thkim0305 commited on
Commit
4a8a570
·
verified ·
1 Parent(s): 0369068

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. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round10.pth +3 -0
  2. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round12.pth +3 -0
  3. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round15.pth +3 -0
  4. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round17.pth +3 -0
  5. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round2.pth +3 -0
  6. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round20.pth +3 -0
  7. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round5.pth +3 -0
  8. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round7.pth +3 -0
  9. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_trainer_state.json +378 -0
  10. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round10.pth +3 -0
  11. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round12.pth +3 -0
  12. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round15.pth +3 -0
  13. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round17.pth +3 -0
  14. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round2.pth +3 -0
  15. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round20.pth +3 -0
  16. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round5.pth +3 -0
  17. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round7.pth +3 -0
  18. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_trainer_state.json +378 -0
  19. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round10.pth +3 -0
  20. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round12.pth +3 -0
  21. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round15.pth +3 -0
  22. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round17.pth +3 -0
  23. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round2.pth +3 -0
  24. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round20.pth +3 -0
  25. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round5.pth +3 -0
  26. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round7.pth +3 -0
  27. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_trainer_state.json +378 -0
  28. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round10.pth +3 -0
  29. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round12.pth +3 -0
  30. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round15.pth +3 -0
  31. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round17.pth +3 -0
  32. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round2.pth +3 -0
  33. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round20.pth +3 -0
  34. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round5.pth +3 -0
  35. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round7.pth +3 -0
  36. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_trainer_state.json +378 -0
  37. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round10.pth +3 -0
  38. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round12.pth +3 -0
  39. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round15.pth +3 -0
  40. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round17.pth +3 -0
  41. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round2.pth +3 -0
  42. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round20.pth +3 -0
  43. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round5.pth +3 -0
  44. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round7.pth +3 -0
  45. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_trainer_state.json +378 -0
  46. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round10.pth +3 -0
  47. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round12.pth +3 -0
  48. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round15.pth +3 -0
  49. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round17.pth +3 -0
  50. client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round2.pth +3 -0
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:93abded2771eae8b38fc450177b9be96b52d6098b2a5f1670ffec1db7787acb9
3
+ size 368443438
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4af2b7a06155662bc2590c3e7d62f0415bc6b31883ec0b62d1ea01e4ac5d536
3
+ size 368443438
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95a72a5403616b0807a64da9ab22d73b174c102bc1f1a5170ff44091cc09cce2
3
+ size 368443438
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round17.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f930dde96502fd535b4ac8de03eb4a6fe96e6ca65ccb8a8245a2baa46cbdc40
3
+ size 368443438
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e1654311c798ae23be901b2384e4a58128109887a3799403026888b2ee21de6
3
+ size 368442474
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round20.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:16f3c8faa82b6d3419c08dabee3ddee34d1660d5eec37f76375e0ac5f612530b
3
+ size 368443438
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a56b50235a0f308a7b3c4e38cd41b9c92ec1ee21f545f5b94e2f8c63139fa17d
3
+ size 368442474
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ced7370f0f14ff7207bd5c193a7c7ebd660a7f2a6cc2be6b15143147dea0605
3
+ size 368442474
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_trainer_state.json ADDED
@@ -0,0 +1,378 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 97,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.020618556701030927,
13
+ "grad_norm": 2.4410085678100586,
14
+ "learning_rate": 2e-05,
15
+ "loss": 1.3332,
16
+ "step": 2
17
+ },
18
+ {
19
+ "epoch": 0.041237113402061855,
20
+ "grad_norm": 2.9084312915802,
21
+ "learning_rate": 2e-05,
22
+ "loss": 1.2347,
23
+ "step": 4
24
+ },
25
+ {
26
+ "epoch": 0.061855670103092786,
27
+ "grad_norm": 1.524423360824585,
28
+ "learning_rate": 2e-05,
29
+ "loss": 1.0421,
30
+ "step": 6
31
+ },
32
+ {
33
+ "epoch": 0.08247422680412371,
34
+ "grad_norm": 3.025947332382202,
35
+ "learning_rate": 2e-05,
36
+ "loss": 1.9983,
37
+ "step": 8
38
+ },
39
+ {
40
+ "epoch": 0.10309278350515463,
41
+ "grad_norm": 6.507871627807617,
42
+ "learning_rate": 2e-05,
43
+ "loss": 0.6466,
44
+ "step": 10
45
+ },
46
+ {
47
+ "epoch": 0.12371134020618557,
48
+ "grad_norm": 0.6918491721153259,
49
+ "learning_rate": 2e-05,
50
+ "loss": 0.67,
51
+ "step": 12
52
+ },
53
+ {
54
+ "epoch": 0.14432989690721648,
55
+ "grad_norm": 0.8197646141052246,
56
+ "learning_rate": 2e-05,
57
+ "loss": 1.339,
58
+ "step": 14
59
+ },
60
+ {
61
+ "epoch": 0.16494845360824742,
62
+ "grad_norm": 1.4165186882019043,
63
+ "learning_rate": 2e-05,
64
+ "loss": 1.3645,
65
+ "step": 16
66
+ },
67
+ {
68
+ "epoch": 0.18556701030927836,
69
+ "grad_norm": 0.7580226063728333,
