2
+ run:
3
+ dir: ./outputs/${experiment_group}/${run_name}
4
+ sweep:
5
+ dir: ./outputs/multirun/${experiment_group}
6
+ subdir: ${run_name}_${hydra.job.id}
7
+ launcher:
8
+ submitit_folder: ${hydra.sweep.dir}/.submitit/%j
9
+ timeout_min: 360
10
+ cpus_per_task: null
11
+ gpus_per_node: null
12
+ tasks_per_node: 1
13
+ mem_gb: null
14
+ nodes: 1
15
+ name: ${experiment_group}
16
+ stderr_to_stdout: false
17
+ _target_: hydra_plugins.hydra_submitit_launcher.submitit_launcher.SlurmLauncher
18
+ partition: ampere
19
+ qos: null
20
+ comment: null
21
+ constraint: null
22
+ exclude: null
23
+ gres: gpu:1
24
+ cpus_per_gpu: null
25
+ gpus_per_task: null
26
+ mem_per_gpu: null
27
+ mem_per_cpu: null
28
+ account: VLACHOS-SL3-GPU
29
+ signal_delay_s: 120
30
+ max_num_timeout: 0
31
+ additional_parameters: {}
32
+ array_parallelism: 256
33
+ setup:
34
+ - . /etc/profile.d/modules.sh
35
+ - module list
36
+ - echo -e 'loading other modules'
37
+ - module load rhel8/default-amp
38
+ - module load cuda/12.1
39
+ - module load cudnn/8.9_cuda-12.1
40
+ - module list
41
+ sweeper:
42
+ _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
43
+ max_batch_size: null
44
+ params: null
45
+ help:
46
+ app_name: ${hydra.job.name}
47
+ header: '${hydra.help.app_name} is powered by Hydra.
48
+
49
+ '
50
+ footer: 'Powered by Hydra (https://hydra.cc)
51
+
52
+ Use --hydra-help to view Hydra specific help
53
+
54
+ '
55
+ template: '${hydra.help.header}
56
+
57
+ == Configuration groups ==
58
+
59
+ Compose your configuration from those groups (group=option)
60
+
61
+
62
+ $APP_CONFIG_GROUPS
63
+
64
+
65
+ == Config ==
66
+
67
+ Override anything in the config (foo.bar=value)
68
+
69
+
70
+ $CONFIG
71
+
72
+
73
+ ${hydra.help.footer}
74
+
75
+ '
76
+ hydra_help:
77
+ template: 'Hydra (${hydra.runtime.version})
78
+
79
+ See https://hydra.cc for more info.
80
+
81
+
82
+ == Flags ==
83
+
84
+ $FLAGS_HELP
85
+
86
+
87
+ == Configuration groups ==
88
+
89
+ Compose your configuration from those groups (For example, append hydra/job_logging=disabled
90
+ to command line)
91
+
92
+
93
+ $HYDRA_CONFIG_GROUPS
94
+
95
+
96
+ Use ''--cfg hydra'' to Show the Hydra config.
97
+
98
+ '
99
+ hydra_help: ???
