Hugging Face's logo

Matt300209
/
1B_code_only_1e4

Safetensors
llama
Model card Files
xet
Community
Matt300209 commited on
Commit
5d55318
·
verified ·
1 Parent(s): ce0ea66

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 +31 -0
  2. checkpoint-1143/optimizer_0/.metadata +3 -0
  3. checkpoint-1143/optimizer_0/__0_0.distcp +3 -0
  4. checkpoint-1143/optimizer_0/__1_0.distcp +3 -0
  5. checkpoint-1143/optimizer_0/__2_0.distcp +3 -0
  6. checkpoint-1143/optimizer_0/__3_0.distcp +3 -0
  7. checkpoint-1143/pytorch_model_fsdp_0/.metadata +3 -0
  8. checkpoint-1143/pytorch_model_fsdp_0/__0_0.distcp +3 -0
  9. checkpoint-1143/pytorch_model_fsdp_0/__1_0.distcp +3 -0
  10. checkpoint-1143/pytorch_model_fsdp_0/__2_0.distcp +3 -0
  11. checkpoint-1143/pytorch_model_fsdp_0/__3_0.distcp +3 -0
  12. checkpoint-1143/rng_state_0.pth +3 -0
  13. checkpoint-1143/rng_state_1.pth +3 -0
  14. checkpoint-1143/rng_state_2.pth +3 -0
  15. checkpoint-1143/rng_state_3.pth +3 -0
  16. checkpoint-1143/scheduler.pt +3 -0
  17. checkpoint-1143/trainer_state.json +846 -0
  18. checkpoint-2286/optimizer_0/.metadata +3 -0
  19. checkpoint-2286/optimizer_0/__0_0.distcp +3 -0
  20. checkpoint-2286/optimizer_0/__1_0.distcp +3 -0
  21. checkpoint-2286/optimizer_0/__2_0.distcp +3 -0
  22. checkpoint-2286/optimizer_0/__3_0.distcp +3 -0
  23. checkpoint-2286/pytorch_model_fsdp_0/.metadata +3 -0
  24. checkpoint-2286/pytorch_model_fsdp_0/__0_0.distcp +3 -0
  25. checkpoint-2286/pytorch_model_fsdp_0/__1_0.distcp +3 -0
  26. checkpoint-2286/pytorch_model_fsdp_0/__2_0.distcp +3 -0
  27. checkpoint-2286/pytorch_model_fsdp_0/__3_0.distcp +3 -0
  28. checkpoint-2286/rng_state_0.pth +3 -0
  29. checkpoint-2286/rng_state_1.pth +3 -0
  30. checkpoint-2286/rng_state_2.pth +3 -0
  31. checkpoint-2286/rng_state_3.pth +3 -0
  32. checkpoint-2286/scheduler.pt +3 -0
  33. checkpoint-2286/trainer_state.json +1644 -0
  34. checkpoint-3426/optimizer_0/.metadata +3 -0
  35. checkpoint-3426/optimizer_0/__0_0.distcp +3 -0
  36. checkpoint-3426/optimizer_0/__1_0.distcp +3 -0
  37. checkpoint-3426/optimizer_0/__2_0.distcp +3 -0
  38. checkpoint-3426/optimizer_0/__3_0.distcp +3 -0
  39. checkpoint-3426/pytorch_model_fsdp_0/.metadata +3 -0
  40. checkpoint-3426/pytorch_model_fsdp_0/__0_0.distcp +3 -0
  41. checkpoint-3426/pytorch_model_fsdp_0/__1_0.distcp +3 -0
  42. checkpoint-3426/pytorch_model_fsdp_0/__2_0.distcp +3 -0
  43. checkpoint-3426/pytorch_model_fsdp_0/__3_0.distcp +3 -0
  44. checkpoint-3426/rng_state_0.pth +3 -0
  45. checkpoint-3426/rng_state_1.pth +3 -0
  46. checkpoint-3426/rng_state_2.pth +3 -0
  47. checkpoint-3426/rng_state_3.pth +3 -0
  48. checkpoint-3426/scheduler.pt +3 -0
  49. checkpoint-3426/trainer_state.json +2442 -0
  50. config.json +36 -0
.gitattributes CHANGED
@@ -33,3 +33,34 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ checkpoint-1143/optimizer_0/.metadata filter=lfs diff=lfs merge=lfs -text
37
+ checkpoint-1143/optimizer_0/__0_0.distcp filter=lfs diff=lfs merge=lfs -text
38
+ checkpoint-1143/optimizer_0/__1_0.distcp filter=lfs diff=lfs merge=lfs -text
39
+ checkpoint-1143/optimizer_0/__2_0.distcp filter=lfs diff=lfs merge=lfs -text
40
+ checkpoint-1143/optimizer_0/__3_0.distcp filter=lfs diff=lfs merge=lfs -text
41
+ checkpoint-1143/pytorch_model_fsdp_0/.metadata filter=lfs diff=lfs merge=lfs -text
42
+ checkpoint-1143/pytorch_model_fsdp_0/__0_0.distcp filter=lfs diff=lfs merge=lfs -text
43
+ checkpoint-1143/pytorch_model_fsdp_0/__1_0.distcp filter=lfs diff=lfs merge=lfs -text
44
+ checkpoint-1143/pytorch_model_fsdp_0/__2_0.distcp filter=lfs diff=lfs merge=lfs -text
45
+ checkpoint-1143/pytorch_model_fsdp_0/__3_0.distcp filter=lfs diff=lfs merge=lfs -text
46
+ checkpoint-2286/optimizer_0/.metadata filter=lfs diff=lfs merge=lfs -text
47
+ checkpoint-2286/optimizer_0/__0_0.distcp filter=lfs diff=lfs merge=lfs -text
48
+ checkpoint-2286/optimizer_0/__1_0.distcp filter=lfs diff=lfs merge=lfs -text
49
+ checkpoint-2286/optimizer_0/__2_0.distcp filter=lfs diff=lfs merge=lfs -text
50
+ checkpoint-2286/optimizer_0/__3_0.distcp filter=lfs diff=lfs merge=lfs -text
51
+ checkpoint-2286/pytorch_model_fsdp_0/.metadata filter=lfs diff=lfs merge=lfs -text
52
+ checkpoint-2286/pytorch_model_fsdp_0/__0_0.distcp filter=lfs diff=lfs merge=lfs -text
53
+ checkpoint-2286/pytorch_model_fsdp_0/__1_0.distcp filter=lfs diff=lfs merge=lfs -text
54
+ checkpoint-2286/pytorch_model_fsdp_0/__2_0.distcp filter=lfs diff=lfs merge=lfs -text
55
+ checkpoint-2286/pytorch_model_fsdp_0/__3_0.distcp filter=lfs diff=lfs merge=lfs -text
56
+ checkpoint-3426/optimizer_0/.metadata filter=lfs diff=lfs merge=lfs -text
57
+ checkpoint-3426/optimizer_0/__0_0.distcp filter=lfs diff=lfs merge=lfs -text
58
+ checkpoint-3426/optimizer_0/__1_0.distcp filter=lfs diff=lfs merge=lfs -text
59
+ checkpoint-3426/optimizer_0/__2_0.distcp filter=lfs diff=lfs merge=lfs -text
60
+ checkpoint-3426/optimizer_0/__3_0.distcp filter=lfs diff=lfs merge=lfs -text
61
+ checkpoint-3426/pytorch_model_fsdp_0/.metadata filter=lfs diff=lfs merge=lfs -text
62
+ checkpoint-3426/pytorch_model_fsdp_0/__0_0.distcp filter=lfs diff=lfs merge=lfs -text
63
+ checkpoint-3426/pytorch_model_fsdp_0/__1_0.distcp filter=lfs diff=lfs merge=lfs -text
64
+ checkpoint-3426/pytorch_model_fsdp_0/__2_0.distcp filter=lfs diff=lfs merge=lfs -text
65
+ checkpoint-3426/pytorch_model_fsdp_0/__3_0.distcp filter=lfs diff=lfs merge=lfs -text
66
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
checkpoint-1143/optimizer_0/.metadata ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1dd9e6d4beebe1690c8d99358768b1076a1ef1cc8b20ffcd62703a3eedb0a8c1
3
+ size 342948
checkpoint-1143/optimizer_0/__0_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dac5d3d1efa4b6d99d1205acb964c22ea995f4bd6308ffbd12a6a9a61a033edd
3
+ size 2472090861
checkpoint-1143/optimizer_0/__1_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad08430914c0679e24ea6913676ee83efb7824038c5bbea284c334cfed3228da
3
+ size 2472174123
checkpoint-1143/optimizer_0/__2_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43bde2bc55ebaca643a45b848a1f65bd0167fadda186065db992b6c9ced3b1d9
3
+ size 2472182518
checkpoint-1143/optimizer_0/__3_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:125861a679a795fe13b3c02eea22db849c3a207a98f7be2da110e34a20261fe2
3
+ size 2472179190
checkpoint-1143/pytorch_model_fsdp_0/.metadata ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:242ac3ba3946c42f208f825806645989b468fa09b96c8ce90963492896751f1f
3
+ size 129152
checkpoint-1143/pytorch_model_fsdp_0/__0_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2ba534ac3db914d6a6f3843e9623a0d7cf4cdd4717a3ef9f7cdf435477c8f7fa
3
+ size 1498714507
checkpoint-1143/pytorch_model_fsdp_0/__1_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c10517f1ee98d272b691d5be2ab0caf5f706dfd979bba9550f5d5eb3010cc53f
3
+ size 1498714507
checkpoint-1143/pytorch_model_fsdp_0/__2_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5757f685250c771031aa71f39e5e57af65de3a3584f16d428d6703c19a45f2f
3
+ size 1498714507
checkpoint-1143/pytorch_model_fsdp_0/__3_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1fa13d888ab3434a3a556da28c7400c57cdec90df0696570dc11c2f67fb68fe9
3
+ size 1498714507
checkpoint-1143/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4188048c882e5b595f5e2cce07e8c73ea45afa11bc95f5627f2d6191ccaa5e40
3
+ size 15429
checkpoint-1143/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0eb1f7b1acfd920c8cdef076eee48ccfec7d3ed4d8c5d83c2592fbbb4a4f9b38
3
+ size 15429
checkpoint-1143/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5664fd5079c69dc8edcc429ded884f07d45c771f27a5a65fe1ad751348f9e1de
3
+ size 15429
checkpoint-1143/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a3f17c4e252e5b7e99f3573c3bca992070f9ef436dd578ee9e88c333a94cef0
3
+ size 15429
checkpoint-1143/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa4344f06305e3a8793199a02dc90908e6c605511a625eb80252b479a7a39796
3
+ size 1465
checkpoint-1143/trainer_state.json ADDED
@@ -0,0 +1,846 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 1143,
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,
13
+ "eval_loss": 2.1641297340393066,
14
+ "eval_runtime": 4.2967,
15
+ "eval_samples_per_second": 4.655,
16
+ "eval_steps_per_second": 0.465,
17
+ "step": 0
18
+ },
19
+ {
20
+ "epoch": 0.0008750820389411508,
21
+ "grad_norm": 6.37307071685791,
22
+ "learning_rate": 2.915451895043732e-07,
23
+ "loss": 1.6434,
24
+ "step": 1
25
+ },
26
+ {
27
+ "epoch": 0.008750820389411508,
28
+ "grad_norm": 4.0545759201049805,
29
+ "learning_rate": 2.915451895043732e-06,
30
+ "loss": 1.6965,
31
+ "step": 10
32
+ },
33
+ {
34
+ "epoch": 0.017501640778823015,
35
+ "grad_norm": 2.363191604614258,
36
+ "learning_rate": 5.830903790087464e-06,
37
+ "loss": 1.5272,
38
+ "step": 20
39
+ },
40
+ {
41
+ "epoch": 0.02625246116823452,
42
+ "grad_norm": 1.6348848342895508,
43
+ "learning_rate": 8.746355685131195e-06,
44
+ "loss": 1.389,
45
+ "step": 30
46
+ },
47
+ {
48
+ "epoch": 0.03500328155764603,
49
+ "grad_norm": 1.6716902256011963,
50
+ "learning_rate": 1.1661807580174927e-05,
51
+ "loss": 1.348,
52
+ "step": 40
53
+ },
54
+ {
55
+ "epoch": 0.04375410194705753,
56
+ "grad_norm": 1.510228157043457,
57
+ "learning_rate": 1.457725947521866e-05,
58