70
+ "learning_rate": 2e-05,
71
+ "loss": 0.6102,
72
+ "step": 18
73
+ },
74
+ {
75
+ "epoch": 0.20618556701030927,
76
+ "grad_norm": 1.047922134399414,
77
+ "learning_rate": 2e-05,
78
+ "loss": 0.6912,
79
+ "step": 20
80
+ },
81
+ {
82
+ "epoch": 0.2268041237113402,
83
+ "grad_norm": 1.2217419147491455,
84
+ "learning_rate": 2e-05,
85
+ "loss": 0.4418,
86
+ "step": 22
87
+ },
88
+ {
89
+ "epoch": 0.24742268041237114,
90
+ "grad_norm": 1.3620197772979736,
91
+ "learning_rate": 2e-05,
92
+ "loss": 0.5522,
93
+ "step": 24
94
+ },
95
+ {
96
+ "epoch": 0.26804123711340205,
97
+ "grad_norm": 3.1148924827575684,
98
+ "learning_rate": 2e-05,
99
+ "loss": 1.7596,
100
+ "step": 26
101
+ },
102
+ {
103
+ "epoch": 0.28865979381443296,
104
+ "grad_norm": 1.5309940576553345,
105
+ "learning_rate": 2e-05,
106
+ "loss": 1.2457,
107
+ "step": 28
108
+ },
109
+ {
110
+ "epoch": 0.30927835051546393,
111
+ "grad_norm": 3.7326714992523193,
112
+ "learning_rate": 2e-05,
113
+ "loss": 1.515,
114
+ "step": 30
115
+ },
116
+ {
117
+ "epoch": 0.32989690721649484,
118
+ "grad_norm": 0.7158209681510925,
119
+ "learning_rate": 2e-05,
120
+ "loss": 1.4406,
121
+ "step": 32
122
+ },
123
+ {
124
+ "epoch": 0.35051546391752575,
125
+ "grad_norm": 1.514033317565918,
126
+ "learning_rate": 2e-05,
127
+ "loss": 0.7915,
128
+ "step": 34
129
+ },
130
+ {
131
+ "epoch": 0.3711340206185567,
132
+ "grad_norm": 0.5789536833763123,
133
+ "learning_rate": 2e-05,
134
+ "loss": 0.8058,
135
+ "step": 36
136
+ },
137
+ {
138
+ "epoch": 0.3917525773195876,
139
+ "grad_norm": 1.2178142070770264,
140
+ "learning_rate": 2e-05,
141
+ "loss": 0.7588,
142
+ "step": 38
143
+ },
144
+ {
145
+ "epoch": 0.41237113402061853,
146
+ "grad_norm": 1.49998939037323,
147
+ "learning_rate": 2e-05,
148
+ "loss": 0.9498,
149
+ "step": 40
150
+ },
151
+ {
152
+ "epoch": 0.4329896907216495,
153
+ "grad_norm": 1.0307927131652832,
154
+ "learning_rate": 2e-05,
155
+ "loss": 1.6418,
156
+ "step": 42
157
+ },
158
+ {
159
+ "epoch": 0.4536082474226804,
160
+ "grad_norm": 6.7452473640441895,
161
+ "learning_rate": 2e-05,
162
+ "loss": 0.8119,
163
+ "step": 44
164
+ },
165
+ {
166
+ "epoch": 0.4742268041237113,
167
+ "grad_norm": 1.2180413007736206,
168
+ "learning_rate": 2e-05,
169
+ "loss": 1.0612,
170
+ "step": 46
171
+ },
172
+ {
173
+ "epoch": 0.4948453608247423,
174
+ "grad_norm": 2.678840398788452,
175
+ "learning_rate": 2e-05,
176
+ "loss": 1.2415,
177
+ "step": 48
178
+ },
179
+ {
180
+ "epoch": 0.5154639175257731,
181
+ "grad_norm": 2.5963003635406494,
182
+ "learning_rate": 2e-05,
183
+ "loss": 1.4499,
184
+ "step": 50
185
+ },
186
+ {
187
+ "epoch": 0.5360824742268041,
188
+ "grad_norm": 0.7345780730247498,
189
+ "learning_rate": 2e-05,
190
+ "loss": 0.1788,
191
+ "step": 52
192
+ },
193
+ {
194
+ "epoch": 0.5567010309278351,
195
+ "grad_norm": 1.1813082695007324,
196
+ "learning_rate": 2e-05,
197
+ "loss": 1.3107,
198
+ "step": 54
199
+ },
200
+ {
201
+ "epoch": 0.5773195876288659,
202
+ "grad_norm": 2.317573070526123,
203
+ "learning_rate": 2e-05,
204
+ "loss": 1.094,
205
+ "step": 56
206
+ },
207
+ {
208
+ "epoch": 0.5979381443298969,
209
+ "grad_norm": 1.814728021621704,
210
+ "learning_rate": 2e-05,
211
+ "loss": 0.8588,
212
+ "step": 58
213
+ },
214
+ {
215
+ "epoch": 0.6185567010309279,
216
+ "grad_norm": 1.8747090101242065,
217
+ "learning_rate": 2e-05,
218
+ "loss": 0.8991,
219
+ "step": 60
220
+ },
221
+ {
222
+ "epoch": 0.6391752577319587,
223
+ "grad_norm": 3.8781356811523438,
224
+ "learning_rate": 2e-05,
225
+ "loss": 1.2434,
226
+ "step": 62
227
+ },
228
+ {
229
+ "epoch": 0.6597938144329897,
230
+ "grad_norm": 2.7404439449310303,
231
+ "learning_rate": 2e-05,
232
+ "loss": 0.8968,
233
+ "step": 64
234
+ },
235
+ {
236
+ "epoch": 0.6804123711340206,
237
+ "grad_norm": 4.213019371032715,
238
+ "learning_rate": 2e-05,
239
+ "loss": 0.6253,
240
+ "step": 66
241
+ },
242
+ {
243
+ "epoch": 0.7010309278350515,
244
+ "grad_norm": 1.4956772327423096,
245
+ "learning_rate": 2e-05,
246
+ "loss": 0.7943,
247
+ "step": 68
248
+ },
249
+ {
250
+ "epoch": 0.7216494845360825,
251
+ "grad_norm": 1.6160740852355957,
252
+ "learning_rate": 2e-05,
253
+ "loss": 1.0117,
254
+ "step": 70
255
+ },
256
+ {
257
+ "epoch": 0.7422680412371134,
258
+ "grad_norm": 6.043447017669678,
259
+ "learning_rate": 2e-05,
260
+ "loss": 1.5506,
261
+ "step": 72
262
+ },
263
+ {
264
+ "epoch": 0.7628865979381443,
265
+ "grad_norm": 7.466903209686279,
266
+ "learning_rate": 2e-05,
267
+ "loss": 0.9948,
268
+ "step": 74
269
+ },
270
+ {
271
+ "epoch": 0.7835051546391752,
272
+ "grad_norm": 4.36099910736084,
273
+ "learning_rate": 2e-05,
274
+ "loss": 1.1756,
275
+ "step": 76
276
+ },
277
+ {
278
+ "epoch": 0.8041237113402062,
279
+ "grad_norm": 1.8531993627548218,
280
+ "learning_rate": 2e-05,
281
+ "loss": 1.0316,
282
+ "step": 78
283
+ },
284
+ {
285
+ "epoch": 0.8247422680412371,
286
+ "grad_norm": 4.479206562042236,
287
+ "learning_rate": 2e-05,
288
+ "loss": 1.6606,
289
+ "step": 80
290
+ },
291
+ {
292
+ "epoch": 0.845360824742268,
293
+ "grad_norm": 0.5770215392112732,
294
+ "learning_rate": 2e-05,
295
+ "loss": 1.5992,
296
+ "step": 82
297
+ },
298
+ {
299
+ "epoch": 0.865979381443299,
300
+ "grad_norm": 3.7939014434814453,
301
+ "learning_rate": 2e-05,
302
+ "loss": 2.2104,
303
+ "step": 84
304
+ },
305
+ {
306
+ "epoch": 0.8865979381443299,
307
+ "grad_norm": 2.1936018466949463,
308
+ "learning_rate": 2e-05,
309
+ "loss": 1.2776,
310
+ "step": 86
311
+ },
312
+ {
313
+ "epoch": 0.9072164948453608,
314
+ "grad_norm": 1.1358851194381714,
315
+ "learning_rate": 2e-05,
316
+ "loss": 1.9249,
317
+ "step": 88
318
+ },
319
+ {
320
+ "epoch": 0.9278350515463918,
321
+ "grad_norm": 1.4483697414398193,
322
+ "learning_rate": 2e-05,
323
+ "loss": 1.0092,
324
+ "step": 90
325
+ },
326
+ {
327
+ "epoch": 0.9484536082474226,
328
+ "grad_norm": 1.3848443031311035,
329
+ "learning_rate": 2e-05,
330
+ "loss": 1.3407,
331
+ "step": 92
332
+ },
333
+ {
334
+ "epoch": 0.9690721649484536,
335
+ "grad_norm": 1.5741782188415527,
336
+ "learning_rate": 2e-05,
337
+ "loss": 0.6802,
338
+ "step": 94
339
+ },
340
+ {
341
+ "epoch": 0.9896907216494846,
342
+ "grad_norm": 1.6624468564987183,
343
+ "learning_rate": 2e-05,
344
+ "loss": 0.7617,
345
+ "step": 96
346
+ },
347
+ {
348
+ "epoch": 1.0,
349
+ "step": 97,
350
+ "total_flos": 5653713448337408.0,
351
+ "train_loss": 1.1114492072272546,
352
+ "train_runtime": 220.1261,
353
+ "train_samples_per_second": 1.763,
354
+ "train_steps_per_second": 0.441
355
+ }
356
+ ],
357
+ "logging_steps": 2,
358
+ "max_steps": 97,
359
+ "num_input_tokens_seen": 0,
360