100
+ hydra_logging:
101
+ version: 1
102
+ formatters:
103
+ colorlog:
104
+ (): colorlog.ColoredFormatter
105
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
106
+ handlers:
107
+ console:
108
+ class: logging.StreamHandler
109
+ formatter: colorlog
110
+ stream: ext://sys.stdout
111
+ root:
112
+ level: INFO
113
+ handlers:
114
+ - console
115
+ disable_existing_loggers: false
116
+ job_logging:
117
+ version: 1
118
+ formatters:
119
+ simple:
120
+ format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
121
+ colorlog:
122
+ (): colorlog.ColoredFormatter
123
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
124
+ - %(message)s'
125
+ log_colors:
126
+ DEBUG: purple
127
+ INFO: green
128
+ WARNING: yellow
129
+ ERROR: red
130
+ CRITICAL: bold_red
131
+ handlers:
132
+ console:
133
+ class: logging.StreamHandler
134
+ formatter: colorlog
135
+ stream: ext://sys.stdout
136
+ file:
137
+ class: logging.FileHandler
138
+ formatter: simple
139
+ filename: ${hydra.job.name}.log
140
+ root:
141
+ level: INFO
142
+ handlers:
143
+ - console
144
+ - file
145
+ disable_existing_loggers: false
146
+ env: {}
147
+ mode: MULTIRUN
148
+ searchpath: []
149
+ callbacks: {}
150
+ output_subdir: .hydra
151
+ overrides:
152
+ hydra:
153
+ - hydra.launcher.timeout_min=360
154
+ - hydra.mode=MULTIRUN
155
+ task:
156
+ - experiment_group=additional_randomsubset_1000
157
+ - dataset=amazon_agri
158
+ - strategy=randomsubset_entropy
159
+ - data.seed=654321
160
+ - model.seed=123456
161
+ - active_data.seed=123456
162
+ - model.name=bert-base-uncased
163
+ - +launcher=slurm
164
+ - strategy.args.subpool_size=1000
165
+ job:
166
+ name: active_train
167
+ chdir: true
168
+ override_dirname: +launcher=slurm,active_data.seed=123456,data.seed=654321,dataset=amazon_agri,experiment_group=additional_randomsubset_1000,model.name=bert-base-uncased,model.seed=123456,strategy.args.subpool_size=1000,strategy=randomsubset_entropy
169
+ id: '49537867_3'
170
+ num: 3
171
+ config_name: conf
172
+ env_set: {}
173
+ env_copy: []
174
+ config:
175
+ override_dirname:
176
+ kv_sep: '='
177
+ item_sep: ','
178
+ exclude_keys: []
179
+ runtime:
180
+ version: 1.3.2
181
+ version_base: '1.3'
182
+ cwd: /rds/user/pl487/hpc-work/anchoral
183
+ config_sources:
184
+ - path: hydra.conf
185
+ schema: pkg
186
+ provider: hydra
187
+ - path: /rds/user/pl487/hpc-work/anchoral/conf
188
+ schema: file
189
+ provider: main
190
+ - path: hydra_plugins.hydra_colorlog.conf
191
+ schema: pkg
192
+ provider: hydra-colorlog
193
+ - path: ''
194
+ schema: structured
195
+ provider: schema
196
+ output_dir: /rds/user/pl487/hpc-work/anchoral/outputs/multirun/additional_randomsubset_1000/amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3
197
+ choices:
198
+ launcher: slurm
199
+ dataset: amazon_agri
200
+ strategy: randomsubset_entropy
201
+ hydra/env: default
202
+ hydra/callbacks: null
203
+ hydra/job_logging: colorlog
204
+ hydra/hydra_logging: colorlog
205
+ hydra/hydra_help: default
206
+ hydra/help: default
207
+ hydra/sweeper: basic
208
+ hydra/launcher: submitit_slurm
209
+ hydra/output: default
210
+ verbose: false
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/.hydra/overrides.yaml ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ - experiment_group=additional_randomsubset_1000
2
+ - dataset=amazon_agri
3
+ - strategy=randomsubset_entropy
4
+ - data.seed=654321
5
+ - model.seed=123456
6
+ - active_data.seed=123456
7
+ - model.name=bert-base-uncased
8
+ - +launcher=slurm
9
+ - strategy.args.subpool_size=1000
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/active_train.log ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [2024-04-05 21:45:44,959][hydra][INFO] -
2
+ estimator:
3
+ accelerator: gpu
4
+ precision: 32
5
+ deterministic: true
6
+ tf32_mode: high
7
+ callbacks:
8
+ timer:
9
+ _target_: energizer.active_learning.callbacks.Timer
10
+ save_outputs:
11
+ _target_: src.callbacks.SaveOutputs
12
+ dirpath: ./logs/
13
+ instance_level: false
14
+ batch_level: false
15
+ epoch_level: false
16
+ early_stopping:
17