+ "loss": 1.2997,
59
+ "step": 50
60
+ },
61
+ {
62
+ "epoch": 0.05250492233646904,
63
+ "grad_norm": 1.6153247356414795,
64
+ "learning_rate": 1.749271137026239e-05,
65
+ "loss": 1.2665,
66
+ "step": 60
67
+ },
68
+ {
69
+ "epoch": 0.06125574272588055,
70
+ "grad_norm": 1.446387767791748,
71
+ "learning_rate": 2.0408163265306123e-05,
72
+ "loss": 1.237,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.07000656311529206,
77
+ "grad_norm": 1.3039863109588623,
78
+ "learning_rate": 2.3323615160349855e-05,
79
+ "loss": 1.2341,
80
+ "step": 80
81
+ },
82
+ {
83
+ "epoch": 0.07875738350470357,
84
+ "grad_norm": 1.4866383075714111,
85
+ "learning_rate": 2.6239067055393584e-05,
86
+ "loss": 1.2338,
87
+ "step": 90
88
+ },
89
+ {
90
+ "epoch": 0.08750820389411507,
91
+ "grad_norm": 1.358859658241272,
92
+ "learning_rate": 2.915451895043732e-05,
93
+ "loss": 1.2331,
94
+ "step": 100
95
+ },
96
+ {
97
+ "epoch": 0.09625902428352658,
98
+ "grad_norm": 1.847506046295166,
99
+ "learning_rate": 3.206997084548105e-05,
100
+ "loss": 1.1628,
101
+ "step": 110
102
+ },
103
+ {
104
+ "epoch": 0.10500984467293809,
105
+ "grad_norm": 1.7742356061935425,
106
+ "learning_rate": 3.498542274052478e-05,
107
+ "loss": 1.1846,
108
+ "step": 120
109
+ },
110
+ {
111
+ "epoch": 0.1137606650623496,
112
+ "grad_norm": 1.3415714502334595,
113
+ "learning_rate": 3.790087463556852e-05,
114
+ "loss": 1.1699,
115
+ "step": 130
116
+ },
117
+ {
118
+ "epoch": 0.1225114854517611,
119
+ "grad_norm": 1.5319396257400513,
120
+ "learning_rate": 4.0816326530612245e-05,
121
+ "loss": 1.181,
122
+ "step": 140
123
+ },
124
+ {
125
+ "epoch": 0.1312623058411726,
126
+ "grad_norm": 1.3915642499923706,
127
+ "learning_rate": 4.373177842565598e-05,
128
+ "loss": 1.189,
129
+ "step": 150
130
+ },
131
+ {
132
+ "epoch": 0.14001312623058412,
133
+ "grad_norm": 1.175917625427246,
134
+ "learning_rate": 4.664723032069971e-05,
135
+ "loss": 1.1605,
136
+ "step": 160
137
+ },
138
+ {
139
+ "epoch": 0.14876394661999562,
140
+ "grad_norm": 1.3125519752502441,
141
+ "learning_rate": 4.956268221574344e-05,
142
+ "loss": 1.1734,
143
+ "step": 170
144
+ },
145
+ {
146
+ "epoch": 0.15751476700940714,
147
+ "grad_norm": 1.2966923713684082,
148
+ "learning_rate": 5.247813411078717e-05,
149
+ "loss": 1.2075,
150
+ "step": 180
151
+ },
152
+ {
153
+ "epoch": 0.16626558739881864,
154
+ "grad_norm": 1.2630099058151245,
155
+ "learning_rate": 5.539358600583091e-05,
156
+ "loss": 1.1791,
157
+ "step": 190
158
+ },
159
+ {
160
+ "epoch": 0.17501640778823013,
161
+ "grad_norm": 1.306665301322937,
162
+ "learning_rate": 5.830903790087464e-05,
163
+ "loss": 1.1469,
164
+ "step": 200
165
+ },
166
+ {
167
+ "epoch": 0.18376722817764166,
168
+ "grad_norm": 1.19467031955719,
169
+ "learning_rate": 6.122448979591838e-05,
170
+ "loss": 1.185,
171
+ "step": 210
172
+ },
173
+ {
174
+ "epoch": 0.19251804856705315,
175
+ "grad_norm": 1.416387677192688,
176
+ "learning_rate": 6.41399416909621e-05,
177
+ "loss": 1.2061,
178
+ "step": 220
179
+ },
180
+ {
181
+ "epoch": 0.20126886895646467,
182
+ "grad_norm": 1.3546264171600342,
183
+ "learning_rate": 6.705539358600583e-05,
184
+ "loss": 1.2016,
185
+ "step": 230
186
+ },
187
+ {
188
+ "epoch": 0.21001968934587617,
189
+ "grad_norm": 1.2417850494384766,
190
+ "learning_rate": 6.997084548104956e-05,
191
+ "loss": 1.1797,
192
+ "step": 240
193
+ },
194
+ {
195
+ "epoch": 0.2187705097352877,
196
+ "grad_norm": 1.2703570127487183,
197
+ "learning_rate": 7.28862973760933e-05,
198
+ "loss": 1.2059,
199
+ "step": 250
200
+ },
201
+ {
202
+ "epoch": 0.2275213301246992,
203
+ "grad_norm": 1.2479429244995117,
204
+ "learning_rate": 7.580174927113704e-05,
205
+ "loss": 1.1958,
206
+ "step": 260
207
+ },
208
+ {
209
+ "epoch": 0.23627215051411068,
210
+ "grad_norm": 1.507601261138916,
211
+ "learning_rate": 7.871720116618077e-05,
212
+ "loss": 1.2151,
213
+ "step": 270
214
+ },
215
+ {
216
+ "epoch": 0.2450229709035222,
217
+ "grad_norm": 1.2083854675292969,
218
+ "learning_rate": 8.163265306122449e-05,
219
+ "loss": 1.2202,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2537737912929337,
224
+ "grad_norm": 1.2939566373825073,
225
+ "learning_rate": 8.454810495626822e-05,
226
+ "loss": 1.1916,
227
+ "step": 290
228
+ },
229
+ {
230
+ "epoch": 0.2625246116823452,
231
+ "grad_norm": 1.2871382236480713,
232
+ "learning_rate": 8.746355685131195e-05,
233
+ "loss": 1.222,
234
+ "step": 300
235
+ },
236
+ {
237
+ "epoch": 0.27127543207175675,
238
+ "grad_norm": 1.1558256149291992,
239
+ "learning_rate": 9.037900874635569e-05,
240
+ "loss": 1.2094,
241
+ "step": 310
242
+ },
243
+ {
244
+ "epoch": 0.28002625246116825,
245
+ "grad_norm": 1.0638883113861084,
246
+ "learning_rate": 9.329446064139942e-05,
247
+ "loss": 1.2,
248
+ "step": 320
249
+ },
250
+ {
251
+ "epoch": 0.28877707285057974,
252
+ "grad_norm": 1.1072458028793335,
253
+ "learning_rate": 9.620991253644317e-05,
254
+ "loss": 1.224,
255
+ "step": 330
256
+ },
257
+ {
258
+ "epoch": 0.29752789323999124,
259
+ "grad_norm": 1.1332110166549683,
260
+ "learning_rate": 9.912536443148688e-05,
261
+ "loss": 1.2266,
262
+ "step": 340
263
+ },
264
+ {
265
+ "epoch": 0.30627871362940273,
266
+ "grad_norm": 1.0537033081054688,
267
+ "learning_rate": 9.999872800049233e-05,
268
+ "loss": 1.2173,
269
+ "step": 350
270
+ },
271
+ {
272
+ "epoch": 0.3150295340188143,
273
+ "grad_norm": 1.090928316116333,
274
+ "learning_rate": 9.999249795462235e-05,
275
+ "loss": 1.2416,
276
+ "step": 360
277
+ },
278
+ {
279
+ "epoch": 0.3237803544082258,
280
+ "grad_norm": 1.0084835290908813,
281
+ "learning_rate": 9.998107687592326e-05,
282
+ "loss": 1.2094,
283
+ "step": 370
284
+ },
285
+ {
286
+ "epoch": 0.3325311747976373,
287
+ "grad_norm": 1.82576584815979,
288
+ "learning_rate": 9.996446595031691e-05,
289
+ "loss": 1.2246,
290
+ "step": 380
291
+ },
292
+ {
293
+ "epoch": 0.34128199518704877,
294
+ "grad_norm": 1.2305560111999512,
295
+ "learning_rate": 9.994266690261928e-05,
296
+ "loss": 1.2647,
297
+ "step": 390
298
+ },
299
+ {
300
+ "epoch": 0.35003281557646027,
301
+ "grad_norm": 1.170067548751831,
302
+ "learning_rate": 9.991568199636148e-05,
303
+ "loss": 1.1829,
304
+ "step": 400
305
+ },
306
+ {
307
+ "epoch": 0.3587836359658718,
308
+ "grad_norm": 1.0299819707870483,
309
+ "learning_rate": 9.988351403355454e-05,
310
+ "loss": 1.2471,
311
+ "step": 410
312
+ },
313
+ {
314
+ "epoch": 0.3675344563552833,
315
+ "grad_norm": 0.9742593765258789,
316
+ "learning_rate": 9.98461663543987e-05,
317
+ "loss": 1.2049,
318
+ "step": 420
319
+ },
320
+ {
321
+ "epoch": 0.3762852767446948,
322
+ "grad_norm": 0.9377045631408691,
323
+ "learning_rate": 9.980364283693638e-05,
324
+ "loss": 1.1963,
325
+ "step": 430
326
+ },
327
+ {
328
+ "epoch": 0.3850360971341063,
329
+ "grad_norm": 1.0425899028778076,
330
+ "learning_rate": 9.97559478966496e-05,
331
+ "loss": 1.2167,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.39378691752351785,
336
+ "grad_norm": 0.9749513864517212,
337
+ "learning_rate": 9.97030864860014e-05,
338
+ "loss": 1.2166,
339
+ "step": 450
340
+ },
341
+ {
342
+ "epoch": 0.40253773791292935,
343
+ "grad_norm": 1.0552780628204346,
344
+ "learning_rate": 9.964506409392176e-05,
345
+ "loss": 1.2228,
346
+ "step": 460
347
+ },
348
+ {
349
+ "epoch": 0.41128855830234085,
350
+ "grad_norm": 1.0261808633804321,
351
+ "learning_rate": 9.958188674523748e-05,
352
+ "loss": 1.2252,
353
+ "step": 470
354
+ },
355
+ {
356
+ "epoch": 0.42003937869175234,
357
+ "grad_norm": 0.9530945420265198,
358
+ "learning_rate": 9.951356100004665e-05,
359
+ "loss": 1.1845,
360
+ "step": 480
361
+ },
362
+ {
363
+ "epoch": 0.42879019908116384,
364
+ "grad_norm": 0.8990481495857239,
365
+ "learning_rate": 9.94400939530375e-05,
366
+ "loss": 1.2283,
367
+ "step": 490
368
+ },
369
+ {
370
+ "epoch": 0.4375410194705754,
371
+ "grad_norm": 0.9713592529296875,
372
+ "learning_rate": 9.936149323275169e-05,
373
+ "loss": 1.1945,
374
+ "step": 500
375
+ },
376
+ {
377
+ "epoch": 0.4462918398599869,
378
+ "grad_norm": 1.003854513168335,
379
+ "learning_rate": 9.927776700079216e-05,
380
+ "loss": 1.2322,
381
+ "step": 510
382
+ },
383
+ {
384
+ "epoch": 0.4550426602493984,
385
+ "grad_norm": 0.935400664806366,
386
+ "learning_rate": 9.918892395097572e-05,
387
+ "loss": 1.2272,
388
+ "step": 520
389
+ },
390
+ {
391
+ "epoch": 0.4637934806388099,
392
+ "grad_norm": 1.075519323348999,
393
+ "learning_rate": 9.90949733084303e-05,
394
+ "loss": 1.2069,
395
+ "step": 530
396
+ },
397
+ {
398
+ "epoch": 0.47254430102822137,
399
+ "grad_norm": 0.9949712157249451,
400
+ "learning_rate": 9.8995924828637e-05,
401
+ "loss": 1.2083,
402
+ "step": 540
403
+ },
404
+ {
405
+ "epoch": 0.4812951214176329,
406
+ "grad_norm": 0.8942195773124695,
407
+ "learning_rate": 9.889178879641717e-05,
408
+ "loss": 1.2137,
409
+ "step": 550
410
+ },
411
+ {
412
+ "epoch": 0.4900459418070444,
413
+ "grad_norm": 0.9525908827781677,
414
+ "learning_rate": 9.878257602486445e-05,
415
+ "loss": 1.2348,
416
+ "step": 560
417
+ },
418
+ {
419
+ "epoch": 0.4987967621964559,
420
+ "grad_norm": 0.895167350769043,
421
+ "learning_rate": 9.8668297854222e-05,
422
+ "loss": 1.2034,
423
+ "step": 570
424
+ },
425
+ {
426
+ "epoch": 0.5075475825858674,
427
+ "grad_norm": 1.0244312286376953,
428