+ "num_train_epochs": 1,
361
+ "save_steps": 500,
362
+ "stateful_callbacks": {
363
+ "TrainerControl": {
364
+ "args": {
365
+ "should_epoch_stop": false,
366
+ "should_evaluate": false,
367
+ "should_log": false,
368
+ "should_save": false,
369
+ "should_training_stop": false
370
+ },
371
+ "attributes": {}
372
+ }
373
+ },
374
+ "total_flos": 5653713448337408.0,
375
+ "train_batch_size": 1,
376
+ "trial_name": null,
377
+ "trial_params": null
378
+ }
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b59fe809a5d2ff1ff756b1de152477ac8a80c025104a6cd66a90e49ca8f3f84
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92e2fe2a51dde3243caf5634353f0a09f4f71c5d629547f3763e42f5c912af3b
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:71d24055fd3128a43890e17810c0de225c352b7f033aa727b8036c70c99dd87b
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round17.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ce5d4aa7ade8f6d8109d98638e794d2b83b23105bde25a3d799c6b327f764fa2
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a3cf74f924c43ca4a7904a481d442221fa1aa818c3ba7f13b8bd40ce3e1198d9
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round20.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45f6bf0220bd0598ca2813c435122aa3518991b0f620c5b179add96fd94cee85
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:675895499139b95e75faabf47c9537891802ceb09552a240cc2c51ba48b48f0b
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:487b5d47689a8fff149b46f997bedcea9afd6eacde315b0561d4d694e82dca27
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_trainer_state.json ADDED
@@ -0,0 +1,378 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 97,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.020618556701030927,
13
+ "grad_norm": 2.1019058227539062,
14
+ "learning_rate": 2e-05,
15
+ "loss": 1.2384,
16
+ "step": 2
17
+ },
18
+ {
19
+ "epoch": 0.041237113402061855,
20
+ "grad_norm": 2.015745162963867,
21
+ "learning_rate": 2e-05,
22
+ "loss": 2.2301,
23
+ "step": 4
24
+ },
25
+ {
26
+ "epoch": 0.061855670103092786,
27
+ "grad_norm": 1.6932185888290405,
28
+ "learning_rate": 2e-05,
29
+ "loss": 1.332,
30
+ "step": 6
31
+ },
32
+ {
33
+ "epoch": 0.08247422680412371,
34
+ "grad_norm": 0.8953085541725159,
35
+ "learning_rate": 2e-05,
36
+ "loss": 0.7714,
37
+ "step": 8
38
+ },
39
+ {
40
+ "epoch": 0.10309278350515463,
41
+ "grad_norm": 0.3113327622413635,
42
+ "learning_rate": 2e-05,
43
+ "loss": 0.773,
44
+ "step": 10
45
+ },
46
+ {
47
+ "epoch": 0.12371134020618557,
48
+ "grad_norm": 1.1422592401504517,
49
+ "learning_rate": 2e-05,
50
+ "loss": 1.167,
51
+ "step": 12
52
+ },
53
+ {
54
+ "epoch": 0.14432989690721648,
55
+ "grad_norm": 1.0945158004760742,
56
+ "learning_rate": 2e-05,
57
+ "loss": 0.9648,
58
+ "step": 14
59
+ },
60
+ {
61
+ "epoch": 0.16494845360824742,
62
+ "grad_norm": 0.8330481052398682,
63
+ "learning_rate": 2e-05,
64
+ "loss": 1.245,
65
+ "step": 16
66
+ },
67
+ {
68
+ "epoch": 0.18556701030927836,
69
+ "grad_norm": 1.1668928861618042,
70
+ "learning_rate": 2e-05,
71
+ "loss": 1.0801,
72
+ "step": 18
73
+ },
74
+ {
75
+ "epoch": 0.20618556701030927,
76
+ "grad_norm": 1.4704902172088623,
77
+ "learning_rate": 2e-05,
78
+ "loss": 1.6117,
79
+ "step": 20
80
+ },
81
+ {
82
+ "epoch": 0.2268041237113402,
83
+ "grad_norm": 0.9273108839988708,
84
+ "learning_rate": 2e-05,
85
+ "loss": 1.1484,
86
+ "step": 22
87
+ },
88
+ {
89
+ "epoch": 0.24742268041237114,
90
+ "grad_norm": 0.47007447481155396,
91
+ "learning_rate": 2e-05,
92
+ "loss": 1.2086,
93
+ "step": 24
94
+ },
95
+ {
96
+ "epoch": 0.26804123711340205,
97
+ "grad_norm": 1.2891818284988403,
98
+ "learning_rate": 2e-05,
99
+ "loss": 0.8721,
100
+ "step": 26
101
+ },
102
+ {
103
+ "epoch": 0.28865979381443296,
104
+ "grad_norm": 2.4511640071868896,
105
+ "learning_rate": 2e-05,
106
+ "loss": 1.7582,
107
+ "step": 28
108
+ },
109
+ {
110
+ "epoch": 0.30927835051546393,
111
+ "grad_norm": 1.4482309818267822,
112
+ "learning_rate": 2e-05,
113
+ "loss": 1.1509,
114
+ "step": 30
115
+ },
116
+ {
117
+ "epoch": 0.32989690721649484,
118
+ "grad_norm": 1.1071757078170776,
119
+ "learning_rate": 2e-05,
120
+ "loss": 0.5992,
121
+ "step": 32
122
+ },
123
+ {
124
+ "epoch": 0.35051546391752575,
125
+ "grad_norm": 0.9718566536903381,
126
+ "learning_rate": 2e-05,
127
+ "loss": 0.6382,
128
+ "step": 34
129
+ },
130
+ {
131
+ "epoch": 0.3711340206185567,
132
+ "grad_norm": 1.8578362464904785,
133
+ "learning_rate": 2e-05,
134
+ "loss": 0.568,
135
+ "step": 36
136
+ },
137
+ {
138
+ "epoch": 0.3917525773195876,
139
+ "grad_norm": 0.9442087411880493,
140
+ "learning_rate": 2e-05,
141
+ "loss": 0.5858,
142
+ "step": 38
143
+ },
144
+ {
145
+ "epoch": 0.41237113402061853,
146
+ "grad_norm": 0.8823479413986206,
147
+ "learning_rate": 2e-05,
148
+ "loss": 0.756,
149
+ "step": 40
150
+ },
151
+ {
152
+ "epoch": 0.4329896907216495,
153
+ "grad_norm": 2.2437326908111572,
154
+ "learning_rate": 2e-05,
155
+ "loss": 0.8872,
156
+ "step": 42
157
+ },
158
+ {
159
+ "epoch": 0.4536082474226804,
160
+ "grad_norm": 1.061965823173523,
161
+ "learning_rate": 2e-05,
162
+ "loss": 1.8276,
163
+ "step": 44
164
+ },
165
+ {
166
+ "epoch": 0.4742268041237113,
167
+ "grad_norm": 0.6379388570785522,
168
+ "learning_rate": 2e-05,
169
+ "loss": 0.5883,
170
+ "step": 46
171
+ },
172
+ {
173
+ "epoch": 0.4948453608247423,
174
+ "grad_norm": 0.6430217027664185,
175
+ "learning_rate": 2e-05,
176
+ "loss": 0.8486,
177
+ "step": 48
178
+ },
179
+ {
180
+ "epoch": 0.5154639175257731,
181
+ "grad_norm": 1.2530596256256104,
182
+ "learning_rate": 2e-05,
183
+ "loss": 0.752,
184
+ "step": 50
185
+ },
186
+ {
187
+ "epoch": 0.5360824742268041,
188
+ "grad_norm": 1.1313493251800537,
189
+ "learning_rate": 2e-05,
190
+ "loss": 1.1919,
191
+ "step": 52
192
+ },
193
+ {
194
+ "epoch": 0.5567010309278351,
195
+ "grad_norm": 1.936036467552185,
196
+ "learning_rate": 2e-05,
197
+ "loss": 0.8134,
198
+ "step": 54
199
+ },
200
+ {
201
+ "epoch": 0.5773195876288659,
202
+ "grad_norm": 1.9745476245880127,
203
+ "learning_rate": 2e-05,
204
+ "loss": 0.9518,
205
+ "step": 56
206
+ },
207
+ {
208
+ "epoch": 0.5979381443298969,
209
+ "grad_norm": 3.271169900894165,
210
+ "learning_rate": 2e-05,
211
+ "loss": 1.5382,
212
+ "step": 58
213
+ },
214
+ {
215
+ "epoch": 0.6185567010309279,
216
+ "grad_norm": 0.9468516111373901,
217
+ "learning_rate": 2e-05,
218
+ "loss": 1.1189,
219
+ "step": 60
220
+ },
221
+ {
222
+ "epoch": 0.6391752577319587,
223
+ "grad_norm": 3.363159418106079,
224
+ "learning_rate": 2e-05,
225
+ "loss": 1.0864,
226
+ "step": 62
227
+ },
228
+ {
229
+ "epoch": 0.6597938144329897,
230