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
18
+ monitor: train/avg_f1_minclass
19
+ stage: train
20
+ interval: epoch
21
+ mode: max
22
+ min_delta: 1.0e-05
23
+ patience: 10
24
+ stopping_threshold: null
25
+ divergence_threshold: null
26
+ verbose: true
27
+ model_checkpoint:
28
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
29
+ dirpath: .checkpoints
30
+ monitor: train/avg_f1_minclass
31
+ stage: train
32
+ mode: max
33
+ save_last: false
34
+ save_top_k: 1
35
+ verbose: true
36
+ loggers:
37
+ tensorboard:
38
+ _target_: energizer.loggers.TensorBoardLogger
39
+ root_dir: ./
40
+ name: tb_logs
41
+ version: null
42
+ data:
43
+ batch_size: 32
44
+ eval_batch_size: 256
45
+ num_workers: 32
46
+ pin_memory: true
47
+ drop_last: false
48
+ persistent_workers: true
49
+ shuffle: true
50
+ seed: 654321
51
+ replacement: false
52
+ max_length: 512
53
+ active_data:
54
+ budget: 100
55
+ positive_budget: 5
56
+ seed: 123456
57
+ fit:
58
+ min_steps: 100
59
+ max_epochs: 10
60
+ learning_rate: 4.0e-05
61
+ optimizer: adamw
62
+ log_interval: 1
63
+ enable_progress_bar: false
64
+ limit_train_batches: null
65
+ limit_validation_batches: null
66
+ active_fit:
67
+ max_budget: 5000
68
+ query_size: 25
69
+ reinit_model: true
70
+ limit_pool_batches: null
71
+ limit_test_batches: null
72
+ test:
73
+ log_interval: 1
74
+ enable_progress_bar: false
75
+ limit_batches: null
76
+ strategy:
77
+ name: randomsubset_entropy
78
+ args:
79
+ seed: 42
80
+ subpool_size: 1000
81
+ model:
82
+ name: bert-base-uncased
83
+ seed: 123456
84
+ dataset:
85
+ name: amazon-agri
86
+ text_column: text
87
+ label_column: labels
88
+ uid_column: uid
89
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
90
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
91
+ minority_classes:
92
+ - 1
93
+ index_metric: all-mpnet-base-v2_cosine
94
+ log_interval: 1
95
+ enable_progress_bar: false
96
+ limit_batches: null
97
+ seed: 42
98
+ experiment_group: additional_randomsubset_1000
99
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
100
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
101
+
102
+ ======================================================================
103
+ [2024-04-05 21:45:44,966][hydra][INFO] - Running active learning with strategy {'name': 'randomsubset_entropy', 'args': {'seed': 42, 'subpool_size': 1000}}
104
+ [2024-04-05 21:45:44,992][hydra][INFO] - Seed enabled: 42
105
+ [2024-04-05 21:45:52,370][hydra][INFO] - Labelled size: 100 Pool size: 1186139 Test size: 5285
106
+ Label distribution:
107
+ | | labels | count | perc |
108
+ |---:|:---------|--------:|-------:|
109
+ | 0 | Negative | 95 | 0.95 |
110
+ | 1 | Positive | 5 | 0.05 |
111
+ [2024-04-05 21:45:52,524][hydra][INFO] - Batch:
112
+ {<InputKeys.INPUT_IDS: 'input_ids'>: tensor([[ 101, 3521, 13903, 102]]), <InputKeys.ATT_MASK: 'attention_mask'>: tensor([[1, 1, 1, 1]]), <InputKeys.LABELS: 'labels'>: tensor([0]), <InputKeys.ON_CPU: 'on_cpu'>: {<SpecialKeys.ID: 'uid'>: [1462254]}}
113
+ [2024-04-05 21:45:57,989][hydra][INFO] - Loggers: {'tensorboard': <energizer.loggers.tensorboard.TensorBoardLogger object at 0x14ed16274b20>}
114
+ [2024-04-05 21:45:57,989][hydra][INFO] - Callbacks: {'timer': <energizer.active_learning.callbacks.Timer object at 0x14ed0eb50130>, 'save_outputs': <src.callbacks.SaveOutputs object at 0x14ed0eb758e0>, 'early_stopping': <energizer.callbacks.early_stopping.EarlyStopping object at 0x14ed0eb75be0>, 'model_checkpoint': <energizer.callbacks.model_checkpoint.ModelCheckpoint object at 0x14ed161b05b0>}
115
+ [2024-04-05 21:45:58,013][hydra][INFO] -
116
+ | Name | Type | Params
117
+ -----------------------------------------
118
+ 0 | bert | BertModel | 109 M
119
+ 1 | dropout | Dropout | 0
120
+ 2 | classifier | Linear | 1.5 K
121
+ -----------------------------------------
122
+ 109 M Trainable params
123
+ 0 Non-trainable params
124
+ 109 M Total params
125
+ 437.935 Total estimated model params size (MB)
126
+ 0.00 GB CUDA Memory used
127
+ [2024-04-06 03:42:58,881][submitit][INFO] - Job has timed out. Ran 357 minutes out of requested 360 minutes.