+ "learning_rate": 9.854896615070495e-05,
429
+ "loss": 1.1942,
430
+ "step": 580
431
+ },
432
+ {
433
+ "epoch": 0.516298402975279,
434
+ "grad_norm": 0.9835163950920105,
435
+ "learning_rate": 9.842459330526826e-05,
436
+ "loss": 1.2182,
437
+ "step": 590
438
+ },
439
+ {
440
+ "epoch": 0.5250492233646904,
441
+ "grad_norm": 0.9238565564155579,
442
+ "learning_rate": 9.829519223232006e-05,
443
+ "loss": 1.1992,
444
+ "step": 600
445
+ },
446
+ {
447
+ "epoch": 0.533800043754102,
448
+ "grad_norm": 0.9063233733177185,
449
+ "learning_rate": 9.81607763683808e-05,
450
+ "loss": 1.2135,
451
+ "step": 610
452
+ },
453
+ {
454
+ "epoch": 0.5425508641435135,
455
+ "grad_norm": 0.9810764193534851,
456
+ "learning_rate": 9.802135967068782e-05,
457
+ "loss": 1.1909,
458
+ "step": 620
459
+ },
460
+ {
461
+ "epoch": 0.5513016845329249,
462
+ "grad_norm": 0.9388274550437927,
463
+ "learning_rate": 9.787695661574632e-05,
464
+ "loss": 1.1693,
465
+ "step": 630
466
+ },
467
+ {
468
+ "epoch": 0.5600525049223365,
469
+ "grad_norm": 0.933283805847168,
470
+ "learning_rate": 9.772758219782601e-05,
471
+ "loss": 1.1857,
472
+ "step": 640
473
+ },
474
+ {
475
+ "epoch": 0.5688033253117479,
476
+ "grad_norm": 0.947989284992218,
477
+ "learning_rate": 9.757325192740425e-05,
478
+ "loss": 1.1901,
479
+ "step": 650
480
+ },
481
+ {
482
+ "epoch": 0.5775541457011595,
483
+ "grad_norm": 0.8272686004638672,
484
+ "learning_rate": 9.741398182955548e-05,
485
+ "loss": 1.2087,
486
+ "step": 660
487
+ },
488
+ {
489
+ "epoch": 0.586304966090571,
490
+ "grad_norm": 0.8907813429832458,
491
+ "learning_rate": 9.724978844228717e-05,
492
+ "loss": 1.201,
493
+ "step": 670
494
+ },
495
+ {
496
+ "epoch": 0.5950557864799825,
497
+ "grad_norm": 0.8374966382980347,
498
+ "learning_rate": 9.708068881482271e-05,
499
+ "loss": 1.2028,
500
+ "step": 680
501
+ },
502
+ {
503
+ "epoch": 0.603806606869394,
504
+ "grad_norm": 0.9283761382102966,
505
+ "learning_rate": 9.690670050583095e-05,
506
+ "loss": 1.1995,
507
+ "step": 690
508
+ },
509
+ {
510
+ "epoch": 0.6125574272588055,
511
+ "grad_norm": 0.8573452234268188,
512
+ "learning_rate": 9.672784158160307e-05,
513
+ "loss": 1.1372,
514
+ "step": 700
515
+ },
516
+ {
517
+ "epoch": 0.621308247648217,
518
+ "grad_norm": 0.8831506371498108,
519
+ "learning_rate": 9.654413061417655e-05,
520
+ "loss": 1.1626,
521
+ "step": 710
522
+ },
523
+ {
524
+ "epoch": 0.6300590680376286,
525
+ "grad_norm": 0.9180812239646912,
526
+ "learning_rate": 9.635558667940683e-05,
527
+ "loss": 1.1466,
528
+ "step": 720
529
+ },
530
+ {
531
+ "epoch": 0.63880988842704,
532
+ "grad_norm": 0.8657775521278381,
533
+ "learning_rate": 9.616222935498646e-05,
534
+ "loss": 1.2052,
535
+ "step": 730
536
+ },
537
+ {
538
+ "epoch": 0.6475607088164516,
539
+ "grad_norm": 0.9488115310668945,
540
+ "learning_rate": 9.596407871841229e-05,
541
+ "loss": 1.1651,
542
+ "step": 740
543
+ },
544
+ {
545
+ "epoch": 0.656311529205863,
546
+ "grad_norm": 0.8592752814292908,
547
+ "learning_rate": 9.57611553449006e-05,
548
+ "loss": 1.1953,
549
+ "step": 750
550
+ },
551
+ {
552
+ "epoch": 0.6650623495952745,
553
+ "grad_norm": 0.889445960521698,
554
+ "learning_rate": 9.555348030525076e-05,
555
+ "loss": 1.1647,
556
+ "step": 760
557
+ },
558
+ {
559
+ "epoch": 0.6738131699846861,
560
+ "grad_norm": 0.8997176885604858,
561
+ "learning_rate": 9.534107516365725e-05,
562
+ "loss": 1.167,
563
+ "step": 770
564
+ },
565
+ {
566
+ "epoch": 0.6825639903740975,
567
+ "grad_norm": 0.8208065032958984,
568
+ "learning_rate": 9.512396197547053e-05,
569
+ "loss": 1.1631,
570
+ "step": 780
571
+ },
572
+ {
573
+ "epoch": 0.6913148107635091,
574
+ "grad_norm": 0.8419785499572754,
575
+ "learning_rate": 9.490216328490692e-05,
576
+ "loss": 1.1857,
577
+ "step": 790
578
+ },
579
+ {
580
+ "epoch": 0.7000656311529205,
581
+ "grad_norm": 0.8125676512718201,
582
+ "learning_rate": 9.467570212270767e-05,
583
+ "loss": 1.2426,
584
+ "step": 800
585
+ },
586
+ {
587
+ "epoch": 0.7088164515423321,
588
+ "grad_norm": 0.9222508072853088,
589
+ "learning_rate": 9.444460200374748e-05,
590
+ "loss": 1.1452,
591
+ "step": 810
592
+ },
593
+ {
594
+ "epoch": 0.7175672719317436,
595
+ "grad_norm": 0.8932722210884094,
596
+ "learning_rate": 9.420888692459291e-05,
597
+ "loss": 1.1647,
598
+ "step": 820
599
+ },
600
+ {
601
+ "epoch": 0.7263180923211551,
602
+ "grad_norm": 0.817932665348053,
603
+ "learning_rate": 9.396858136101066e-05,
604
+ "loss": 1.1618,
605
+ "step": 830
606
+ },
607
+ {
608
+ "epoch": 0.7350689127105666,
609
+ "grad_norm": 0.8717408180236816,
610
+ "learning_rate": 9.372371026542595e-05,
611
+ "loss": 1.1517,
612
+ "step": 840
613
+ },
614
+ {
615
+ "epoch": 0.7438197330999782,
616
+ "grad_norm": 0.8578230142593384,
617
+ "learning_rate": 9.347429906433174e-05,
618
+ "loss": 1.2009,
619
+ "step": 850
620
+ },
621
+ {
622
+ "epoch": 0.7525705534893896,
623
+ "grad_norm": 0.8276301622390747,
624
+ "learning_rate": 9.322037365564845e-05,
625
+ "loss": 1.1627,
626
+ "step": 860
627
+ },
628
+ {
629
+ "epoch": 0.7613213738788012,
630
+ "grad_norm": 0.7934173941612244,
631
+ "learning_rate": 9.296196040603475e-05,
632
+ "loss": 1.1735,
633
+ "step": 870
634
+ },
635
+ {
636
+ "epoch": 0.7700721942682126,
637
+ "grad_norm": 0.8270857930183411,
638
+ "learning_rate": 9.269908614814995e-05,
639
+ "loss": 1.149,
640
+ "step": 880
641
+ },
642
+ {
643
+ "epoch": 0.7788230146576242,
644
+ "grad_norm": 1.0632795095443726,
645
+ "learning_rate": 9.243177817786757e-05,
646
+ "loss": 1.1937,
647
+ "step": 890
648
+ },
649
+ {
650
+ "epoch": 0.7875738350470357,
651
+ "grad_norm": 0.8449203968048096,
652
+ "learning_rate": 9.216006425144118e-05,
653
+ "loss": 1.1787,
654
+ "step": 900
655
+ },
656
+ {
657
+ "epoch": 0.7963246554364471,
658
+ "grad_norm": 0.7634311318397522,
659
+ "learning_rate": 9.188397258262221e-05,
660
+ "loss": 1.1445,
661
+ "step": 910
662
+ },
663
+ {
664
+ "epoch": 0.8050754758258587,
665
+ "grad_norm": 0.7856956720352173,
666
+ "learning_rate": 9.160353183973046e-05,
667
+ "loss": 1.1519,
668
+ "step": 920
669
+ },
670
+ {
671
+ "epoch": 0.8138262962152701,
672
+ "grad_norm": 0.8162206411361694,
673
+ "learning_rate": 9.131877114267711e-05,
674
+ "loss": 1.1282,
675
+ "step": 930
676
+ },
677
+ {
678
+ "epoch": 0.8225771166046817,
679
+ "grad_norm": 0.8263719081878662,
680
+ "learning_rate": 9.102972005994122e-05,
681
+ "loss": 1.1127,
682
+ "step": 940
683
+ },
684
+ {
685
+ "epoch": 0.8313279369940932,
686
+ "grad_norm": 0.831969141960144,
687
+ "learning_rate": 9.073640860549931e-05,
688
+ "loss": 1.1446,
689
+ "step": 950
690
+ },
691
+ {
692
+ "epoch": 0.8400787573835047,
693
+ "grad_norm": 0.8127765655517578,
694
+ "learning_rate": 9.043886723570884e-05,
695
+ "loss": 1.1649,
696
+ "step": 960
697
+ },
698
+ {
699
+ "epoch": 0.8488295777729162,
700
+ "grad_norm": 0.7642777562141418,
701
+ "learning_rate": 9.013712684614582e-05,
702
+ "loss": 1.1656,
703
+ "step": 970
704
+ },
705
+ {
706
+ "epoch": 0.8575803981623277,
707
+ "grad_norm": 0.8071634769439697,
708
+ "learning_rate": 8.983121876839657e-05,
709
+ "loss": 1.1414,
710
+ "step": 980
711
+ },
712
+ {
713
+ "epoch": 0.8663312185517392,
714
+ "grad_norm": 0.7789682149887085,
715
+ "learning_rate": 8.95211747668046e-05,
716
+ "loss": 1.1127,
717
+ "step": 990
718
+ },
719
+ {
720
+ "epoch": 0.8750820389411508,
721
+ "grad_norm": 0.7899104356765747,
722
+ "learning_rate": 8.920702703517204e-05,
723
+ "loss": 1.1218,
724
+ "step": 1000
725
+ },
726
+ {
727
+ "epoch": 0.8838328593305622,
728
+ "grad_norm": 0.726874053478241,
729
+ "learning_rate": 8.888880819341702e-05,
730
+ "loss": 1.1345,
731
+ "step": 1010
732
+ },
733
+ {
734
+ "epoch": 0.8925836797199738,
735
+ "grad_norm": 0.8020316958427429,
736
+ "learning_rate": 8.856655128418637e-05,
737
+ "loss": 1.1229,
738
+ "step": 1020
739
+ },
740
+ {
741
+ "epoch": 0.9013345001093852,
742
+ "grad_norm": 0.8466740250587463,
743
+ "learning_rate": 8.824028976942465e-05,
744
+ "loss": 1.1451,
745
+ "step": 1030
746
+ },
747
+ {
748
+ "epoch": 0.9100853204987968,
749
+ "grad_norm": 0.7729008793830872,
750
+ "learning_rate": 8.791005752689958e-05,
751
+ "loss": 1.1204,
752
+ "step": 1040
753
+ },
754
+ {
755
+ "epoch": 0.9188361408882083,
756
+ "grad_norm": 0.8055347204208374,
757
+ "learning_rate": 8.757588884668434e-05,
758
+ "loss": 1.1372,
759
+ "step": 1050
760
+ },
761
+ {
762
+ "epoch": 0.9275869612776197,
763
+ "grad_norm": 0.8180489540100098,
764
+ "learning_rate": 8.723781842759695e-05,
765
+ "loss": 1.1632,
766
+ "step": 1060
767
+ },
768
+ {
769
+ "epoch": 0.9363377816670313,
770
+ "grad_norm": 0.816849946975708,
771
+ "learning_rate": 8.689588137359732e-05,
772
+ "loss": 1.1226,
773
+ "step": 1070
774
+ },
775
+ {
776
+ "epoch": 0.9450886020564427,
777
+ "grad_norm": 0.8028609752655029,
778
+ "learning_rate": 8.655011319014216e-05,
779
+ "loss": 1.1559,
780
+ "step": 1080
781
+ },
782
+ {
783
+ "epoch": 0.9538394224458543,
784
+ "grad_norm": 0.8719303011894226,
785
+ "learning_rate": 8.620054978049829e-05,
786
+ "loss": 1.1522,
787
+ "step": 1090
788
+ },
789
+ {
790
+ "epoch": 0.9625902428352658,
791
+ "grad_norm": 0.766177237033844,
792
+ "learning_rate": 8.584722744201447e-05,
793
+ "loss": 1.1186,
794
+ "step": 1100
795
+ },
796
+ {
797