+ "grad_norm": 1.307508945465088,
231
+ "learning_rate": 2e-05,
232
+ "loss": 0.7174,
233
+ "step": 64
234
+ },
235
+ {
236
+ "epoch": 0.6804123711340206,
237
+ "grad_norm": 2.224381923675537,
238
+ "learning_rate": 2e-05,
239
+ "loss": 1.2527,
240
+ "step": 66
241
+ },
242
+ {
243
+ "epoch": 0.7010309278350515,
244
+ "grad_norm": 0.8224920630455017,
245
+ "learning_rate": 2e-05,
246
+ "loss": 0.981,
247
+ "step": 68
248
+ },
249
+ {
250
+ "epoch": 0.7216494845360825,
251
+ "grad_norm": 1.8318278789520264,
252
+ "learning_rate": 2e-05,
253
+ "loss": 1.1245,
254
+ "step": 70
255
+ },
256
+ {
257
+ "epoch": 0.7422680412371134,
258
+ "grad_norm": 1.883228063583374,
259
+ "learning_rate": 2e-05,
260
+ "loss": 0.8595,
261
+ "step": 72
262
+ },
263
+ {
264
+ "epoch": 0.7628865979381443,
265
+ "grad_norm": 1.8525029420852661,
266
+ "learning_rate": 2e-05,
267
+ "loss": 1.4871,
268
+ "step": 74
269
+ },
270
+ {
271
+ "epoch": 0.7835051546391752,
272
+ "grad_norm": 0.8241564035415649,
273
+ "learning_rate": 2e-05,
274
+ "loss": 0.6477,
275
+ "step": 76
276
+ },
277
+ {
278
+ "epoch": 0.8041237113402062,
279
+ "grad_norm": 1.532976508140564,
280
+ "learning_rate": 2e-05,
281
+ "loss": 1.1193,
282
+ "step": 78
283
+ },
284
+ {
285
+ "epoch": 0.8247422680412371,
286
+ "grad_norm": 0.6114745736122131,
287
+ "learning_rate": 2e-05,
288
+ "loss": 0.7454,
289
+ "step": 80
290
+ },
291
+ {
292
+ "epoch": 0.845360824742268,
293
+ "grad_norm": 1.4930524826049805,
294
+ "learning_rate": 2e-05,
295
+ "loss": 1.1418,
296
+ "step": 82
297
+ },
298
+ {
299
+ "epoch": 0.865979381443299,
300
+ "grad_norm": 1.7472268342971802,
301
+ "learning_rate": 2e-05,
302
+ "loss": 1.0005,
303
+ "step": 84
304
+ },
305
+ {
306
+ "epoch": 0.8865979381443299,
307
+ "grad_norm": 3.3397369384765625,
308
+ "learning_rate": 2e-05,
309
+ "loss": 1.4718,
310
+ "step": 86
311
+ },
312
+ {
313
+ "epoch": 0.9072164948453608,
314
+ "grad_norm": 1.0568790435791016,
315
+ "learning_rate": 2e-05,
316
+ "loss": 0.5434,
317
+ "step": 88
318
+ },
319
+ {
320
+ "epoch": 0.9278350515463918,
321
+ "grad_norm": 1.3748642206192017,
322
+ "learning_rate": 2e-05,
323
+ "loss": 0.8214,
324
+ "step": 90
325
+ },
326
+ {
327
+ "epoch": 0.9484536082474226,
328
+ "grad_norm": 0.7457247972488403,
329
+ "learning_rate": 2e-05,
330
+ "loss": 0.5015,
331
+ "step": 92
332
+ },
333
+ {
334
+ "epoch": 0.9690721649484536,
335
+ "grad_norm": 3.519141912460327,
336
+ "learning_rate": 2e-05,
337
+ "loss": 0.9263,
338
+ "step": 94
339
+ },
340
+ {
341
+ "epoch": 0.9896907216494846,
342
+ "grad_norm": 1.3801841735839844,
343
+ "learning_rate": 2e-05,
344
+ "loss": 1.0602,
345
+ "step": 96
346
+ },
347
+ {
348
+ "epoch": 1.0,
349
+ "step": 97,
350
+ "total_flos": 1.3522822525616128e+16,
351
+ "train_loss": 1.030662929888853,
352
+ "train_runtime": 308.3257,
353
+ "train_samples_per_second": 1.258,
354
+ "train_steps_per_second": 0.315
355
+ }
356
+ ],
357
+ "logging_steps": 2,
358
+ "max_steps": 97,
359
+ "num_input_tokens_seen": 0,
360
+ "num_train_epochs": 1,
361
+ "save_steps": 500,
362
+ "stateful_callbacks": {
363
+ "TrainerControl": {
364
+ "args": {
365
+ "should_epoch_stop": false,
366
+ "should_evaluate": false,
367
+ "should_log": false,
368
+ "should_save": false,
369
+ "should_training_stop": false
370
+ },
371
+ "attributes": {}
372
+ }
373
+ },
374
+ "total_flos": 1.3522822525616128e+16,
375
+ "train_batch_size": 1,
376
+ "trial_name": null,
377
+ "trial_params": null
378
+ }
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81ea4e99fe25c65d86ed9776cbac424de6b2b8d43fb47a1d8890dfec93b96961
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:09d2bdf2b9ce13a9ec50994d339bd81d63e6e52b28b259ece5e9cb8811fe1dfc
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e55e09f61f612220399d5a2849a488d4144eb0b10b3dbc3835dff0385d14acd1
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round17.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7449279a8525fe98a2816e50bbb277d0114d5562ad7ca0b1547a6dbb7ce341f
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d718525b9d18dda4393fe0da15a0470861d3ec12000b3aeccd8ce377394152db
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round20.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca1375ce5cfee303954ee1fc8b7d7f0aa4b80de93593dbf6f7bbf6cc85c5baef
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:860503692aa4b1db122fdc0b4f8c1d200d809123818bf78160f364bc91d0bca2
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e8a02edc5041d4c3b870135e29fbde11fa7c99eb2453e41149d91613f0d2e0c
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_trainer_state.json ADDED
@@ -0,0 +1,378 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 97,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.020618556701030927,
13
+ "grad_norm": 1.8023158311843872,
14
+ "learning_rate": 2e-05,
15
+ "loss": 0.4777,
16
+ "step": 2
17
+ },
18
+ {
19
+ "epoch": 0.041237113402061855,
20
+ "grad_norm": 2.333343267440796,
21
+ "learning_rate": 2e-05,
22
+ "loss": 0.6568,
23
+ "step": 4
24
+ },
25
+ {
26
+ "epoch": 0.061855670103092786,
27
+ "grad_norm": 2.633899211883545,
28
+ "learning_rate": 2e-05,
29
+ "loss": 2.8199,
30
+ "step": 6
31
+ },
32
+ {
33
+ "epoch": 0.08247422680412371,
34
+ "grad_norm": 1.4781279563903809,
35
+ "learning_rate": 2e-05,
36
+ "loss": 0.5231,
37
+ "step": 8
38
+ },
39
+ {
40
+ "epoch": 0.10309278350515463,
41
+ "grad_norm": 2.995490074157715,
42
+ "learning_rate": 2e-05,
43
+ "loss": 1.3005,
44
+ "step": 10
45
+ },
46
+ {
47
+ "epoch": 0.12371134020618557,
48
+ "grad_norm": 0.4523420035839081,
49
+ "learning_rate": 2e-05,
50
+ "loss": 0.4011,
51
+ "step": 12
52
+ },
53
+ {
54
+ "epoch": 0.14432989690721648,
55
+ "grad_norm": 2.69547438621521,
56
+ "learning_rate": 2e-05,
57
+ "loss": 1.2758,
58
+ "step": 14
59
+ },
60
+ {
61
+ "epoch": 0.16494845360824742,
62
+ "grad_norm": 2.170813798904419,
63
+ "learning_rate": 2e-05,
64
+ "loss": 0.9653,
65
+ "step": 16
66
+ },
67
+ {
68
+ "epoch": 0.18556701030927836,
69
+ "grad_norm": 1.8196375370025635,
70
+ "learning_rate": 2e-05,
71
+ "loss": 0.8301,
72
+ "step": 18
73
+ },
74
+ {
75
+ "epoch": 0.20618556701030927,
76
+ "grad_norm": 2.484487533569336,
77
+ "learning_rate": 2e-05,
78
+ "loss": 0.8427,
79
+ "step": 20
80
+ },
81
+ {
82
+ "epoch": 0.2268041237113402,
83
+ "grad_norm": 1.03840970993042,
84
+ "learning_rate": 2e-05,
85
+ "loss": 0.7525,
86
+ "step": 22
87
+ },
88
+ {
89
+ "epoch": 0.24742268041237114,
90
+ "grad_norm": 0.58939129114151,
91
+ "learning_rate": 2e-05,
92
+ "loss": 0.4422,
93
+ "step": 24
94
+ },
95
+ {
96
+ "epoch": 0.26804123711340205,
97
+ "grad_norm": 1.8537667989730835,
98