128
+ [2024-04-06 03:42:58,950][submitit][WARNING] - Caught signal SIGUSR2 on gpu-q-11: this job is timed-out.
129
+ [2024-04-06 03:42:58,975][submitit][INFO] - Calling checkpoint method.
130
+ [2024-04-06 03:42:59,023][submitit][INFO] - Job not requeued because: timed-out too many times.
131
+ [2024-04-06 03:42:59,023][submitit][WARNING] - Bypassing signal SIGCONT
132
+ [2024-04-06 03:42:59,028][submitit][INFO] - Job completed successfully
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/hparams.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 654321
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 123456
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: 1
62
+ enable_progress_bar: false
63
+ limit_train_batches: null
64
+ limit_validation_batches: null
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: null
70
+ limit_test_batches: null
71
+ test:
72
+ log_interval: 1
73
+ enable_progress_bar: false
74
+ limit_batches: null
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 123456
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
89
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/logs/labelled_dataset.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ce754d2dca678ea42cf3e001792a4da9d72d80a37c90016e17da4832440ff6cf
3
+ size 9931425
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/logs/subpool_ids.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/tb_logs/version_0/events.out.tfevents.1712349958.gpu-q-11.665611.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9378cad52d178e5900a8a7a0058bae572304a7cc9363fd6ddc009e219d7b1ae6
3
+ size 7674684
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_3/tensorboard_logs.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44e805c104f815bc852528ec18b754f1ea1988a203419f6ee476bb46574d374d
3
+ size 1318230
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/.early_stopping.jsonl ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ {"best_score":1.0,"best_step":12,"stage":"train","interval":"epoch","stopping_step":22,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
2
+ {"best_score":1.0,"best_step":35,"stage":"train","interval":"epoch","stopping_step":45,"reason":"Monitored metric `train/avg_f1_minclass` did not improve in the last 10 epochs."}
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/.hydra/config.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 123456
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 654321
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: ${log_interval}
62
+ enable_progress_bar: ${enable_progress_bar}
63
+ limit_train_batches: ${limit_batches}
64
+ limit_validation_batches: ${limit_batches}
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: ${limit_batches}
70
+ limit_test_batches: ${limit_batches}
71
+ test:
72
+ log_interval: ${log_interval}
73
+ enable_progress_bar: ${enable_progress_bar}
74
+ limit_batches: ${limit_batches}
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 654321
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: ${data_path}/prepared/amazoncat-agri
89
+ processed_path: ${data_path}/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: ${dataset.name}/${model.name}_${strategy.name}_${now:%Y-%m-%d}T${now:%H-%M-%S}
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/.hydra/hydra.yaml ADDED
@@ -0,0 +1,210 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ hydra:
2
+ run:
3
+ dir: ./outputs/${experiment_group}/${run_name}
4
+ sweep:
5
+ dir: ./outputs/multirun/${experiment_group}
6
+ subdir: ${run_name}_${hydra.job.id}
7
+ launcher:
8
+ submitit_folder: ${hydra.sweep.dir}/.submitit/%j
9
+ timeout_min: 360
10
+ cpus_per_task: null
11
+ gpus_per_node: null
12
+ tasks_per_node: 1
13
+ mem_gb: null
14
+ nodes: 1
15
+ name: ${experiment_group}
16
+ stderr_to_stdout: false
17
+ _target_: hydra_plugins.hydra_submitit_launcher.submitit_launcher.SlurmLauncher
18
+ partition: ampere
19
+ qos: null
20
+ comment: null
21
+ constraint: null
22
+ exclude: null
23
+ gres: gpu:1
24
+ cpus_per_gpu: null
25
+ gpus_per_task: null
26
+ mem_per_gpu: null
27
+ mem_per_cpu: null
28
+ account: VLACHOS-SL3-GPU
29
+ signal_delay_s: 120
30
+ max_num_timeout: 0
31
+ additional_parameters: {}
32
+ array_parallelism: 256
33
+ setup:
34
+ - . /etc/profile.d/modules.sh
35
+ - module list
36
+ - echo -e 'loading other modules'
37
+ - module load rhel8/default-amp
38
+ - module load cuda/12.1
39
+ - module load cudnn/8.9_cuda-12.1
40
+ - module list
41
+ sweeper:
42
+ _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
43
+ max_batch_size: null
44
+ params: null
45
+ help:
46
+ app_name: ${hydra.job.name}
47
+ header: '${hydra.help.app_name} is powered by Hydra.