+ "epoch": 0.9713410632246773,
798
+ "grad_norm": 0.7790910005569458,
799
+ "learning_rate": 8.549018286235255e-05,
800
+ "loss": 1.1227,
801
+ "step": 1110
802
+ },
803
+ {
804
+ "epoch": 0.9800918836140888,
805
+ "grad_norm": 0.7803717255592346,
806
+ "learning_rate": 8.512945311567786e-05,
807
+ "loss": 1.1186,
808
+ "step": 1120
809
+ },
810
+ {
811
+ "epoch": 0.9888427040035004,
812
+ "grad_norm": 0.7084935307502747,
813
+ "learning_rate": 8.476507565880957e-05,
814
+ "loss": 1.1085,
815
+ "step": 1130
816
+ },
817
+ {
818
+ "epoch": 0.9975935243929118,
819
+ "grad_norm": 0.7696670889854431,
820
+ "learning_rate": 8.439708832733142e-05,
821
+ "loss": 1.1123,
822
+ "step": 1140
823
+ }
824
+ ],
825
+ "logging_steps": 10,
826
+ "max_steps": 3426,
827
+ "num_input_tokens_seen": 0,
828
+ "num_train_epochs": 3,
829
+ "save_steps": 500.0,
830
+ "stateful_callbacks": {
831
+ "TrainerControl": {
832
+ "args": {
833
+ "should_epoch_stop": false,
834
+ "should_evaluate": false,
835
+ "should_log": false,
836
+ "should_save": true,
837
+ "should_training_stop": false
838
+ },
839
+ "attributes": {}
840
+ }
841
+ },
842
+ "total_flos": 1.7750297385159885e+17,
843
+ "train_batch_size": 4,
844
+ "trial_name": null,
845
+ "trial_params": null
846
+ }
checkpoint-2286/optimizer_0/.metadata ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a2773be92fb3ede0513f2f27aca8b8ba4505a31612d614008b1aa98ee1a62a3
3
+ size 342948
checkpoint-2286/optimizer_0/__0_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:043869a496a8ff7ce40f8fad38d8a7ef1933dd31415eb2078e48c9ef58be892c
3
+ size 2472090861
checkpoint-2286/optimizer_0/__1_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d8fd1038dc14d3be310e17d4574716f78685a9264608f7471a5ccb2a9645a17c
3
+ size 2472174123
checkpoint-2286/optimizer_0/__2_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:448b99f03997542f61a57d0f9c79ac7eb0c54df43a61ec6e86b4cb17dac21393
3
+ size 2472182518
checkpoint-2286/optimizer_0/__3_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:135cfa2dd2b8dfd4e579dd4959853708428ac877d4337968a4c3cd6c50f6cb0b
3
+ size 2472179190
checkpoint-2286/pytorch_model_fsdp_0/.metadata ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:53b22471600c91452508b6a9f609effd598701ef136ebd4135b1feb8d1638ce1
3
+ size 129152
checkpoint-2286/pytorch_model_fsdp_0/__0_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81144f312eb9afd64fa13c0ec7eb578328a0ce3544aa94899a7423a6df0f36bb
3
+ size 1498714507
checkpoint-2286/pytorch_model_fsdp_0/__1_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:07cf921bee5c72ddd159fcc15162e0dcbe6089064e9a3c9967163eb73da239b7
3
+ size 1498714507
checkpoint-2286/pytorch_model_fsdp_0/__2_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:781a06015c2f4ff6faf67c5b24b33b23910cca0723b44d8c4614801f67d4c81f
3
+ size 1498714507
checkpoint-2286/pytorch_model_fsdp_0/__3_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f55166c0572f422cc31f2f6f435ba6777a2e7d1440ea2ff0397d2d6b571d7374
3
+ size 1498714507
checkpoint-2286/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9eee49a2ad561e75c5f0f3e8056308fc27088e19e19620813602767a759d8e5a
3
+ size 15429
checkpoint-2286/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e34c5da13614815d740bb91790974fffad60cd4e1e61540d824668f639ceb8c4
3
+ size 15429
checkpoint-2286/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:84cf36eb77e76f4d60fcc92f3755828769fe8aab6a2ca29f9573827e93d0c9f5
3
+ size 15429
checkpoint-2286/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5001f16e2075ef4f6a1c0dbc75d9121e1906a36d5ad35279fed02e36a6c7d13
3
+ size 15429
checkpoint-2286/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:91c78e6d56e32e42431584c89e071a6b0fbafa0beb69a4fcc15d1b8ed5f44422
3
+ size 1465
checkpoint-2286/trainer_state.json ADDED
@@ -0,0 +1,1644 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.0,
5
+ "eval_steps": 500,
6
+ "global_step": 2286,
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,
13
+ "eval_loss": 2.1641297340393066,
14
+ "eval_runtime": 4.2967,
15
+ "eval_samples_per_second": 4.655,
16
+ "eval_steps_per_second": 0.465,
17
+ "step": 0
18
+ },
19
+ {
20
+ "epoch": 0.0008750820389411508,
21
+ "grad_norm": 6.37307071685791,
22
+ "learning_rate": 2.915451895043732e-07,
23
+ "loss": 1.6434,
24
+ "step": 1
25
+ },
26
+ {
27
+ "epoch": 0.008750820389411508,
28
+ "grad_norm": 4.0545759201049805,
29
+ "learning_rate": 2.915451895043732e-06,
30
+ "loss": 1.6965,
31
+ "step": 10
32
+ },
33
+ {
34
+ "epoch": 0.017501640778823015,
35
+ "grad_norm": 2.363191604614258,
36
+ "learning_rate": 5.830903790087464e-06,
37
+ "loss": 1.5272,
38
+ "step": 20
39
+ },
40
+ {
41
+ "epoch": 0.02625246116823452,
42
+ "grad_norm": 1.6348848342895508,
43
+ "learning_rate": 8.746355685131195e-06,
44
+ "loss": 1.389,
45
+ "step": 30
46
+ },
47
+ {
48
+ "epoch": 0.03500328155764603,
49
+ "grad_norm": 1.6716902256011963,
50
+ "learning_rate": 1.1661807580174927e-05,
51
+ "loss": 1.348,
52
+ "step": 40
53
+ },
54
+ {
55
+ "epoch": 0.04375410194705753,
56
+ "grad_norm": 1.510228157043457,
57
+ "learning_rate": 1.457725947521866e-05,
58
+ "loss": 1.2997,
59
+ "step": 50
60
+ },
61
+ {
62
+ "epoch": 0.05250492233646904,
63
+ "grad_norm": 1.6153247356414795,
64
+ "learning_rate": 1.749271137026239e-05,
65
+ "loss": 1.2665,
66
+ "step": 60
67
+ },
68
+ {
69
+ "epoch": 0.06125574272588055,
70
+ "grad_norm": 1.446387767791748,
71
+ "learning_rate": 2.0408163265306123e-05,
72
+ "loss": 1.237,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.07000656311529206,
77
+ "grad_norm": 1.3039863109588623,
78
+ "learning_rate": 2.3323615160349855e-05,
79
+ "loss": 1.2341,
80
+ "step": 80
81
+ },
82
+ {
83
+ "epoch": 0.07875738350470357,
84
+ "grad_norm": 1.4866383075714111,
85
+ "learning_rate": 2.6239067055393584e-05,
86
+ "loss": 1.2338,
87
+ "step": 90
88
+ },
89
+ {
90
+ "epoch": 0.08750820389411507,
91
+ "grad_norm": 1.358859658241272,
92
+ "learning_rate": 2.915451895043732e-05,
93
+ "loss": 1.2331,
94
+ "step": 100
95
+ },
96
+ {
97
+ "epoch": 0.09625902428352658,
98
+ "grad_norm": 1.847506046295166,
99
+ "learning_rate": 3.206997084548105e-05,
100
+ "loss": 1.1628,
101
+ "step": 110
102
+ },
103
+ {
104
+ "epoch": 0.10500984467293809,
105
+ "grad_norm": 1.7742356061935425,
106
+ "learning_rate": 3.498542274052478e-05,
107
+ "loss": 1.1846,
108
+ "step": 120
109
+ },
110
+ {
111
+ "epoch": 0.1137606650623496,
112
+ "grad_norm": 1.3415714502334595,
113
+ "learning_rate": 3.790087463556852e-05,
114
+ "loss": 1.1699,
115
+ "step": 130
116
+ },
117
+ {
118
+ "epoch": 0.1225114854517611,
119
+ "grad_norm": 1.5319396257400513,
120
+ "learning_rate": 4.0816326530612245e-05,
121
+ "loss": 1.181,
122
+ "step": 140
123
+ },
124
+ {
125
+ "epoch": 0.1312623058411726,
126
+ "grad_norm": 1.3915642499923706,
127
+ "learning_rate": 4.373177842565598e-05,
128
+ "loss": 1.189,
129
+ "step": 150
130
+ },
131
+ {
132
+ "epoch": 0.14001312623058412,
133
+ "grad_norm": 1.175917625427246,
134
+ "learning_rate": 4.664723032069971e-05,
135
+ "loss": 1.1605,
136
+ "step": 160
137
+ },
138
+ {
139
+ "epoch": 0.14876394661999562,
140
+ "grad_norm": 1.3125519752502441,
141
+ "learning_rate": 4.956268221574344e-05,
142
+ "loss": 1.1734,
143
+ "step": 170
144
+ },
145
+ {
146
+ "epoch": 0.15751476700940714,
147
+ "grad_norm": 1.2966923713684082,
148
+ "learning_rate": 5.247813411078717e-05,
149
+ "loss": 1.2075,
150
+ "step": 180
151
+ },
152
+ {
153
+ "epoch": 0.16626558739881864,
154
+ "grad_norm": 1.2630099058151245,
155
+ "learning_rate": 5.539358600583091e-05,
156
+ "loss": 1.1791,
157
+ "step": 190
158
+ },
159
+ {
160
+ "epoch": 0.17501640778823013,
161
+ "grad_norm": 1.306665301322937,
162
+ "learning_rate": 5.830903790087464e-05,
163
+ "loss": 1.1469,
164
+ "step": 200
165
+ },
166
+ {
167
+ "epoch": 0.18376722817764166,
168
+ "grad_norm": 1.19467031955719,
169
+ "learning_rate": 6.122448979591838e-05,
170
+ "loss": 1.185,
171
+ "step": 210
172
+ },
173
+ {
174
+ "epoch": 0.19251804856705315,
175
+ "grad_norm": 1.416387677192688,
176
+ "learning_rate": 6.41399416909621e-05,
177
+ "loss": 1.2061,
178
+ "step": 220
179
+ },
180
+ {
181
+ "epoch": 0.20126886895646467,
182
+ "grad_norm": 1.3546264171600342,
183
+ "learning_rate": 6.705539358600583e-05,
184
+ "loss": 1.2016,
185
+ "step": 230
186
+ },
187
+ {
188
+ "epoch": 0.21001968934587617,
189
+ "grad_norm": 1.2417850494384766,
190
+ "learning_rate": 6.997084548104956e-05,
191
+ "loss": 1.1797,
192
+ "step": 240
193
+ },
194
+ {
195
+ "epoch": 0.2187705097352877,
196
+ "grad_norm": 1.2703570127487183,
197
+ "learning_rate": 7.28862973760933e-05,
198
+ "loss": 1.2059,
199
+ "step": 250
200
+ },
201
+ {
202
+ "epoch": 0.2275213301246992,
203
+ "grad_norm": 1.2479429244995117,
204
+ "learning_rate": 7.580174927113704e-05,
205
+ "loss": 1.1958,
206
+ "step": 260
207
+ },
208
+ {
209
+ "epoch": 0.23627215051411068,
210
+ "grad_norm": 1.507601261138916,
211
+ "learning_rate": 7.871720116618077e-05,
212
+ "loss": 1.2151,
213
+ "step": 270
214
+ },
215
+ {
216
+ "epoch": 0.2450229709035222,
217
+ "grad_norm": 1.2083854675292969,
218
+ "learning_rate": 8.163265306122449e-05,
219
+ "loss": 1.2202,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2537737912929337,
224
+ "grad_norm": 1.2939566373825073,
225