+ "learning_rate": 2e-05,
99
+ "loss": 0.5406,
100
+ "step": 26
101
+ },
102
+ {
103
+ "epoch": 0.28865979381443296,
104
+ "grad_norm": 1.9497562646865845,
105
+ "learning_rate": 2e-05,
106
+ "loss": 1.8268,
107
+ "step": 28
108
+ },
109
+ {
110
+ "epoch": 0.30927835051546393,
111
+ "grad_norm": 1.3029683828353882,
112
+ "learning_rate": 2e-05,
113
+ "loss": 0.4993,
114
+ "step": 30
115
+ },
116
+ {
117
+ "epoch": 0.32989690721649484,
118
+ "grad_norm": 1.48475182056427,
119
+ "learning_rate": 2e-05,
120
+ "loss": 0.3259,
121
+ "step": 32
122
+ },
123
+ {
124
+ "epoch": 0.35051546391752575,
125
+ "grad_norm": 2.472151041030884,
126
+ "learning_rate": 2e-05,
127
+ "loss": 0.9236,
128
+ "step": 34
129
+ },
130
+ {
131
+ "epoch": 0.3711340206185567,
132
+ "grad_norm": 0.365927517414093,
133
+ "learning_rate": 2e-05,
134
+ "loss": 0.5899,
135
+ "step": 36
136
+ },
137
+ {
138
+ "epoch": 0.3917525773195876,
139
+ "grad_norm": 0.7687128186225891,
140
+ "learning_rate": 2e-05,
141
+ "loss": 0.2881,
142
+ "step": 38
143
+ },
144
+ {
145
+ "epoch": 0.41237113402061853,
146
+ "grad_norm": 1.4524534940719604,
147
+ "learning_rate": 2e-05,
148
+ "loss": 0.7623,
149
+ "step": 40
150
+ },
151
+ {
152
+ "epoch": 0.4329896907216495,
153
+ "grad_norm": 8.207484245300293,
154
+ "learning_rate": 2e-05,
155
+ "loss": 2.418,
156
+ "step": 42
157
+ },
158
+ {
159
+ "epoch": 0.4536082474226804,
160
+ "grad_norm": 1.92764151096344,
161
+ "learning_rate": 2e-05,
162
+ "loss": 0.5673,
163
+ "step": 44
164
+ },
165
+ {
166
+ "epoch": 0.4742268041237113,
167
+ "grad_norm": 2.470285177230835,
168
+ "learning_rate": 2e-05,
169
+ "loss": 1.9934,
170
+ "step": 46
171
+ },
172
+ {
173
+ "epoch": 0.4948453608247423,
174
+ "grad_norm": 1.807496190071106,
175
+ "learning_rate": 2e-05,
176
+ "loss": 1.6274,
177
+ "step": 48
178
+ },
179
+ {
180
+ "epoch": 0.5154639175257731,
181
+ "grad_norm": 1.729468822479248,
182
+ "learning_rate": 2e-05,
183
+ "loss": 1.0121,
184
+ "step": 50
185
+ },
186
+ {
187
+ "epoch": 0.5360824742268041,
188
+ "grad_norm": 0.3935474455356598,
189
+ "learning_rate": 2e-05,
190
+ "loss": 0.3824,
191
+ "step": 52
192
+ },
193
+ {
194
+ "epoch": 0.5567010309278351,
195
+ "grad_norm": 2.567225456237793,
196
+ "learning_rate": 2e-05,
197
+ "loss": 1.352,
198
+ "step": 54
199
+ },
200
+ {
201
+ "epoch": 0.5773195876288659,
202
+ "grad_norm": 3.6223816871643066,
203
+ "learning_rate": 2e-05,
204
+ "loss": 1.4641,
205
+ "step": 56
206
+ },
207
+ {
208
+ "epoch": 0.5979381443298969,
209
+ "grad_norm": 1.3244529962539673,
210
+ "learning_rate": 2e-05,
211
+ "loss": 0.6464,
212
+ "step": 58
213
+ },
214
+ {
215
+ "epoch": 0.6185567010309279,
216
+ "grad_norm": 3.0408148765563965,
217
+ "learning_rate": 2e-05,
218
+ "loss": 0.7199,
219
+ "step": 60
220
+ },
221
+ {
222
+ "epoch": 0.6391752577319587,
223
+ "grad_norm": 1.9795098304748535,
224
+ "learning_rate": 2e-05,
225
+ "loss": 0.7565,
226
+ "step": 62
227
+ },
228
+ {
229
+ "epoch": 0.6597938144329897,
230
+ "grad_norm": 2.4508280754089355,
231
+ "learning_rate": 2e-05,
232
+ "loss": 0.9399,
233
+ "step": 64
234
+ },
235
+ {
236
+ "epoch": 0.6804123711340206,
237
+ "grad_norm": 0.9279292821884155,
238
+ "learning_rate": 2e-05,
239
+ "loss": 0.4619,
240
+ "step": 66
241
+ },
242
+ {
243
+ "epoch": 0.7010309278350515,
244
+ "grad_norm": 0.538191020488739,
245
+ "learning_rate": 2e-05,
246
+ "loss": 0.2551,
247
+ "step": 68
248
+ },
249
+ {
250
+ "epoch": 0.7216494845360825,
251
+ "grad_norm": 5.2244110107421875,
252
+ "learning_rate": 2e-05,
253
+ "loss": 2.5304,
254
+ "step": 70
255
+ },
256
+ {
257
+ "epoch": 0.7422680412371134,
258
+ "grad_norm": 2.26310133934021,
259
+ "learning_rate": 2e-05,
260
+ "loss": 0.9515,
261
+ "step": 72
262
+ },
263
+ {
264
+ "epoch": 0.7628865979381443,
265
+ "grad_norm": 0.4313948452472687,
266
+ "learning_rate": 2e-05,
267
+ "loss": 0.2802,
268
+ "step": 74
269
+ },
270
+ {
271
+ "epoch": 0.7835051546391752,
272
+ "grad_norm": 5.081795692443848,
273
+ "learning_rate": 2e-05,
274
+ "loss": 1.6002,
275
+ "step": 76
276
+ },
277
+ {
278
+ "epoch": 0.8041237113402062,
279
+ "grad_norm": 2.1677563190460205,
280
+ "learning_rate": 2e-05,
281
+ "loss": 0.9818,
282
+ "step": 78
283
+ },
284
+ {
285
+ "epoch": 0.8247422680412371,
286
+ "grad_norm": 1.9904687404632568,
287
+ "learning_rate": 2e-05,
288
+ "loss": 0.8751,
289
+ "step": 80
290
+ },
291
+ {
292
+ "epoch": 0.845360824742268,
293
+ "grad_norm": 3.359741449356079,
294
+ "learning_rate": 2e-05,
295
+ "loss": 0.5831,
296
+ "step": 82
297
+ },
298
+ {
299
+ "epoch": 0.865979381443299,
300
+ "grad_norm": 1.1926054954528809,
301
+ "learning_rate": 2e-05,
302
+ "loss": 0.3873,
303
+ "step": 84
304
+ },
305
+ {
306
+ "epoch": 0.8865979381443299,
307
+ "grad_norm": 1.2176306247711182,
308
+ "learning_rate": 2e-05,
309
+ "loss": 0.6256,
310
+ "step": 86
311
+ },
312
+ {
313
+ "epoch": 0.9072164948453608,
314
+ "grad_norm": 0.7016535997390747,
315
+ "learning_rate": 2e-05,
316
+ "loss": 0.3552,
317
+ "step": 88
318
+ },
319
+ {
320
+ "epoch": 0.9278350515463918,
321
+ "grad_norm": 1.8117469549179077,
322
+ "learning_rate": 2e-05,
323
+ "loss": 0.5001,
324
+ "step": 90
325
+ },
326
+ {
327
+ "epoch": 0.9484536082474226,
328
+ "grad_norm": 2.748814344406128,
329
+ "learning_rate": 2e-05,
330
+ "loss": 1.1114,
331
+ "step": 92
332
+ },
333
+ {
334
+ "epoch": 0.9690721649484536,
335
+ "grad_norm": 3.6248812675476074,
336
+ "learning_rate": 2e-05,
337
+ "loss": 0.6721,
338
+ "step": 94
339
+ },
340
+ {
341
+ "epoch": 0.9896907216494846,
342
+ "grad_norm": 2.769014596939087,
343
+ "learning_rate": 2e-05,
344
+ "loss": 1.4068,
345
+ "step": 96
346
+ },
347
+ {
348
+ "epoch": 1.0,
349
+ "step": 97,
350
+ "total_flos": 1.5329114357497856e+16,
351
+ "train_loss": 0.9247780927677745,
352
+ "train_runtime": 349.7946,
353
+ "train_samples_per_second": 1.109,
354
+ "train_steps_per_second": 0.277
355
+ }
356
+ ],
357
+ "logging_steps": 2,
358
+ "max_steps": 97,
359
+ "num_input_tokens_seen": 0,
360
+ "num_train_epochs": 1,
361
+ "save_steps": 500,
362
+ "stateful_callbacks": {
363
+ "TrainerControl": {
364
+ "args": {
365
+ "should_epoch_stop": false,
366
+ "should_evaluate": false,
367
+ "should_log": false,
368
+ "should_save": false,
369
+ "should_training_stop": false
370
+ },
371
+ "attributes": {}
372
+ }
373
+ },
374
+ "total_flos": 1.5329114357497856e+16,
375
+ "train_batch_size": 1,
376
+ "trial_name": null,
377
+ "trial_params": null
378
+ }
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2908b60c5d4eb0083024ffb2cd3047034dc685fccc27b5981c9d1e2d47aef02a
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bfce8cc1ef5a648c9351912e1c734c810036f4b0ab22e7f29043378864a65a34