48
+
49
+ '
50
+ footer: 'Powered by Hydra (https://hydra.cc)
51
+
52
+ Use --hydra-help to view Hydra specific help
53
+
54
+ '
55
+ template: '${hydra.help.header}
56
+
57
+ == Configuration groups ==
58
+
59
+ Compose your configuration from those groups (group=option)
60
+
61
+
62
+ $APP_CONFIG_GROUPS
63
+
64
+
65
+ == Config ==
66
+
67
+ Override anything in the config (foo.bar=value)
68
+
69
+
70
+ $CONFIG
71
+
72
+
73
+ ${hydra.help.footer}
74
+
75
+ '
76
+ hydra_help:
77
+ template: 'Hydra (${hydra.runtime.version})
78
+
79
+ See https://hydra.cc for more info.
80
+
81
+
82
+ == Flags ==
83
+
84
+ $FLAGS_HELP
85
+
86
+
87
+ == Configuration groups ==
88
+
89
+ Compose your configuration from those groups (For example, append hydra/job_logging=disabled
90
+ to command line)
91
+
92
+
93
+ $HYDRA_CONFIG_GROUPS
94
+
95
+
96
+ Use ''--cfg hydra'' to Show the Hydra config.
97
+
98
+ '
99
+ hydra_help: ???
100
+ hydra_logging:
101
+ version: 1
102
+ formatters:
103
+ colorlog:
104
+ (): colorlog.ColoredFormatter
105
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
106
+ handlers:
107
+ console:
108
+ class: logging.StreamHandler
109
+ formatter: colorlog
110
+ stream: ext://sys.stdout
111
+ root:
112
+ level: INFO
113
+ handlers:
114
+ - console
115
+ disable_existing_loggers: false
116
+ job_logging:
117
+ version: 1
118
+ formatters:
119
+ simple:
120
+ format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
121
+ colorlog:
122
+ (): colorlog.ColoredFormatter
123
+ format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
124
+ - %(message)s'
125
+ log_colors:
126
+ DEBUG: purple
127
+ INFO: green
128
+ WARNING: yellow
129
+ ERROR: red
130
+ CRITICAL: bold_red
131
+ handlers:
132
+ console:
133
+ class: logging.StreamHandler
134
+ formatter: colorlog
135
+ stream: ext://sys.stdout
136
+ file:
137
+ class: logging.FileHandler
138
+ formatter: simple
139
+ filename: ${hydra.job.name}.log
140
+ root:
141
+ level: INFO
142
+ handlers:
143
+ - console
144
+ - file
145
+ disable_existing_loggers: false
146
+ env: {}
147
+ mode: MULTIRUN
148
+ searchpath: []
149
+ callbacks: {}
150
+ output_subdir: .hydra
151
+ overrides:
152
+ hydra:
153
+ - hydra.launcher.timeout_min=360
154
+ - hydra.mode=MULTIRUN
155
+ task:
156
+ - experiment_group=additional_randomsubset_1000
157
+ - dataset=amazon_agri
158
+ - strategy=randomsubset_entropy
159
+ - data.seed=123456
160
+ - model.seed=654321
161
+ - active_data.seed=654321
162
+ - model.name=bert-base-uncased
163
+ - +launcher=slurm
164
+ - strategy.args.subpool_size=1000
165
+ job:
166
+ name: active_train
167
+ chdir: true
168
+ override_dirname: +launcher=slurm,active_data.seed=654321,data.seed=123456,dataset=amazon_agri,experiment_group=additional_randomsubset_1000,model.name=bert-base-uncased,model.seed=654321,strategy.args.subpool_size=1000,strategy=randomsubset_entropy
169
+ id: '49537867_4'
170
+ num: 4
171
+ config_name: conf
172
+ env_set: {}
173
+ env_copy: []
174
+ config:
175
+ override_dirname:
176
+ kv_sep: '='
177
+ item_sep: ','
178
+ exclude_keys: []
179