+ "learning_rate": 8.454810495626822e-05,
226
+ "loss": 1.1916,
227
+ "step": 290
228
+ },
229
+ {
230
+ "epoch": 0.2625246116823452,
231
+ "grad_norm": 1.2871382236480713,
232
+ "learning_rate": 8.746355685131195e-05,
233
+ "loss": 1.222,
234
+ "step": 300
235
+ },
236
+ {
237
+ "epoch": 0.27127543207175675,
238
+ "grad_norm": 1.1558256149291992,
239
+ "learning_rate": 9.037900874635569e-05,
240
+ "loss": 1.2094,
241
+ "step": 310
242
+ },
243
+ {
244
+ "epoch": 0.28002625246116825,
245
+ "grad_norm": 1.0638883113861084,
246
+ "learning_rate": 9.329446064139942e-05,
247
+ "loss": 1.2,
248
+ "step": 320
249
+ },
250
+ {
251
+ "epoch": 0.28877707285057974,
252
+ "grad_norm": 1.1072458028793335,
253
+ "learning_rate": 9.620991253644317e-05,
254
+ "loss": 1.224,
255
+ "step": 330
256
+ },
257
+ {
258
+ "epoch": 0.29752789323999124,
259
+ "grad_norm": 1.1332110166549683,
260
+ "learning_rate": 9.912536443148688e-05,
261
+ "loss": 1.2266,
262
+ "step": 340
263
+ },
264
+ {
265
+ "epoch": 0.30627871362940273,
266
+ "grad_norm": 1.0537033081054688,
267
+ "learning_rate": 9.999872800049233e-05,
268
+ "loss": 1.2173,
269
+ "step": 350
270
+ },
271
+ {
272
+ "epoch": 0.3150295340188143,
273
+ "grad_norm": 1.090928316116333,
274
+ "learning_rate": 9.999249795462235e-05,
275
+ "loss": 1.2416,
276
+ "step": 360
277
+ },
278
+ {
279
+ "epoch": 0.3237803544082258,
280
+ "grad_norm": 1.0084835290908813,
281
+ "learning_rate": 9.998107687592326e-05,
282
+ "loss": 1.2094,
283
+ "step": 370
284
+ },
285
+ {
286
+ "epoch": 0.3325311747976373,
287
+ "grad_norm": 1.82576584815979,
288
+ "learning_rate": 9.996446595031691e-05,
289
+ "loss": 1.2246,
290
+ "step": 380
291
+ },
292
+ {
293
+ "epoch": 0.34128199518704877,
294
+ "grad_norm": 1.2305560111999512,
295
+ "learning_rate": 9.994266690261928e-05,
296
+ "loss": 1.2647,
297
+ "step": 390
298
+ },
299
+ {
300
+ "epoch": 0.35003281557646027,
301
+ "grad_norm": 1.170067548751831,
302
+ "learning_rate": 9.991568199636148e-05,
303
+ "loss": 1.1829,
304
+ "step": 400
305
+ },
306
+ {
307
+ "epoch": 0.3587836359658718,
308
+ "grad_norm": 1.0299819707870483,
309
+ "learning_rate": 9.988351403355454e-05,
310
+ "loss": 1.2471,
311
+ "step": 410
312
+ },
313
+ {
314
+ "epoch": 0.3675344563552833,
315
+ "grad_norm": 0.9742593765258789,
316
+ "learning_rate": 9.98461663543987e-05,
317
+ "loss": 1.2049,
318
+ "step": 420
319
+ },
320
+ {
321
+ "epoch": 0.3762852767446948,
322
+ "grad_norm": 0.9377045631408691,
323
+ "learning_rate": 9.980364283693638e-05,
324
+ "loss": 1.1963,
325
+ "step": 430
326
+ },
327
+ {
328
+ "epoch": 0.3850360971341063,
329
+ "grad_norm": 1.0425899028778076,
330
+ "learning_rate": 9.97559478966496e-05,
331
+ "loss": 1.2167,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.39378691752351785,
336
+ "grad_norm": 0.9749513864517212,
337
+ "learning_rate": 9.97030864860014e-05,
338
+ "loss": 1.2166,
339
+ "step": 450
340
+ },
341
+ {
342
+ "epoch": 0.40253773791292935,
343
+ "grad_norm": 1.0552780628204346,
344
+ "learning_rate": 9.964506409392176e-05,
345
+ "loss": 1.2228,
346
+ "step": 460
347
+ },
348
+ {
349
+ "epoch": 0.41128855830234085,
350
+ "grad_norm": 1.0261808633804321,
351
+ "learning_rate": 9.958188674523748e-05,
352
+ "loss": 1.2252,
353
+ "step": 470
354
+ },
355
+ {
356
+ "epoch": 0.42003937869175234,
357
+ "grad_norm": 0.9530945420265198,
358
+ "learning_rate": 9.951356100004665e-05,
359
+ "loss": 1.1845,
360
+ "step": 480
361
+ },
362
+ {
363
+ "epoch": 0.42879019908116384,
364
+ "grad_norm": 0.8990481495857239,
365
+ "learning_rate": 9.94400939530375e-05,
366
+ "loss": 1.2283,
367
+ "step": 490
368
+ },
369
+ {
370
+ "epoch": 0.4375410194705754,
371
+ "grad_norm": 0.9713592529296875,
372
+ "learning_rate": 9.936149323275169e-05,
373
+ "loss": 1.1945,
374
+ "step": 500
375
+ },
376
+ {
377
+ "epoch": 0.4462918398599869,
378
+ "grad_norm": 1.003854513168335,
379
+ "learning_rate": 9.927776700079216e-05,
380
+ "loss": 1.2322,
381
+ "step": 510
382
+ },
383
+ {
384
+ "epoch": 0.4550426602493984,
385
+ "grad_norm": 0.935400664806366,
386
+ "learning_rate": 9.918892395097572e-05,
387
+ "loss": 1.2272,
388
+ "step": 520
389
+ },
390
+ {
391
+ "epoch": 0.4637934806388099,
392
+ "grad_norm": 1.075519323348999,
393
+ "learning_rate": 9.90949733084303e-05,
394
+ "loss": 1.2069,
395
+ "step": 530
396
+ },
397
+ {
398
+ "epoch": 0.47254430102822137,
399
+ "grad_norm": 0.9949712157249451,
400
+ "learning_rate": 9.8995924828637e-05,
401
+ "loss": 1.2083,
402
+ "step": 540
403
+ },
404
+ {
405
+ "epoch": 0.4812951214176329,
406
+ "grad_norm": 0.8942195773124695,
407
+ "learning_rate": 9.889178879641717e-05,
408
+ "loss": 1.2137,
409
+ "step": 550
410
+ },
411
+ {
412
+ "epoch": 0.4900459418070444,
413
+ "grad_norm": 0.9525908827781677,
414
+ "learning_rate": 9.878257602486445e-05,
415
+ "loss": 1.2348,
416
+ "step": 560
417
+ },
418
+ {
419
+ "epoch": 0.4987967621964559,
420
+ "grad_norm": 0.895167350769043,
421
+ "learning_rate": 9.8668297854222e-05,
422
+ "loss": 1.2034,
423
+ "step": 570
424
+ },
425
+ {
426
+ "epoch": 0.5075475825858674,
427
+ "grad_norm": 1.0244312286376953,
428
+ "learning_rate": 9.854896615070495e-05,
429
+ "loss": 1.1942,
430
+ "step": 580
431
+ },
432
+ {
433
+ "epoch": 0.516298402975279,
434
+ "grad_norm": 0.9835163950920105,
435
+ "learning_rate": 9.842459330526826e-05,
436
+ "loss": 1.2182,
437
+ "step": 590
438
+ },
439
+ {
440
+ "epoch": 0.5250492233646904,
441
+ "grad_norm": 0.9238565564155579,
442
+ "learning_rate": 9.829519223232006e-05,
443
+ "loss": 1.1992,
444
+ "step": 600
445
+ },
446
+ {
447
+ "epoch": 0.533800043754102,
448
+ "grad_norm": 0.9063233733177185,
449
+ "learning_rate": 9.81607763683808e-05,
450
+ "loss": 1.2135,
451
+ "step": 610
452
+ },
453
+ {
454
+ "epoch": 0.5425508641435135,
455
+ "grad_norm": 0.9810764193534851,
456
+ "learning_rate": 9.802135967068782e-05,
457
+ "loss": 1.1909,
458
+ "step": 620
459
+ },
460
+ {
461
+ "epoch": 0.5513016845329249,
462
+ "grad_norm": 0.9388274550437927,
463
+ "learning_rate": 9.787695661574632e-05,
464
+ "loss": 1.1693,
465
+ "step": 630
466
+ },
467
+ {
468
+ "epoch": 0.5600525049223365,
469
+ "grad_norm": 0.933283805847168,
470
+ "learning_rate": 9.772758219782601e-05,
471
+ "loss": 1.1857,
472
+ "step": 640
473
+ },
474
+ {
475
+ "epoch": 0.5688033253117479,
476
+ "grad_norm": 0.947989284992218,
477
+ "learning_rate": 9.757325192740425e-05,
478
+ "loss": 1.1901,
479
+ "step": 650
480
+ },
481
+ {
482
+ "epoch": 0.5775541457011595,
483
+ "grad_norm": 0.8272686004638672,
484
+ "learning_rate": 9.741398182955548e-05,
485
+ "loss": 1.2087,
486
+ "step": 660
487
+ },
488
+ {
489
+ "epoch": 0.586304966090571,
490
+ "grad_norm": 0.8907813429832458,
491
+ "learning_rate": 9.724978844228717e-05,
492
+ "loss": 1.201,
493
+ "step": 670
494
+ },
495
+ {
496
+ "epoch": 0.5950557864799825,
497
+ "grad_norm": 0.8374966382980347,
498
+ "learning_rate": 9.708068881482271e-05,
499
+ "loss": 1.2028,
500
+ "step": 680
501
+ },
502
+ {
503
+ "epoch": 0.603806606869394,
504
+ "grad_norm": 0.9283761382102966,
505
+ "learning_rate": 9.690670050583095e-05,
506
+ "loss": 1.1995,
507
+ "step": 690
508
+ },
509
+ {
510
+ "epoch": 0.6125574272588055,
511
+ "grad_norm": 0.8573452234268188,
512
+ "learning_rate": 9.672784158160307e-05,
513
+ "loss": 1.1372,
514
+ "step": 700
515
+ },
516
+ {
517
+ "epoch": 0.621308247648217,
518
+ "grad_norm": 0.8831506371498108,
519
+ "learning_rate": 9.654413061417655e-05,
520
+ "loss": 1.1626,
521
+ "step": 710
522
+ },
523
+ {
524
+ "epoch": 0.6300590680376286,
525
+ "grad_norm": 0.9180812239646912,
526
+ "learning_rate": 9.635558667940683e-05,
527
+ "loss": 1.1466,
528
+ "step": 720
529
+ },
530
+ {
531
+ "epoch": 0.63880988842704,
532
+ "grad_norm": 0.8657775521278381,
533
+ "learning_rate": 9.616222935498646e-05,
534
+ "loss": 1.2052,
535
+ "step": 730
536
+ },
537
+ {
538
+ "epoch": 0.6475607088164516,
539
+ "grad_norm": 0.9488115310668945,
540
+ "learning_rate": 9.596407871841229e-05,
541
+ "loss": 1.1651,
542
+ "step": 740
543
+ },
544
+ {
545
+ "epoch": 0.656311529205863,
546
+ "grad_norm": 0.8592752814292908,
547
+ "learning_rate": 9.57611553449006e-05,
548
+ "loss": 1.1953,
549
+ "step": 750
550
+ },
551
+ {
552
+ "epoch": 0.6650623495952745,
553
+ "grad_norm": 0.889445960521698,
554
+ "learning_rate": 9.555348030525076e-05,
555
+ "loss": 1.1647,
556
+ "step": 760
557
+ },
558
+ {
559
+ "epoch": 0.6738131699846861,
560
+ "grad_norm": 0.8997176885604858,
561
+ "learning_rate": 9.534107516365725e-05,
562
+ "loss": 1.167,
563
+ "step": 770
564
+ },
565
+ {
566
+ "epoch": 0.6825639903740975,
567
+ "grad_norm": 0.8208065032958984,
568
+ "learning_rate": 9.512396197547053e-05,
569
+ "loss": 1.1631,
570
+ "step": 780
571
+ },
572
+ {
573
+ "epoch": 0.6913148107635091,
574
+ "grad_norm": 0.8419785499572754,
575
+ "learning_rate": 9.490216328490692e-05,
576
+ "loss": 1.1857,
577
+ "step": 790
578
+ },
579
+ {
580
+ "epoch": 0.7000656311529205,
581
+ "grad_norm": 0.8125676512718201,
582
+ "learning_rate": 9.467570212270767e-05,
583
+ "loss": 1.2426,
584
+ "step": 800
585
+ },
586
+ {
587
+ "epoch": 0.7088164515423321,
588
+ "grad_norm": 0.9222508072853088,
589
+ "learning_rate": 9.444460200374748e-05,
590
+ "loss": 1.1452,
591
+ "step": 810
592
+ },
593
+ {
594
+ "epoch": 0.7175672719317436,