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7acc8875ab4ca1ba366ff14db348ec5ef1210605c6e2c6fa18d8373713fd41c3
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round17.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c5178b170d8fd01fc7cb4b267584a74ca5d41ac38c953a1797fd290732f0c9c
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3c04a8967061f0ed5178762ca311b1378846ab4fa8e22bb2c5515a724a697978
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round20.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b85fce0437f13e9e428cf1b401be1d1869888082ce08cab06bb31b7f79fccc0
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d795de3e40192cc7193d575e906b2b05cd1b099b05fe9038b9401294b3ff2363
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e5a490cf63cb307af1430eea838fe180370af281d611d944337af1e61f08155
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_trainer_state.json ADDED
@@ -0,0 +1,378 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 97,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.020618556701030927,
13
+ "grad_norm": 2.5633797645568848,
14
+ "learning_rate": 2e-05,
15
+ "loss": 1.974,
16
+ "step": 2
17
+ },
18
+ {
19
+ "epoch": 0.041237113402061855,
20
+ "grad_norm": 5.461665153503418,
21
+ "learning_rate": 2e-05,
22
+ "loss": 3.5071,
23
+ "step": 4
24
+ },
25
+ {
26
+ "epoch": 0.061855670103092786,
27
+ "grad_norm": 0.6179507374763489,
28
+ "learning_rate": 2e-05,
29
+ "loss": 0.6818,
30
+ "step": 6
31
+ },
32
+ {
33
+ "epoch": 0.08247422680412371,
34
+ "grad_norm": 0.5985448360443115,
35
+ "learning_rate": 2e-05,
36
+ "loss": 0.8984,
37
+ "step": 8
38
+ },
39
+ {
40
+ "epoch": 0.10309278350515463,
41
+ "grad_norm": 1.3306835889816284,
42
+ "learning_rate": 2e-05,
43
+ "loss": 1.4385,
44
+ "step": 10
45
+ },
46
+ {
47
+ "epoch": 0.12371134020618557,
48
+ "grad_norm": 0.5487744808197021,
49
+ "learning_rate": 2e-05,
50
+ "loss": 1.1519,
51
+ "step": 12
52
+ },
53
+ {
54
+ "epoch": 0.14432989690721648,
55
+ "grad_norm": 1.5741511583328247,
56
+ "learning_rate": 2e-05,
57
+ "loss": 0.6216,
58
+ "step": 14
59
+ },
60
+ {
61
+ "epoch": 0.16494845360824742,
62
+ "grad_norm": 0.6786820292472839,
63
+ "learning_rate": 2e-05,
64
+ "loss": 0.7161,
65
+ "step": 16
66
+ },
67
+ {
68
+ "epoch": 0.18556701030927836,
69
+ "grad_norm": 0.9131273627281189,
70
+ "learning_rate": 2e-05,
71
+ "loss": 1.1003,
72
+ "step": 18
73
+ },
74
+ {
75
+ "epoch": 0.20618556701030927,
76
+ "grad_norm": 1.3463377952575684,
77
+ "learning_rate": 2e-05,
78
+ "loss": 1.71,
79
+ "step": 20
80
+ },
81
+ {
82
+ "epoch": 0.2268041237113402,
83
+ "grad_norm": 0.5329614281654358,
84
+ "learning_rate": 2e-05,
85
+ "loss": 0.6123,
86
+ "step": 22
87
+ },
88
+ {
89
+ "epoch": 0.24742268041237114,
90
+ "grad_norm": 0.7505204081535339,
91
+ "learning_rate": 2e-05,
92
+ "loss": 0.9683,
93
+ "step": 24
94
+ },
95
+ {
96
+ "epoch": 0.26804123711340205,
97
+ "grad_norm": 0.7533180713653564,
98
+ "learning_rate": 2e-05,
99
+ "loss": 1.1445,
100
+ "step": 26
101
+ },
102
+ {
103
+ "epoch": 0.28865979381443296,
104
+ "grad_norm": 0.6992803812026978,
105
+ "learning_rate": 2e-05,
106
+ "loss": 1.1177,
107
+ "step": 28
108
+ },
109
+ {
110
+ "epoch": 0.30927835051546393,
111
+ "grad_norm": 0.6891196966171265,
112
+ "learning_rate": 2e-05,
113
+ "loss": 1.3975,
114
+ "step": 30
115
+ },
116
+ {
117
+ "epoch": 0.32989690721649484,
118
+ "grad_norm": 1.2143183946609497,
119
+ "learning_rate": 2e-05,
120
+ "loss": 1.1021,
121
+ "step": 32
122
+ },
123
+ {
124
+ "epoch": 0.35051546391752575,
125
+ "grad_norm": 0.840191662311554,
126
+ "learning_rate": 2e-05,
127
+ "loss": 0.8009,
128
+ "step": 34
129
+ },
130
+ {
131
+ "epoch": 0.3711340206185567,
132
+ "grad_norm": 0.6156558394432068,
133
+ "learning_rate": 2e-05,
134
+ "loss": 1.3974,
135
+ "step": 36
136
+ },
137
+ {
138
+ "epoch": 0.3917525773195876,
139
+ "grad_norm": 0.5059778690338135,
140
+ "learning_rate": 2e-05,
141
+ "loss": 1.2592,
142
+ "step": 38
143
+ },
144
+ {
145
+ "epoch": 0.41237113402061853,
146
+ "grad_norm": 1.269295573234558,
147
+ "learning_rate": 2e-05,
148
+ "loss": 0.9996,
149
+ "step": 40
150
+ },
151
+ {
152
+ "epoch": 0.4329896907216495,
153
+ "grad_norm": 0.716225266456604,
154
+ "learning_rate": 2e-05,
155
+ "loss": 0.8199,
156
+ "step": 42
157
+ },
158
+ {
159
+ "epoch": 0.4536082474226804,
160
+ "grad_norm": 0.531692624092102,
161
+ "learning_rate": 2e-05,
162
+ "loss": 0.5912,
163
+ "step": 44
164
+ },
165
+ {
166
+ "epoch": 0.4742268041237113,
167
+ "grad_norm": 0.715979814529419,
168
+ "learning_rate": 2e-05,
169
+ "loss": 0.9426,
170
+ "step": 46
171
+ },
172
+ {
173
+ "epoch": 0.4948453608247423,
174
+ "grad_norm": 1.2602847814559937,
175
+ "learning_rate": 2e-05,
176
+ "loss": 1.4139,
177
+ "step": 48
178
+ },
179
+ {
180
+ "epoch": 0.5154639175257731,
181
+ "grad_norm": 1.5276769399642944,
182
+ "learning_rate": 2e-05,
183
+ "loss": 1.2681,
184
+ "step": 50
185
+ },
186
+ {
187
+ "epoch": 0.5360824742268041,
188
+ "grad_norm": 1.8456441164016724,
189
+ "learning_rate": 2e-05,
190
+ "loss": 0.6275,
191
+ "step": 52
192
+ },
193
+ {
194
+ "epoch": 0.5567010309278351,
195
+ "grad_norm": 0.8769627213478088,
196
+ "learning_rate": 2e-05,
197
+ "loss": 1.0259,
198
+ "step": 54
199
+ },
200
+ {
201
+ "epoch": 0.5773195876288659,
202
+ "grad_norm": 3.3462865352630615,
203
+ "learning_rate": 2e-05,
204
+ "loss": 1.479,
205
+ "step": 56
206
+ },
207
+ {
208
+ "epoch": 0.5979381443298969,
209
+ "grad_norm": 1.0609265565872192,
210
+ "learning_rate": 2e-05,
211
+ "loss": 0.6238,
212
+ "step": 58
213
+ },
214
+ {
215
+ "epoch": 0.6185567010309279,
216
+ "grad_norm": 1.0079237222671509,
217
+ "learning_rate": 2e-05,
218
+ "loss": 0.6956,
219
+ "step": 60
220
+ },
221
+ {
222
+ "epoch": 0.6391752577319587,
223
+ "grad_norm": 4.337386131286621,
224
+ "learning_rate": 2e-05,
225
+ "loss": 1.3864,
226
+ "step": 62
227
+ },
228
+ {
229
+ "epoch": 0.6597938144329897,
230
+ "grad_norm": 1.1737914085388184,
231
+ "learning_rate": 2e-05,
232
+ "loss": 1.0942,
233
+ "step": 64
234
+ },
235
+ {
236
+ "epoch": 0.6804123711340206,
237
+ "grad_norm": 0.5829662680625916,
238
+ "learning_rate": 2e-05,
239
+ "loss": 0.8114,
240
+ "step": 66
241
+ },
242
+ {
243
+ "epoch": 0.7010309278350515,
244
+ "grad_norm": 0.25017768144607544,
245
+ "learning_rate": 2e-05,
246
+ "loss": 0.4132,
247
+ "step": 68
248
+ },
249
+ {
250
+ "epoch": 0.7216494845360825,
251
+ "grad_norm": 1.2530516386032104,
252
+ "learning_rate": 2e-05,
253
+ "loss": 0.5202,
254
+ "step": 70
255
+ },
256
+ {
257
+ "epoch": 0.7422680412371134,
258