+ runtime:
180
+ version: 1.3.2
181
+ version_base: '1.3'
182
+ cwd: /rds/user/pl487/hpc-work/anchoral
183
+ config_sources:
184
+ - path: hydra.conf
185
+ schema: pkg
186
+ provider: hydra
187
+ - path: /rds/user/pl487/hpc-work/anchoral/conf
188
+ schema: file
189
+ provider: main
190
+ - path: hydra_plugins.hydra_colorlog.conf
191
+ schema: pkg
192
+ provider: hydra-colorlog
193
+ - path: ''
194
+ schema: structured
195
+ provider: schema
196
+ output_dir: /rds/user/pl487/hpc-work/anchoral/outputs/multirun/additional_randomsubset_1000/amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4
197
+ choices:
198
+ launcher: slurm
199
+ dataset: amazon_agri
200
+ strategy: randomsubset_entropy
201
+ hydra/env: default
202
+ hydra/callbacks: null
203
+ hydra/job_logging: colorlog
204
+ hydra/hydra_logging: colorlog
205
+ hydra/hydra_help: default
206
+ hydra/help: default
207
+ hydra/sweeper: basic
208
+ hydra/launcher: submitit_slurm
209
+ hydra/output: default
210
+ verbose: false
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/.hydra/overrides.yaml ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ - experiment_group=additional_randomsubset_1000
2
+ - dataset=amazon_agri
3
+ - strategy=randomsubset_entropy
4
+ - data.seed=123456
5
+ - model.seed=654321
6
+ - active_data.seed=654321
7
+ - model.name=bert-base-uncased
8
+ - +launcher=slurm
9
+ - strategy.args.subpool_size=1000
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/active_train.log ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [2024-04-05 21:45:44,959][hydra][INFO] -
2
+ estimator:
3
+ accelerator: gpu
4
+ precision: 32
5
+ deterministic: true
6
+ tf32_mode: high
7
+ callbacks:
8
+ timer:
9
+ _target_: energizer.active_learning.callbacks.Timer
10
+ save_outputs:
11
+ _target_: src.callbacks.SaveOutputs
12
+ dirpath: ./logs/
13
+ instance_level: false
14
+ batch_level: false
15
+ epoch_level: false
16
+ early_stopping:
17
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
18
+ monitor: train/avg_f1_minclass
19
+ stage: train
20
+ interval: epoch
21
+ mode: max
22
+ min_delta: 1.0e-05
23
+ patience: 10
24
+ stopping_threshold: null
25
+ divergence_threshold: null
26
+ verbose: true
27
+ model_checkpoint:
28
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
29
+ dirpath: .checkpoints
30
+ monitor: train/avg_f1_minclass
31
+ stage: train
32
+ mode: max
33
+ save_last: false
34
+ save_top_k: 1
35
+ verbose: true
36
+ loggers:
37
+ tensorboard:
38
+ _target_: energizer.loggers.TensorBoardLogger
39
+ root_dir: ./
40
+ name: tb_logs
41
+ version: null
42
+ data:
43
+ batch_size: 32
44
+ eval_batch_size: 256
45
+ num_workers: 32
46
+ pin_memory: true
47
+ drop_last: false
48
+ persistent_workers: true
49
+ shuffle: true
50
+ seed: 123456
51
+ replacement: false
52
+ max_length: 512
53
+ active_data:
54
+ budget: 100
55
+ positive_budget: 5
56
+ seed: 654321
57
+ fit:
58
+ min_steps: 100
59
+ max_epochs: 10
60
+ learning_rate: 4.0e-05
61
+ optimizer: adamw
62
+ log_interval: 1
63
+ enable_progress_bar: false
64
+ limit_train_batches: null
65