595
+ "grad_norm": 0.8932722210884094,
596
+ "learning_rate": 9.420888692459291e-05,
597
+ "loss": 1.1647,
598
+ "step": 820
599
+ },
600
+ {
601
+ "epoch": 0.7263180923211551,
602
+ "grad_norm": 0.817932665348053,
603
+ "learning_rate": 9.396858136101066e-05,
604
+ "loss": 1.1618,
605
+ "step": 830
606
+ },
607
+ {
608
+ "epoch": 0.7350689127105666,
609
+ "grad_norm": 0.8717408180236816,
610
+ "learning_rate": 9.372371026542595e-05,
611
+ "loss": 1.1517,
612
+ "step": 840
613
+ },
614
+ {
615
+ "epoch": 0.7438197330999782,
616
+ "grad_norm": 0.8578230142593384,
617
+ "learning_rate": 9.347429906433174e-05,
618
+ "loss": 1.2009,
619
+ "step": 850
620
+ },
621
+ {
622
+ "epoch": 0.7525705534893896,
623
+ "grad_norm": 0.8276301622390747,
624
+ "learning_rate": 9.322037365564845e-05,
625
+ "loss": 1.1627,
626
+ "step": 860
627
+ },
628
+ {
629
+ "epoch": 0.7613213738788012,
630
+ "grad_norm": 0.7934173941612244,
631
+ "learning_rate": 9.296196040603475e-05,
632
+ "loss": 1.1735,
633
+ "step": 870
634
+ },
635
+ {
636
+ "epoch": 0.7700721942682126,
637
+ "grad_norm": 0.8270857930183411,
638
+ "learning_rate": 9.269908614814995e-05,
639
+ "loss": 1.149,
640
+ "step": 880
641
+ },
642
+ {
643
+ "epoch": 0.7788230146576242,
644
+ "grad_norm": 1.0632795095443726,
645
+ "learning_rate": 9.243177817786757e-05,
646
+ "loss": 1.1937,
647
+ "step": 890
648
+ },
649
+ {
650
+ "epoch": 0.7875738350470357,
651
+ "grad_norm": 0.8449203968048096,
652
+ "learning_rate": 9.216006425144118e-05,
653
+ "loss": 1.1787,
654
+ "step": 900
655
+ },
656
+ {
657
+ "epoch": 0.7963246554364471,
658
+ "grad_norm": 0.7634311318397522,
659
+ "learning_rate": 9.188397258262221e-05,
660
+ "loss": 1.1445,
661
+ "step": 910
662
+ },
663
+ {
664
+ "epoch": 0.8050754758258587,
665
+ "grad_norm": 0.7856956720352173,
666
+ "learning_rate": 9.160353183973046e-05,
667
+ "loss": 1.1519,
668
+ "step": 920
669
+ },
670
+ {
671
+ "epoch": 0.8138262962152701,
672
+ "grad_norm": 0.8162206411361694,
673
+ "learning_rate": 9.131877114267711e-05,
674
+ "loss": 1.1282,
675
+ "step": 930
676
+ },
677
+ {
678
+ "epoch": 0.8225771166046817,
679
+ "grad_norm": 0.8263719081878662,
680
+ "learning_rate": 9.102972005994122e-05,
681
+ "loss": 1.1127,
682
+ "step": 940
683
+ },
684
+ {
685
+ "epoch": 0.8313279369940932,
686
+ "grad_norm": 0.831969141960144,
687
+ "learning_rate": 9.073640860549931e-05,
688
+ "loss": 1.1446,
689
+ "step": 950
690
+ },
691
+ {
692
+ "epoch": 0.8400787573835047,
693
+ "grad_norm": 0.8127765655517578,
694
+ "learning_rate": 9.043886723570884e-05,
695
+ "loss": 1.1649,
696
+ "step": 960
697
+ },
698
+ {
699
+ "epoch": 0.8488295777729162,
700
+ "grad_norm": 0.7642777562141418,
701
+ "learning_rate": 9.013712684614582e-05,
702
+ "loss": 1.1656,
703
+ "step": 970
704
+ },
705
+ {
706
+ "epoch": 0.8575803981623277,
707
+ "grad_norm": 0.8071634769439697,
708
+ "learning_rate": 8.983121876839657e-05,
709
+ "loss": 1.1414,
710
+ "step": 980
711
+ },
712
+ {
713
+ "epoch": 0.8663312185517392,
714
+ "grad_norm": 0.7789682149887085,
715
+ "learning_rate": 8.95211747668046e-05,
716
+ "loss": 1.1127,
717
+ "step": 990
718
+ },
719
+ {
720
+ "epoch": 0.8750820389411508,
721
+ "grad_norm": 0.7899104356765747,
722
+ "learning_rate": 8.920702703517204e-05,
723
+ "loss": 1.1218,
724
+ "step": 1000
725
+ },
726
+ {
727
+ "epoch": 0.8838328593305622,
728
+ "grad_norm": 0.726874053478241,
729
+ "learning_rate": 8.888880819341702e-05,
730
+ "loss": 1.1345,
731
+ "step": 1010
732
+ },
733
+ {
734
+ "epoch": 0.8925836797199738,
735
+ "grad_norm": 0.8020316958427429,
736
+ "learning_rate": 8.856655128418637e-05,
737
+ "loss": 1.1229,
738
+ "step": 1020
739
+ },
740
+ {
741
+ "epoch": 0.9013345001093852,
742
+ "grad_norm": 0.8466740250587463,
743
+ "learning_rate": 8.824028976942465e-05,
744
+ "loss": 1.1451,
745
+ "step": 1030
746
+ },
747
+ {
748
+ "epoch": 0.9100853204987968,
749
+ "grad_norm": 0.7729008793830872,
750
+ "learning_rate": 8.791005752689958e-05,
751
+ "loss": 1.1204,
752
+ "step": 1040
753
+ },
754
+ {
755
+ "epoch": 0.9188361408882083,
756
+ "grad_norm": 0.8055347204208374,
757
+ "learning_rate": 8.757588884668434e-05,
758
+ "loss": 1.1372,
759
+ "step": 1050
760
+ },
761
+ {
762
+ "epoch": 0.9275869612776197,
763
+ "grad_norm": 0.8180489540100098,
764
+ "learning_rate": 8.723781842759695e-05,
765
+ "loss": 1.1632,
766
+ "step": 1060
767
+ },
768
+ {
769
+ "epoch": 0.9363377816670313,
770
+ "grad_norm": 0.816849946975708,
771
+ "learning_rate": 8.689588137359732e-05,
772
+ "loss": 1.1226,
773
+ "step": 1070
774
+ },
775
+ {
776
+ "epoch": 0.9450886020564427,
777
+ "grad_norm": 0.8028609752655029,
778
+ "learning_rate": 8.655011319014216e-05,
779
+ "loss": 1.1559,
780
+ "step": 1080
781
+ },
782
+ {
783
+ "epoch": 0.9538394224458543,
784
+ "grad_norm": 0.8719303011894226,
785
+ "learning_rate": 8.620054978049829e-05,
786
+ "loss": 1.1522,
787
+ "step": 1090
788
+ },
789
+ {
790
+ "epoch": 0.9625902428352658,
791
+ "grad_norm": 0.766177237033844,
792
+ "learning_rate": 8.584722744201447e-05,
793
+ "loss": 1.1186,
794
+ "step": 1100
795
+ },
796
+ {
797
+ "epoch": 0.9713410632246773,
798
+ "grad_norm": 0.7790910005569458,
799
+ "learning_rate": 8.549018286235255e-05,
800
+ "loss": 1.1227,
801
+ "step": 1110
802
+ },
803
+ {
804
+ "epoch": 0.9800918836140888,
805
+ "grad_norm": 0.7803717255592346,
806
+ "learning_rate": 8.512945311567786e-05,
807
+ "loss": 1.1186,
808
+ "step": 1120
809
+ },
810
+ {
811
+ "epoch": 0.9888427040035004,
812
+ "grad_norm": 0.7084935307502747,
813
+ "learning_rate": 8.476507565880957e-05,
814
+ "loss": 1.1085,
815
+ "step": 1130
816
+ },
817
+ {
818
+ "epoch": 0.9975935243929118,
819
+ "grad_norm": 0.7696670889854431,
820
+ "learning_rate": 8.439708832733142e-05,
821
+ "loss": 1.1123,
822
+ "step": 1140
823
+ },
824
+ {
825
+ "epoch": 1.006125574272588,
826
+ "grad_norm": 0.9948705434799194,
827
+ "learning_rate": 8.402552933166288e-05,
828
+ "loss": 0.9984,
829
+ "step": 1150
830
+ },
831
+ {
832
+ "epoch": 1.0148763946619996,
833
+ "grad_norm": 0.8488800525665283,
834
+ "learning_rate": 8.36504372530916e-05,
835
+ "loss": 0.9603,
836
+ "step": 1160
837
+ },
838
+ {
839
+ "epoch": 1.023627215051411,
840
+ "grad_norm": 0.8748326897621155,
841
+ "learning_rate": 8.32718510397673e-05,
842
+ "loss": 0.9535,
843
+ "step": 1170
844
+ },
845
+ {
846
+ "epoch": 1.0323780354408225,
847
+ "grad_norm": 0.8636799454689026,
848
+ "learning_rate": 8.288981000265748e-05,
849
+ "loss": 0.9705,
850
+ "step": 1180
851
+ },
852
+ {
853
+ "epoch": 1.041128855830234,
854
+ "grad_norm": 0.7722444534301758,
855
+ "learning_rate": 8.250435381146549e-05,
856
+ "loss": 0.9411,
857
+ "step": 1190
858
+ },
859
+ {
860
+ "epoch": 1.0498796762196456,
861
+ "grad_norm": 0.792931318283081,
862
+ "learning_rate": 8.211552249051149e-05,
863
+ "loss": 0.9455,
864
+ "step": 1200
865
+ },
866
+ {
867
+ "epoch": 1.0586304966090572,
868
+ "grad_norm": 0.8179608583450317,
869
+ "learning_rate": 8.17233564145764e-05,
870
+ "loss": 0.9418,
871
+ "step": 1210
872
+ },
873
+ {
874
+ "epoch": 1.0673813169984685,
875
+ "grad_norm": 0.8081513047218323,
876
+ "learning_rate": 8.132789630470952e-05,
877
+ "loss": 0.9451,
878
+ "step": 1220
879
+ },
880
+ {
881
+ "epoch": 1.07613213738788,
882
+ "grad_norm": 0.8941613435745239,
883
+ "learning_rate": 8.092918322400027e-05,
884
+ "loss": 0.9336,
885
+ "step": 1230
886
+ },
887
+ {
888
+ "epoch": 1.0848829577772916,
889
+ "grad_norm": 0.8505945801734924,
890
+ "learning_rate": 8.052725857331429e-05,
891
+ "loss": 0.9432,
892
+ "step": 1240
893
+ },
894
+ {
895
+ "epoch": 1.0936337781667032,
896
+ "grad_norm": 0.7953508496284485,
897
+ "learning_rate": 8.012216408699455e-05,
898
+ "loss": 0.9608,
899
+ "step": 1250
900
+ },
901
+ {
902
+ "epoch": 1.1023845985561147,
903
+ "grad_norm": 0.9065775871276855,
904
+ "learning_rate": 7.971394182852789e-05,
905
+ "loss": 0.9583,
906
+ "step": 1260
907
+ },
908
+ {
909
+ "epoch": 1.1111354189455263,
910
+ "grad_norm": 0.8752979040145874,
911
+ "learning_rate": 7.930263418617713e-05,
912
+ "loss": 0.9752,
913
+ "step": 1270
914
+ },
915
+ {
916
+ "epoch": 1.1198862393349376,
917
+ "grad_norm": 0.8173210024833679,
918
+ "learning_rate": 7.888828386857985e-05,
919
+ "loss": 0.9505,
920
+ "step": 1280
921
+ },
922
+ {
923
+ "epoch": 1.1286370597243491,
924
+ "grad_norm": 0.7964232563972473,
925
+ "learning_rate": 7.84709339003135e-05,
926
+ "loss": 0.9346,
927
+ "step": 1290
928
+ },
929
+ {
930
+ "epoch": 1.1373878801137607,
931
+ "grad_norm": 0.8364039063453674,
932
+ "learning_rate": 7.805062761742799e-05,
933
+ "loss": 0.9606,
934
+ "step": 1300
935
+ },
936
+ {
937
+ "epoch": 1.1461387005031722,
938
+ "grad_norm": 0.8122743964195251,
939
+ "learning_rate": 7.76274086629458e-05,
940
+ "loss": 0.9455,
941
+ "step": 1310
942
+ },
943
+ {
944
+ "epoch": 1.1548895208925836,
945
+ "grad_norm": 0.8229530453681946,
946
+ "learning_rate": 7.720132098233031e-05,
947
+ "loss": 0.9465,
948
+ "step": 1320
949
+ },
950
+ {
951
+ "epoch": 1.1636403412819951,
952
+ "grad_norm": 0.7859545946121216,
953
+ "learning_rate": 7.677240881892258e-05,
954
+ "loss": 0.9421,
955
+ "step": 1330
956
+ },
957
+ {
958
+ "epoch": 1.1723911616714067,
959
+ "grad_norm": 0.8277194499969482,
960
+ "learning_rate": 7.634071670934734e-05,