+ "grad_norm": 0.6008771061897278,
259
+ "learning_rate": 2e-05,
260
+ "loss": 0.6505,
261
+ "step": 72
262
+ },
263
+ {
264
+ "epoch": 0.7628865979381443,
265
+ "grad_norm": 0.48716089129447937,
266
+ "learning_rate": 2e-05,
267
+ "loss": 0.7202,
268
+ "step": 74
269
+ },
270
+ {
271
+ "epoch": 0.7835051546391752,
272
+ "grad_norm": 1.642950177192688,
273
+ "learning_rate": 2e-05,
274
+ "loss": 1.134,
275
+ "step": 76
276
+ },
277
+ {
278
+ "epoch": 0.8041237113402062,
279
+ "grad_norm": 0.7713057398796082,
280
+ "learning_rate": 2e-05,
281
+ "loss": 0.5877,
282
+ "step": 78
283
+ },
284
+ {
285
+ "epoch": 0.8247422680412371,
286
+ "grad_norm": 1.294180989265442,
287
+ "learning_rate": 2e-05,
288
+ "loss": 0.5226,
289
+ "step": 80
290
+ },
291
+ {
292
+ "epoch": 0.845360824742268,
293
+ "grad_norm": 0.2003328651189804,
294
+ "learning_rate": 2e-05,
295
+ "loss": 0.2094,
296
+ "step": 82
297
+ },
298
+ {
299
+ "epoch": 0.865979381443299,
300
+ "grad_norm": 0.6338465213775635,
301
+ "learning_rate": 2e-05,
302
+ "loss": 0.7058,
303
+ "step": 84
304
+ },
305
+ {
306
+ "epoch": 0.8865979381443299,
307
+ "grad_norm": 1.3183602094650269,
308
+ "learning_rate": 2e-05,
309
+ "loss": 1.2619,
310
+ "step": 86
311
+ },
312
+ {
313
+ "epoch": 0.9072164948453608,
314
+ "grad_norm": 0.9359031915664673,
315
+ "learning_rate": 2e-05,
316
+ "loss": 1.1756,
317
+ "step": 88
318
+ },
319
+ {
320
+ "epoch": 0.9278350515463918,
321
+ "grad_norm": 0.5005592703819275,
322
+ "learning_rate": 2e-05,
323
+ "loss": 1.1162,
324
+ "step": 90
325
+ },
326
+ {
327
+ "epoch": 0.9484536082474226,
328
+ "grad_norm": 0.2807972729206085,
329
+ "learning_rate": 2e-05,
330
+ "loss": 0.583,
331
+ "step": 92
332
+ },
333
+ {
334
+ "epoch": 0.9690721649484536,
335
+ "grad_norm": 0.6926721334457397,
336
+ "learning_rate": 2e-05,
337
+ "loss": 0.9605,
338
+ "step": 94
339
+ },
340
+ {
341
+ "epoch": 0.9896907216494846,
342
+ "grad_norm": 0.21759887039661407,
343
+ "learning_rate": 2e-05,
344
+ "loss": 0.6858,
345
+ "step": 96
346
+ },
347
+ {
348
+ "epoch": 1.0,
349
+ "step": 97,
350
+ "total_flos": 2.080121124238131e+16,
351
+ "train_loss": 1.007467240402379,
352
+ "train_runtime": 338.0776,
353
+ "train_samples_per_second": 1.148,
354
+ "train_steps_per_second": 0.287
355
+ }
356
+ ],
357
+ "logging_steps": 2,
358
+ "max_steps": 97,
359
+ "num_input_tokens_seen": 0,
360
+ "num_train_epochs": 1,
361
+ "save_steps": 500,
362
+ "stateful_callbacks": {
363
+ "TrainerControl": {
364
+ "args": {
365
+ "should_epoch_stop": false,
366
+ "should_evaluate": false,
367
+ "should_log": false,
368
+ "should_save": false,
369
+ "should_training_stop": false
370
+ },
371
+ "attributes": {}
372
+ }
373
+ },
374
+ "total_flos": 2.080121124238131e+16,
375
+ "train_batch_size": 1,
376
+ "trial_name": null,
377
+ "trial_params": null
378
+ }
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bf03d29093163ceaed9ea2e19a65e3718afe1320f63b74047aff22ed1c0d77b5
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f315cc312f35c52fe96b624f2bb9eb4a87784a50eedd3170e48370e79c732c99
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6cc2975d15c47cf62e7a2e666e8637057e26ed262e313f81de63e5aab42a794c
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round17.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dcb23591f1a1572cf31f81cb79aecf34b5f970362e548e6e105db0dae4b1d108
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:27536f011217cab70e402a60c194f6e187e639b9d1ae42b8ba72eaefc3d3c920
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round20.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:27a301d978e2097010eff626f239c57da0c0c2fa356a5a3587b6dad294411382
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3618bf6e8af32458bda5e7a158280e6583de6c0126cef75cf75f85b0b11827c3
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc11c279615d6faf756aa45ae052b1b2fdc8af408d6b8d7d0edd12c617fb0dc3
3
+ size 791576546
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_trainer_state.json ADDED
@@ -0,0 +1,378 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 97,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.020618556701030927,
13
+ "grad_norm": 0.8590981364250183,
14
+ "learning_rate": 2e-05,
15
+ "loss": 0.3685,
16
+ "step": 2
17
+ },
18
+ {
19
+ "epoch": 0.041237113402061855,
20
+ "grad_norm": 1.5185532569885254,
21
+ "learning_rate": 2e-05,
22
+ "loss": 0.9951,
23
+ "step": 4
24
+ },
25
+ {
26
+ "epoch": 0.061855670103092786,
27
+ "grad_norm": 1.0182133913040161,
28
+ "learning_rate": 2e-05,
29
+ "loss": 0.6803,
30
+ "step": 6
31
+ },
32
+ {
33
+ "epoch": 0.08247422680412371,
34
+ "grad_norm": 4.001636981964111,
35
+ "learning_rate": 2e-05,
36
+ "loss": 1.0399,
37
+ "step": 8
38
+ },
39
+ {
40
+ "epoch": 0.10309278350515463,
41
+ "grad_norm": 4.477555751800537,
42
+ "learning_rate": 2e-05,
43
+ "loss": 3.0105,
44
+ "step": 10
45
+ },
46
+ {
47
+ "epoch": 0.12371134020618557,
48
+ "grad_norm": 3.720902442932129,
49
+ "learning_rate": 2e-05,
50
+ "loss": 0.913,
51
+ "step": 12
52
+ },
53
+ {
54
+ "epoch": 0.14432989690721648,
55
+ "grad_norm": 0.19767700135707855,
56
+ "learning_rate": 2e-05,
57
+ "loss": 0.3786,
58
+ "step": 14
59
+ },
60
+ {
61
+ "epoch": 0.16494845360824742,
62
+ "grad_norm": 0.6825243234634399,
63
+ "learning_rate": 2e-05,
64
+ "loss": 1.5835,
65
+ "step": 16
66
+ },
67
+ {
68
+ "epoch": 0.18556701030927836,
69
+ "grad_norm": 0.9178080558776855,
70
+ "learning_rate": 2e-05,
71
+ "loss": 0.9607,
72
+ "step": 18
73
+ },
74
+ {
75
+ "epoch": 0.20618556701030927,
76
+ "grad_norm": 1.5082707405090332,
77
+ "learning_rate": 2e-05,
78
+ "loss": 0.9881,
79
+ "step": 20
80
+ },
81
+ {
82
+ "epoch": 0.2268041237113402,
83
+ "grad_norm": 0.8916629552841187,
84
+ "learning_rate": 2e-05,
85
+ "loss": 1.0688,
86
+ "step": 22
87
+ },
88
+ {
89
+ "epoch": 0.24742268041237114,
90
+ "grad_norm": 0.5061885714530945,
91
+ "learning_rate": 2e-05,
92
+ "loss": 1.2571,
93
+ "step": 24
94
+ },
95
+ {
96
+ "epoch": 0.26804123711340205,
97
+ "grad_norm": 1.7212241888046265,
98
+ "learning_rate": 2e-05,
99
+ "loss": 1.1221,
100
+ "step": 26
101
+ },
102
+ {
103
+ "epoch": 0.28865979381443296,
104
+ "grad_norm": 2.3262715339660645,
105
+ "learning_rate": 2e-05,
106
+ "loss": 1.0792,
107
+ "step": 28
108
+ },
109
+ {
110
+ "epoch": 0.30927835051546393,
111
+ "grad_norm": 1.4995611906051636,
112
+ "learning_rate": 2e-05,
113
+ "loss": 1.4011,
114
+ "step": 30
115
+ },
116
+ {
117
+ "epoch": 0.32989690721649484,
118
+ "grad_norm": 1.2664096355438232,
119
+ "learning_rate": 2e-05,
120
+ "loss": 0.9637,
121
+ "step": 32
122
+ },
123
+ {
124
+ "epoch": 0.35051546391752575,
125