+ limit_validation_batches: null
66
+ active_fit:
67
+ max_budget: 5000
68
+ query_size: 25
69
+ reinit_model: true
70
+ limit_pool_batches: null
71
+ limit_test_batches: null
72
+ test:
73
+ log_interval: 1
74
+ enable_progress_bar: false
75
+ limit_batches: null
76
+ strategy:
77
+ name: randomsubset_entropy
78
+ args:
79
+ seed: 42
80
+ subpool_size: 1000
81
+ model:
82
+ name: bert-base-uncased
83
+ seed: 654321
84
+ dataset:
85
+ name: amazon-agri
86
+ text_column: text
87
+ label_column: labels
88
+ uid_column: uid
89
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
90
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
91
+ minority_classes:
92
+ - 1
93
+ index_metric: all-mpnet-base-v2_cosine
94
+ log_interval: 1
95
+ enable_progress_bar: false
96
+ limit_batches: null
97
+ seed: 42
98
+ experiment_group: additional_randomsubset_1000
99
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
100
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
101
+
102
+ ======================================================================
103
+ [2024-04-05 21:45:44,966][hydra][INFO] - Running active learning with strategy {'name': 'randomsubset_entropy', 'args': {'seed': 42, 'subpool_size': 1000}}
104
+ [2024-04-05 21:45:44,992][hydra][INFO] - Seed enabled: 42
105
+ [2024-04-05 21:45:53,167][hydra][INFO] - Labelled size: 100 Pool size: 1186139 Test size: 5285
106
+ Label distribution:
107
+ | | labels | count | perc |
108
+ |---:|:---------|--------:|-------:|
109
+ | 0 | Negative | 95 | 0.95 |
110
+ | 1 | Positive | 5 | 0.05 |
111
+ [2024-04-05 21:45:53,241][hydra][INFO] - Batch:
112
+ {<InputKeys.INPUT_IDS: 'input_ids'>: tensor([[ 101, 3521, 13903, 102]]), <InputKeys.ATT_MASK: 'attention_mask'>: tensor([[1, 1, 1, 1]]), <InputKeys.LABELS: 'labels'>: tensor([0]), <InputKeys.ON_CPU: 'on_cpu'>: {<SpecialKeys.ID: 'uid'>: [1462254]}}
113
+ [2024-04-05 21:45:59,589][hydra][INFO] - Loggers: {'tensorboard': <energizer.loggers.tensorboard.TensorBoardLogger object at 0x1486b90c9df0>}
114
+ [2024-04-05 21:45:59,589][hydra][INFO] - Callbacks: {'timer': <energizer.active_learning.callbacks.Timer object at 0x1486b916c580>, 'save_outputs': <src.callbacks.SaveOutputs object at 0x1486b91514f0>, 'early_stopping': <energizer.callbacks.early_stopping.EarlyStopping object at 0x1486b9151a30>, 'model_checkpoint': <energizer.callbacks.model_checkpoint.ModelCheckpoint object at 0x1486b91517c0>}
115
+ [2024-04-05 21:45:59,612][hydra][INFO] -
116
+ | Name | Type | Params
117
+ -----------------------------------------
118
+ 0 | bert | BertModel | 109 M
119
+ 1 | dropout | Dropout | 0
120
+ 2 | classifier | Linear | 1.5 K
121
+ -----------------------------------------
122
+ 109 M Trainable params
123
+ 0 Non-trainable params
124
+ 109 M Total params
125
+ 437.935 Total estimated model params size (MB)
126
+ 0.00 GB CUDA Memory used
127
+ [2024-04-06 03:42:58,916][submitit][INFO] - Job has timed out. Ran 357 minutes out of requested 360 minutes.
128
+ [2024-04-06 03:42:58,959][submitit][WARNING] - Caught signal SIGUSR2 on gpu-q-45: this job is timed-out.