961
+ "loss": 0.9541,
962
+ "step": 1340
963
+ },
964
+ {
965
+ "epoch": 1.1811419820608182,
966
+ "grad_norm": 0.7695391774177551,
967
+ "learning_rate": 7.590628947888847e-05,
968
+ "loss": 0.9433,
969
+ "step": 1350
970
+ },
971
+ {
972
+ "epoch": 1.1898928024502298,
973
+ "grad_norm": 0.82956862449646,
974
+ "learning_rate": 7.546917223683453e-05,
975
+ "loss": 0.9395,
976
+ "step": 1360
977
+ },
978
+ {
979
+ "epoch": 1.1986436228396413,
980
+ "grad_norm": 0.8398957252502441,
981
+ "learning_rate": 7.502941037179474e-05,
982
+ "loss": 0.9432,
983
+ "step": 1370
984
+ },
985
+ {
986
+ "epoch": 1.2073944432290526,
987
+ "grad_norm": 0.8232219219207764,
988
+ "learning_rate": 7.4587049546986e-05,
989
+ "loss": 0.9887,
990
+ "step": 1380
991
+ },
992
+ {
993
+ "epoch": 1.2161452636184642,
994
+ "grad_norm": 0.7886401414871216,
995
+ "learning_rate": 7.414213569549145e-05,
996
+ "loss": 0.959,
997
+ "step": 1390
998
+ },
999
+ {
1000
+ "epoch": 1.2248960840078758,
1001
+ "grad_norm": 0.7588061690330505,
1002
+ "learning_rate": 7.369471501549087e-05,
1003
+ "loss": 0.9651,
1004
+ "step": 1400
1005
+ },
1006
+ {
1007
+ "epoch": 1.2336469043972873,
1008
+ "grad_norm": 0.8234547972679138,
1009
+ "learning_rate": 7.324483396546371e-05,
1010
+ "loss": 0.9935,
1011
+ "step": 1410
1012
+ },
1013
+ {
1014
+ "epoch": 1.2423977247866986,
1015
+ "grad_norm": 0.7623049020767212,
1016
+ "learning_rate": 7.279253925936498e-05,
1017
+ "loss": 0.95,
1018
+ "step": 1420
1019
+ },
1020
+ {
1021
+ "epoch": 1.2511485451761102,
1022
+ "grad_norm": 0.7172533869743347,
1023
+ "learning_rate": 7.233787786177464e-05,
1024
+ "loss": 0.9441,
1025
+ "step": 1430
1026
+ },
1027
+ {
1028
+ "epoch": 1.2598993655655217,
1029
+ "grad_norm": 0.7973024249076843,
1030
+ "learning_rate": 7.188089698302098e-05,
1031
+ "loss": 0.9682,
1032
+ "step": 1440
1033
+ },
1034
+ {
1035
+ "epoch": 1.2686501859549333,
1036
+ "grad_norm": 0.7969169020652771,
1037
+ "learning_rate": 7.14216440742785e-05,
1038
+ "loss": 0.9541,
1039
+ "step": 1450
1040
+ },
1041
+ {
1042
+ "epoch": 1.2774010063443448,
1043
+ "grad_norm": 0.8296008110046387,
1044
+ "learning_rate": 7.096016682264074e-05,
1045
+ "loss": 0.9597,
1046
+ "step": 1460
1047
+ },
1048
+ {
1049
+ "epoch": 1.2861518267337564,
1050
+ "grad_norm": 0.8538773655891418,
1051
+ "learning_rate": 7.049651314616862e-05,
1052
+ "loss": 0.9225,
1053
+ "step": 1470
1054
+ },
1055
+ {
1056
+ "epoch": 1.2949026471231677,
1057
+ "grad_norm": 0.8068443536758423,
1058
+ "learning_rate": 7.003073118891487e-05,
1059
+ "loss": 0.9282,
1060
+ "step": 1480
1061
+ },
1062
+ {
1063
+ "epoch": 1.3036534675125793,
1064
+ "grad_norm": 0.8286350965499878,
1065
+ "learning_rate": 6.956286931592481e-05,
1066
+ "loss": 0.9645,
1067
+ "step": 1490
1068
+ },
1069
+ {
1070
+ "epoch": 1.3124042879019908,
1071
+ "grad_norm": 0.796650767326355,
1072
+ "learning_rate": 6.909297610821443e-05,
1073
+ "loss": 0.934,
1074
+ "step": 1500
1075
+ },
1076
+ {
1077
+ "epoch": 1.3211551082914024,
1078
+ "grad_norm": 0.7479636073112488,
1079
+ "learning_rate": 6.862110035772589e-05,
1080
+ "loss": 0.9319,
1081
+ "step": 1510
1082
+ },
1083
+ {
1084
+ "epoch": 1.3299059286808137,
1085
+ "grad_norm": 0.7734546065330505,
1086
+ "learning_rate": 6.81472910622611e-05,
1087
+ "loss": 0.9577,
1088
+ "step": 1520
1089
+ },
1090
+ {
1091
+ "epoch": 1.3386567490702252,
1092
+ "grad_norm": 0.821553647518158,
1093
+ "learning_rate": 6.7671597420394e-05,
1094
+ "loss": 0.966,
1095
+ "step": 1530
1096
+ },
1097
+ {
1098
+ "epoch": 1.3474075694596368,
1099
+ "grad_norm": 0.8141497373580933,
1100
+ "learning_rate": 6.719406882636196e-05,
1101
+ "loss": 0.9593,
1102
+ "step": 1540
1103
+ },
1104
+ {
1105
+ "epoch": 1.3561583898490484,
1106
+ "grad_norm": 0.7738481163978577,
1107
+ "learning_rate": 6.671475486493691e-05,
1108
+ "loss": 0.9474,
1109
+ "step": 1550
1110
+ },
1111
+ {
1112
+ "epoch": 1.36490921023846,
1113
+ "grad_norm": 0.7933306694030762,
1114
+ "learning_rate": 6.62337053062766e-05,
1115
+ "loss": 0.9297,
1116
+ "step": 1560
1117
+ },
1118
+ {
1119
+ "epoch": 1.3736600306278715,
1120
+ "grad_norm": 0.7921035885810852,
1121
+ "learning_rate": 6.575097010075664e-05,
1122
+ "loss": 0.9283,
1123
+ "step": 1570
1124
+ },
1125
+ {
1126
+ "epoch": 1.382410851017283,
1127
+ "grad_norm": 0.7895304560661316,
1128
+ "learning_rate": 6.526659937378392e-05,
1129
+ "loss": 0.9559,
1130
+ "step": 1580
1131
+ },
1132
+ {
1133
+ "epoch": 1.3911616714066943,
1134
+ "grad_norm": 0.8518770933151245,
1135
+ "learning_rate": 6.478064342059164e-05,
1136
+ "loss": 0.9571,
1137
+ "step": 1590
1138
+ },
1139
+ {
1140
+ "epoch": 1.3999124917961059,
1141
+ "grad_norm": 0.8479484915733337,
1142
+ "learning_rate": 6.429315270101701e-05,
1143
+ "loss": 0.9578,
1144
+ "step": 1600
1145
+ },
1146
+ {
1147
+ "epoch": 1.4086633121855174,
1148
+ "grad_norm": 0.763520359992981,
1149
+ "learning_rate": 6.380417783426153e-05,
1150
+ "loss": 0.9304,
1151
+ "step": 1610
1152
+ },
1153
+ {
1154
+ "epoch": 1.4174141325749288,
1155
+ "grad_norm": 0.7657537460327148,
1156
+ "learning_rate": 6.3313769593635e-05,
1157
+ "loss": 0.9541,
1158
+ "step": 1620
1159
+ },
1160
+ {
1161
+ "epoch": 1.4261649529643403,
1162
+ "grad_norm": 0.7461577653884888,
1163
+ "learning_rate": 6.282197890128338e-05,
1164
+ "loss": 0.9331,
1165
+ "step": 1630
1166
+ },
1167
+ {
1168
+ "epoch": 1.4349157733537519,
1169
+ "grad_norm": 0.7814643383026123,
1170
+ "learning_rate": 6.23288568229012e-05,
1171
+ "loss": 0.9431,
1172
+ "step": 1640
1173
+ },
1174
+ {
1175
+ "epoch": 1.4436665937431634,
1176
+ "grad_norm": 0.7600048184394836,
1177
+ "learning_rate": 6.183445456242904e-05,
1178
+ "loss": 0.9549,
1179
+ "step": 1650
1180
+ },
1181
+ {
1182
+ "epoch": 1.452417414132575,
1183
+ "grad_norm": 0.7684770226478577,
1184
+ "learning_rate": 6.13388234567369e-05,
1185
+ "loss": 0.9074,
1186
+ "step": 1660
1187
+ },
1188
+ {
1189
+ "epoch": 1.4611682345219865,
1190
+ "grad_norm": 0.7696465849876404,
1191
+ "learning_rate": 6.084201497029333e-05,
1192
+ "loss": 0.9449,
1193
+ "step": 1670
1194
+ },
1195
+ {
1196
+ "epoch": 1.469919054911398,
1197
+ "grad_norm": 0.818397045135498,
1198
+ "learning_rate": 6.034408068982172e-05,
1199
+ "loss": 0.9396,
1200
+ "step": 1680
1201
+ },
1202
+ {
1203
+ "epoch": 1.4786698753008094,
1204
+ "grad_norm": 0.7443113923072815,
1205
+ "learning_rate": 5.984507231894364e-05,
1206
+ "loss": 0.928,
1207
+ "step": 1690
1208
+ },
1209
+ {
1210
+ "epoch": 1.487420695690221,
1211
+ "grad_norm": 0.8166981339454651,
1212
+ "learning_rate": 5.934504167281022e-05,
1213
+ "loss": 0.9323,
1214
+ "step": 1700
1215
+ },
1216
+ {
1217
+ "epoch": 1.4961715160796325,
1218
+ "grad_norm": 0.8331442475318909,
1219
+ "learning_rate": 5.884404067272174e-05,
1220
+ "loss": 0.9458,
1221
+ "step": 1710
1222
+ },
1223
+ {
1224
+ "epoch": 1.5049223364690438,
1225
+ "grad_norm": 0.7990359663963318,
1226
+ "learning_rate": 5.834212134073644e-05,
1227
+ "loss": 0.9094,
1228
+ "step": 1720
1229
+ },
1230
+ {
1231
+ "epoch": 1.5136731568584554,
1232
+ "grad_norm": 0.768055260181427,
1233
+ "learning_rate": 5.7839335794268666e-05,
1234
+ "loss": 0.9385,
1235
+ "step": 1730
1236
+ },
1237
+ {
1238
+ "epoch": 1.522423977247867,
1239
+ "grad_norm": 0.7425386309623718,
1240
+ "learning_rate": 5.7335736240677174e-05,
1241
+ "loss": 0.9446,
1242
+ "step": 1740
1243
+ },
1244
+ {
1245
+ "epoch": 1.5311747976372785,
1246
+ "grad_norm": 0.7698354721069336,
1247
+ "learning_rate": 5.683137497184424e-05,
1248
+ "loss": 0.9057,
1249
+ "step": 1750
1250
+ },
1251
+ {
1252
+ "epoch": 1.53992561802669,
1253
+ "grad_norm": 0.8186885118484497,
1254
+ "learning_rate": 5.6326304358745694e-05,
1255
+ "loss": 0.9758,
1256
+ "step": 1760
1257
+ },
1258
+ {
1259
+ "epoch": 1.5486764384161016,
1260
+ "grad_norm": 0.8006178736686707,
1261
+ "learning_rate": 5.5820576846013086e-05,
1262
+ "loss": 0.9424,
1263
+ "step": 1770
1264
+ },
1265
+ {
1266
+ "epoch": 1.5574272588055131,
1267
+ "grad_norm": 0.7572160363197327,
1268
+ "learning_rate": 5.531424494648789e-05,
1269
+ "loss": 0.945,
1270
+ "step": 1780
1271
+ },
1272
+ {
1273
+ "epoch": 1.5661780791949247,
1274
+ "grad_norm": 0.8019212484359741,
1275
+ "learning_rate": 5.480736123576886e-05,
1276
+ "loss": 0.9528,
1277
+ "step": 1790
1278
+ },
1279
+ {
1280
+ "epoch": 1.574928899584336,
1281
+ "grad_norm": 0.7867367267608643,
1282
+ "learning_rate": 5.4299978346752756e-05,
1283
+ "loss": 0.9499,
1284
+ "step": 1800
1285
+ },
1286
+ {
1287
+ "epoch": 1.5836797199737476,
1288
+ "grad_norm": 0.7404739856719971,
1289
+ "learning_rate": 5.379214896416907e-05,
1290
+ "loss": 0.9177,
1291
+ "step": 1810
1292
+ },
1293
+ {
1294
+ "epoch": 1.592430540363159,
1295
+ "grad_norm": 0.712374746799469,
1296
+ "learning_rate": 5.328392581910961e-05,
1297
+ "loss": 0.9212,
1298
+ "step": 1820
1299
+ },
1300
+ {
1301
+ "epoch": 1.6011813607525704,
1302
+ "grad_norm": 0.7724602222442627,
1303
+ "learning_rate": 5.277536168355293e-05,
1304
+ "loss": 0.9402,
1305
+ "step": 1830
1306
+ },
1307
+ {
1308
+ "epoch": 1.609932181141982,
1309
+ "grad_norm": 0.8452624678611755,
1310
+ "learning_rate": 5.226650936488474e-05,
1311
+ "loss": 0.9193,
1312
+ "step": 1840
1313
+ },
1314
+ {
1315
+ "epoch": 1.6186830015313936,
1316