+ "grad_norm": 1.5707759857177734,
126
+ "learning_rate": 2e-05,
127
+ "loss": 1.0174,
128
+ "step": 34
129
+ },
130
+ {
131
+ "epoch": 0.3711340206185567,
132
+ "grad_norm": 0.5960234999656677,
133
+ "learning_rate": 2e-05,
134
+ "loss": 0.9389,
135
+ "step": 36
136
+ },
137
+ {
138
+ "epoch": 0.3917525773195876,
139
+ "grad_norm": 5.197012901306152,
140
+ "learning_rate": 2e-05,
141
+ "loss": 1.4119,
142
+ "step": 38
143
+ },
144
+ {
145
+ "epoch": 0.41237113402061853,
146
+ "grad_norm": 1.908044695854187,
147
+ "learning_rate": 2e-05,
148
+ "loss": 1.1758,
149
+ "step": 40
150
+ },
151
+ {
152
+ "epoch": 0.4329896907216495,
153
+ "grad_norm": 2.055213212966919,
154
+ "learning_rate": 2e-05,
155
+ "loss": 0.771,
156
+ "step": 42
157
+ },
158
+ {
159
+ "epoch": 0.4536082474226804,
160
+ "grad_norm": 1.1356596946716309,
161
+ "learning_rate": 2e-05,
162
+ "loss": 1.7888,
163
+ "step": 44
164
+ },
165
+ {
166
+ "epoch": 0.4742268041237113,
167
+ "grad_norm": 1.3487610816955566,
168
+ "learning_rate": 2e-05,
169
+ "loss": 0.914,
170
+ "step": 46
171
+ },
172
+ {
173
+ "epoch": 0.4948453608247423,
174
+ "grad_norm": 2.6447250843048096,
175
+ "learning_rate": 2e-05,
176
+ "loss": 0.6373,
177
+ "step": 48
178
+ },
179
+ {
180
+ "epoch": 0.5154639175257731,
181
+ "grad_norm": 3.1843416690826416,
182
+ "learning_rate": 2e-05,
183
+ "loss": 1.2856,
184
+ "step": 50
185
+ },
186
+ {
187
+ "epoch": 0.5360824742268041,
188
+ "grad_norm": 1.0026551485061646,
189
+ "learning_rate": 2e-05,
190
+ "loss": 1.2857,
191
+ "step": 52
192
+ },
193
+ {
194
+ "epoch": 0.5567010309278351,
195
+ "grad_norm": 2.135878562927246,
196
+ "learning_rate": 2e-05,
197
+ "loss": 1.0765,
198
+ "step": 54
199
+ },
200
+ {
201
+ "epoch": 0.5773195876288659,
202
+ "grad_norm": 1.3891205787658691,
203
+ "learning_rate": 2e-05,
204
+ "loss": 1.1711,
205
+ "step": 56
206
+ },
207
+ {
208
+ "epoch": 0.5979381443298969,
209
+ "grad_norm": 2.050652503967285,
210
+ "learning_rate": 2e-05,
211
+ "loss": 1.5952,
212
+ "step": 58
213
+ },
214
+ {
215
+ "epoch": 0.6185567010309279,
216
+ "grad_norm": 2.9438633918762207,
217
+ "learning_rate": 2e-05,
218
+ "loss": 1.0818,
219
+ "step": 60
220
+ },
221
+ {
222
+ "epoch": 0.6391752577319587,
223
+ "grad_norm": 1.6134741306304932,
224
+ "learning_rate": 2e-05,
225
+ "loss": 0.6641,
226
+ "step": 62
227
+ },
228
+ {
229
+ "epoch": 0.6597938144329897,
230
+ "grad_norm": 2.2244954109191895,
231
+ "learning_rate": 2e-05,
232
+ "loss": 0.7769,
233
+ "step": 64
234
+ },
235
+ {
236
+ "epoch": 0.6804123711340206,
237
+ "grad_norm": 1.8583745956420898,
238
+ "learning_rate": 2e-05,
239
+ "loss": 1.1453,
240
+ "step": 66
241
+ },
242
+ {
243
+ "epoch": 0.7010309278350515,
244
+ "grad_norm": 0.707295835018158,
245
+ "learning_rate": 2e-05,
246
+ "loss": 0.6789,
247
+ "step": 68
248
+ },
249
+ {
250
+ "epoch": 0.7216494845360825,
251
+ "grad_norm": 0.23287317156791687,
252
+ "learning_rate": 2e-05,
253
+ "loss": 1.3918,
254
+ "step": 70
255
+ },
256
+ {
257
+ "epoch": 0.7422680412371134,
258
+ "grad_norm": 1.0222351551055908,
259
+ "learning_rate": 2e-05,
260
+ "loss": 0.9926,
261
+ "step": 72
262
+ },
263
+ {
264
+ "epoch": 0.7628865979381443,
265
+ "grad_norm": 2.1297173500061035,
266
+ "learning_rate": 2e-05,
267
+ "loss": 0.9963,
268
+ "step": 74
269
+ },
270
+ {
271
+ "epoch": 0.7835051546391752,
272
+ "grad_norm": 1.9865819215774536,
273
+ "learning_rate": 2e-05,
274
+ "loss": 0.8631,
275
+ "step": 76
276
+ },
277
+ {
278
+ "epoch": 0.8041237113402062,
279
+ "grad_norm": 2.2636280059814453,
280
+ "learning_rate": 2e-05,
281
+ "loss": 0.4846,
282
+ "step": 78
283
+ },
284
+ {
285
+ "epoch": 0.8247422680412371,
286
+ "grad_norm": 1.4322012662887573,
287
+ "learning_rate": 2e-05,
288
+ "loss": 1.0235,
289
+ "step": 80
290
+ },
291
+ {
292
+ "epoch": 0.845360824742268,
293
+ "grad_norm": 2.809218168258667,
294
+ "learning_rate": 2e-05,
295
+ "loss": 0.9339,
296
+ "step": 82
297
+ },
298
+ {
299
+ "epoch": 0.865979381443299,
300
+ "grad_norm": 1.8168107271194458,
301
+ "learning_rate": 2e-05,
302
+ "loss": 0.8049,
303
+ "step": 84
304
+ },
305
+ {
306
+ "epoch": 0.8865979381443299,
307
+ "grad_norm": 1.5112167596817017,
308
+ "learning_rate": 2e-05,
309
+ "loss": 1.373,
310
+ "step": 86
311
+ },
312
+ {
313
+ "epoch": 0.9072164948453608,
314
+ "grad_norm": 4.819437026977539,
315
+ "learning_rate": 2e-05,
316
+ "loss": 1.7935,
317
+ "step": 88
318
+ },
319
+ {
320
+ "epoch": 0.9278350515463918,
321
+ "grad_norm": 0.9341912269592285,
322
+ "learning_rate": 2e-05,
323
+ "loss": 0.7278,
324
+ "step": 90
325
+ },
326
+ {
327
+ "epoch": 0.9484536082474226,
328
+ "grad_norm": 4.472631931304932,
329
+ "learning_rate": 2e-05,
330
+ "loss": 1.4348,
331
+ "step": 92
332
+ },
333
+ {
334
+ "epoch": 0.9690721649484536,
335
+ "grad_norm": 1.809728980064392,
336
+ "learning_rate": 2e-05,
337
+ "loss": 1.6456,
338
+ "step": 94
339
+ },
340
+ {
341
+ "epoch": 0.9896907216494846,
342
+ "grad_norm": 1.6377031803131104,
343
+ "learning_rate": 2e-05,
344
+ "loss": 1.7333,
345
+ "step": 96
346
+ },
347
+ {
348
+ "epoch": 1.0,
349
+ "step": 97,
350
+ "total_flos": 1.975033480085504e+16,
351
+ "train_loss": 1.1055578644742672,
352
+ "train_runtime": 337.8402,
353
+ "train_samples_per_second": 1.148,
354
+ "train_steps_per_second": 0.287
355
+ }
356
+ ],
357
+ "logging_steps": 2,
358
+ "max_steps": 97,
359
+ "num_input_tokens_seen": 0,
360
+ "num_train_epochs": 1,
361
+ "save_steps": 500,
362
+ "stateful_callbacks": {
363
+ "TrainerControl": {
364
+ "args": {
365
+ "should_epoch_stop": false,
366
+ "should_evaluate": false,
367
+ "should_log": false,
368
+ "should_save": false,
369
+ "should_training_stop": false
370
+ },
371
+ "attributes": {}
372
+ }
373
+ },
374
+ "total_flos": 1.975033480085504e+16,
375
+ "train_batch_size": 1,
376
+ "trial_name": null,
377
+ "trial_params": null
378
+ }
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46eab86d8d1e3f5fc1ac8c1e8f6309d58d813f964dc37d68978bffe21b9f7e81
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:968762ad11492a1af1d36913033f4ed757c1850c3d1655201f762cfd84f91dee
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:974f00e411dcaf11620ee11b363d5b25cb18f8d0e43123d8beffd72fdb36e89c
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round17.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:508bfb2ce05dfa871b7ae9a0f74bb3eaab798217eccb5058f0e70f7b19d7e199
3
+ size 791578182
client_states_feddualMulti2pqfullfreeze_back_T05_freq10_NewAorthensure_Borth_bs4_saveoptim_lr2e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8cefac5bc3f5091fd0dbb56e8a1a0e41e969d98e5271e8ae2ac6e52e0f23ecde
3
+ size 791576546