129
+ [2024-04-06 03:42:58,996][submitit][INFO] - Calling checkpoint method.
130
+ [2024-04-06 03:42:59,035][submitit][INFO] - Job not requeued because: timed-out too many times.
131
+ [2024-04-06 03:42:59,035][submitit][WARNING] - Bypassing signal SIGCONT
132
+ [2024-04-06 03:42:59,040][submitit][INFO] - Job completed successfully
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/hparams.yaml ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ estimator:
2
+ accelerator: gpu
3
+ precision: 32
4
+ deterministic: true
5
+ tf32_mode: high
6
+ callbacks:
7
+ timer:
8
+ _target_: energizer.active_learning.callbacks.Timer
9
+ save_outputs:
10
+ _target_: src.callbacks.SaveOutputs
11
+ dirpath: ./logs/
12
+ instance_level: false
13
+ batch_level: false
14
+ epoch_level: false
15
+ early_stopping:
16
+ _target_: energizer.callbacks.early_stopping.EarlyStopping
17
+ monitor: train/avg_f1_minclass
18
+ stage: train
19
+ interval: epoch
20
+ mode: max
21
+ min_delta: 1.0e-05
22
+ patience: 10
23
+ stopping_threshold: null
24
+ divergence_threshold: null
25
+ verbose: true
26
+ model_checkpoint:
27
+ _target_: energizer.callbacks.model_checkpoint.ModelCheckpoint
28
+ dirpath: .checkpoints
29
+ monitor: train/avg_f1_minclass
30
+ stage: train
31
+ mode: max
32
+ save_last: false
33
+ save_top_k: 1
34
+ verbose: true
35
+ loggers:
36
+ tensorboard:
37
+ _target_: energizer.loggers.TensorBoardLogger
38
+ root_dir: ./
39
+ name: tb_logs
40
+ version: null
41
+ data:
42
+ batch_size: 32
43
+ eval_batch_size: 256
44
+ num_workers: 32
45
+ pin_memory: true
46
+ drop_last: false
47
+ persistent_workers: true
48
+ shuffle: true
49
+ seed: 123456
50
+ replacement: false
51
+ max_length: 512
52
+ active_data:
53
+ budget: 100
54
+ positive_budget: 5
55
+ seed: 654321
56
+ fit:
57
+ min_steps: 100
58
+ max_epochs: 10
59
+ learning_rate: 4.0e-05
60
+ optimizer: adamw
61
+ log_interval: 1
62
+ enable_progress_bar: false
63
+ limit_train_batches: null
64
+ limit_validation_batches: null
65
+ active_fit:
66
+ max_budget: 5000
67
+ query_size: 25
68
+ reinit_model: true
69
+ limit_pool_batches: null
70
+ limit_test_batches: null
71
+ test:
72
+ log_interval: 1
73
+ enable_progress_bar: false
74
+ limit_batches: null
75
+ strategy:
76
+ name: randomsubset_entropy
77
+ args:
78
+ seed: 42
79
+ subpool_size: 1000
80
+ model:
81
+ name: bert-base-uncased
82
+ seed: 654321
83
+ dataset:
84
+ name: amazon-agri
85
+ text_column: text
86
+ label_column: labels
87
+ uid_column: uid
88
+ prepared_path: /rds/user/pl487/hpc-work/anchoral/data/prepared/amazoncat-agri
89
+ processed_path: /rds/user/pl487/hpc-work/anchoral/data/processed/amazoncat-13k
90
+ minority_classes:
91
+ - 1
92
+ index_metric: all-mpnet-base-v2_cosine
93
+ log_interval: 1
94
+ enable_progress_bar: false
95
+ limit_batches: null
96
+ seed: 42
97
+ experiment_group: additional_randomsubset_1000
98
+ run_name: amazon-agri/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18
99
+ data_path: /rds/user/pl487/hpc-work/anchoral/data
outputs/ablations/additional_randomsubset_1000/bert-base-uncased_randomsubset_entropy_2024-04-02T18-37-18_49537867_4/logs/labelled_dataset.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ffcdc2c9172c196926c2f6683bb40435c3cd257e23da4126814bccb35c6aa1a
3
+ size 6286136