+ "grad_norm": 0.7993892431259155,
1317
+ "learning_rate": 5.1757421700414645e-05,
1318
+ "loss": 0.9328,
1319
+ "step": 1850
1320
+ },
1321
+ {
1322
+ "epoch": 1.627433821920805,
1323
+ "grad_norm": 0.7931053042411804,
1324
+ "learning_rate": 5.124815155188966e-05,
1325
+ "loss": 0.9759,
1326
+ "step": 1860
1327
+ },
1328
+ {
1329
+ "epoch": 1.6361846423102167,
1330
+ "grad_norm": 0.7516879439353943,
1331
+ "learning_rate": 5.0738751800005226e-05,
1332
+ "loss": 0.9229,
1333
+ "step": 1870
1334
+ },
1335
+ {
1336
+ "epoch": 1.6449354626996282,
1337
+ "grad_norm": 0.6935803890228271,
1338
+ "learning_rate": 5.022927533891434e-05,
1339
+ "loss": 0.9234,
1340
+ "step": 1880
1341
+ },
1342
+ {
1343
+ "epoch": 1.6536862830890398,
1344
+ "grad_norm": 0.8120819926261902,
1345
+ "learning_rate": 4.9719775070735196e-05,
1346
+ "loss": 0.9582,
1347
+ "step": 1890
1348
+ },
1349
+ {
1350
+ "epoch": 1.662437103478451,
1351
+ "grad_norm": 0.781491219997406,
1352
+ "learning_rate": 4.921030390005802e-05,
1353
+ "loss": 0.9177,
1354
+ "step": 1900
1355
+ },
1356
+ {
1357
+ "epoch": 1.6711879238678626,
1358
+ "grad_norm": 0.7723197340965271,
1359
+ "learning_rate": 4.870091472845166e-05,
1360
+ "loss": 0.9382,
1361
+ "step": 1910
1362
+ },
1363
+ {
1364
+ "epoch": 1.679938744257274,
1365
+ "grad_norm": 0.751848578453064,
1366
+ "learning_rate": 4.819166044897047e-05,
1367
+ "loss": 0.9148,
1368
+ "step": 1920
1369
+ },
1370
+ {
1371
+ "epoch": 1.6886895646466855,
1372
+ "grad_norm": 0.8265687823295593,
1373
+ "learning_rate": 4.768259394066216e-05,
1374
+ "loss": 0.9386,
1375
+ "step": 1930
1376
+ },
1377
+ {
1378
+ "epoch": 1.697440385036097,
1379
+ "grad_norm": 0.7489150762557983,
1380
+ "learning_rate": 4.7173768063076956e-05,
1381
+ "loss": 0.9337,
1382
+ "step": 1940
1383
+ },
1384
+ {
1385
+ "epoch": 1.7061912054255086,
1386
+ "grad_norm": 0.8112419247627258,
1387
+ "learning_rate": 4.666523565077889e-05,
1388
+ "loss": 0.917,
1389
+ "step": 1950
1390
+ },
1391
+ {
1392
+ "epoch": 1.7149420258149202,
1393
+ "grad_norm": 0.7611903548240662,
1394
+ "learning_rate": 4.615704950785965e-05,
1395
+ "loss": 0.9338,
1396
+ "step": 1960
1397
+ },
1398
+ {
1399
+ "epoch": 1.7236928462043317,
1400
+ "grad_norm": 0.7738552093505859,
1401
+ "learning_rate": 4.5649262402455704e-05,
1402
+ "loss": 0.9359,
1403
+ "step": 1970
1404
+ },
1405
+ {
1406
+ "epoch": 1.7324436665937433,
1407
+ "grad_norm": 0.7741129994392395,
1408
+ "learning_rate": 4.5141927061268844e-05,
1409
+ "loss": 0.9209,
1410
+ "step": 1980
1411
+ },
1412
+ {
1413
+ "epoch": 1.7411944869831548,
1414
+ "grad_norm": 0.7815123200416565,
1415
+ "learning_rate": 4.463509616409144e-05,
1416
+ "loss": 0.916,
1417
+ "step": 1990
1418
+ },
1419
+ {
1420
+ "epoch": 1.7499453073725662,
1421
+ "grad_norm": 0.6793907284736633,
1422
+ "learning_rate": 4.412882233833629e-05,
1423
+ "loss": 0.9042,
1424
+ "step": 2000
1425
+ },
1426
+ {
1427
+ "epoch": 1.7586961277619777,
1428
+ "grad_norm": 0.711877167224884,
1429
+ "learning_rate": 4.362315815357197e-05,
1430
+ "loss": 0.9072,
1431
+ "step": 2010
1432
+ },
1433
+ {
1434
+ "epoch": 1.767446948151389,
1435
+ "grad_norm": 0.7671152353286743,
1436
+ "learning_rate": 4.311815611606423e-05,
1437
+ "loss": 0.9226,
1438
+ "step": 2020
1439
+ },
1440
+ {
1441
+ "epoch": 1.7761977685408006,
1442
+ "grad_norm": 0.7221760153770447,
1443
+ "learning_rate": 4.261386866332397e-05,
1444
+ "loss": 0.9077,
1445
+ "step": 2030
1446
+ },
1447
+ {
1448
+ "epoch": 1.7849485889302121,
1449
+ "grad_norm": 0.7419185042381287,
1450
+ "learning_rate": 4.211034815866228e-05,
1451
+ "loss": 0.9017,
1452
+ "step": 2040
1453
+ },
1454
+ {
1455
+ "epoch": 1.7936994093196237,
1456
+ "grad_norm": 0.740652859210968,
1457
+ "learning_rate": 4.1607646885753206e-05,
1458
+ "loss": 0.9159,
1459
+ "step": 2050
1460
+ },
1461
+ {
1462
+ "epoch": 1.8024502297090352,
1463
+ "grad_norm": 0.7923645377159119,
1464
+ "learning_rate": 4.110581704320495e-05,
1465
+ "loss": 0.9135,
1466
+ "step": 2060
1467
+ },
1468
+ {
1469
+ "epoch": 1.8112010500984468,
1470
+ "grad_norm": 0.7528581023216248,
1471
+ "learning_rate": 4.060491073913957e-05,
1472
+ "loss": 0.9291,
1473
+ "step": 2070
1474
+ },
1475
+ {
1476
+ "epoch": 1.8199518704878583,
1477
+ "grad_norm": 0.7974929809570312,
1478
+ "learning_rate": 4.010497998578239e-05,
1479
+ "loss": 0.9208,
1480
+ "step": 2080
1481
+ },
1482
+ {
1483
+ "epoch": 1.8287026908772699,
1484
+ "grad_norm": 0.7388021349906921,
1485
+ "learning_rate": 3.960607669406129e-05,
1486
+ "loss": 0.9155,
1487
+ "step": 2090
1488
+ },
1489
+ {
1490
+ "epoch": 1.8374535112666812,
1491
+ "grad_norm": 0.7927631735801697,
1492
+ "learning_rate": 3.910825266821629e-05,
1493
+ "loss": 0.9156,
1494
+ "step": 2100
1495
+ },
1496
+ {
1497
+ "epoch": 1.8462043316560928,
1498
+ "grad_norm": 0.7545862793922424,
1499
+ "learning_rate": 3.861155960042057e-05,
1500
+ "loss": 0.877,
1501
+ "step": 2110
1502
+ },
1503
+ {
1504
+ "epoch": 1.8549551520455043,
1505
+ "grad_norm": 0.8433863520622253,
1506
+ "learning_rate": 3.8116049065412864e-05,
1507
+ "loss": 0.9107,
1508
+ "step": 2120
1509
+ },
1510
+ {
1511
+ "epoch": 1.8637059724349156,
1512
+ "grad_norm": 0.7610001564025879,
1513
+ "learning_rate": 3.762177251514218e-05,
1514
+ "loss": 0.9068,
1515
+ "step": 2130
1516
+ },
1517
+ {
1518
+ "epoch": 1.8724567928243272,
1519
+ "grad_norm": 0.7452635765075684,
1520
+ "learning_rate": 3.712878127342515e-05,
1521
+ "loss": 0.9113,
1522
+ "step": 2140
1523
+ },
1524
+ {
1525
+ "epoch": 1.8812076132137387,
1526
+ "grad_norm": 0.7530847787857056,
1527
+ "learning_rate": 3.663712653061682e-05,
1528
+ "loss": 0.8881,
1529
+ "step": 2150
1530
+ },
1531
+ {
1532
+ "epoch": 1.8899584336031503,
1533
+ "grad_norm": 0.7940835356712341,
1534
+ "learning_rate": 3.614685933829525e-05,
1535
+ "loss": 0.8807,
1536
+ "step": 2160
1537
+ },
1538
+ {
1539
+ "epoch": 1.8987092539925619,
1540
+ "grad_norm": 0.7291942834854126,
1541
+ "learning_rate": 3.5658030603960436e-05,
1542
+ "loss": 0.9112,
1543
+ "step": 2170
1544
+ },
1545
+ {
1546
+ "epoch": 1.9074600743819734,
1547
+ "grad_norm": 0.7321741580963135,
1548
+ "learning_rate": 3.5170691085748324e-05,
1549
+ "loss": 0.8835,
1550
+ "step": 2180
1551
+ },
1552
+ {
1553
+ "epoch": 1.916210894771385,
1554
+ "grad_norm": 0.7166718244552612,
1555
+ "learning_rate": 3.468489138716029e-05,
1556
+ "loss": 0.9115,
1557
+ "step": 2190
1558
+ },
1559
+ {
1560
+ "epoch": 1.9249617151607963,
1561
+ "grad_norm": 0.741908848285675,
1562
+ "learning_rate": 3.42006819518086e-05,
1563
+ "loss": 0.9198,
1564
+ "step": 2200
1565
+ },
1566
+ {
1567
+ "epoch": 1.9337125355502078,
1568
+ "grad_norm": 0.8034552931785583,
1569
+ "learning_rate": 3.371811305817858e-05,
1570
+ "loss": 0.8952,
1571
+ "step": 2210
1572
+ },
1573
+ {
1574
+ "epoch": 1.9424633559396194,
1575
+ "grad_norm": 0.6716712713241577,
1576
+ "learning_rate": 3.32372348144079e-05,
1577
+ "loss": 0.9197,
1578
+ "step": 2220
1579
+ },
1580
+ {
1581
+ "epoch": 1.9512141763290307,
1582
+ "grad_norm": 0.6731935739517212,
1583
+ "learning_rate": 3.275809715308349e-05,
1584
+ "loss": 0.8949,
1585
+ "step": 2230
1586
+ },
1587
+ {
1588
+ "epoch": 1.9599649967184423,
1589
+ "grad_norm": 0.7521949410438538,
1590
+ "learning_rate": 3.228074982605673e-05,
1591
+ "loss": 0.8949,
1592
+ "step": 2240
1593
+ },
1594
+ {
1595
+ "epoch": 1.9687158171078538,
1596
+ "grad_norm": 0.6889147162437439,
1597
+ "learning_rate": 3.180524239927749e-05,
1598
+ "loss": 0.8941,
1599
+ "step": 2250
1600
+ },
1601
+ {
1602
+ "epoch": 1.9774666374972654,
1603
+ "grad_norm": 0.7169541716575623,
1604
+ "learning_rate": 3.1331624247647276e-05,
1605
+ "loss": 0.8922,
1606
+ "step": 2260
1607
+ },
1608
+ {
1609
+ "epoch": 1.986217457886677,
1610
+ "grad_norm": 0.7812347412109375,
1611
+ "learning_rate": 3.0859944549892325e-05,
1612
+ "loss": 0.8915,
1613
+ "step": 2270
1614
+ },
1615
+ {
1616
+ "epoch": 1.9949682782760885,
1617
+ "grad_norm": 0.7609400153160095,
1618
+ "learning_rate": 3.039025228345725e-05,
1619
+ "loss": 0.9227,
1620
+ "step": 2280
1621
+ }
1622
+ ],
1623
+ "logging_steps": 10,
1624
+ "max_steps": 3426,
1625
+ "num_input_tokens_seen": 0,
1626
+ "num_train_epochs": 3,
1627
+ "save_steps": 500.0,
1628
+ "stateful_callbacks": {
1629
+ "TrainerControl": {
1630
+ "args": {
1631
+ "should_epoch_stop": false,
1632
+ "should_evaluate": false,
1633
+ "should_log": false,
1634
+ "should_save": true,
1635
+ "should_training_stop": false
1636
+ },
1637
+ "attributes": {}
1638
+ }
1639
+ },
1640
+ "total_flos": 3.5506578379505664e+17,
1641
+ "train_batch_size": 4,
1642
+ "trial_name": null,
1643
+ "trial_params": null
1644
+ }
checkpoint-3426/optimizer_0/.metadata ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d5c2ae482a23462ab828266f49e94c90bc98f16eca9d704c846e15ff326036b0
3
+ size 342948
checkpoint-3426/optimizer_0/__0_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e28eb80ae17acea6cbbdffde797c2e073087dc3258069919cf6a739dd4a2606a
3
+ size 2472090861
checkpoint-3426/optimizer_0/__1_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cedc6e67a3192909e8d053106409582eee53e87398139e38da601372c295bc88
3
+ size 2472174123
checkpoint-3426/optimizer_0/__2_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58737096141869c1f119f831ee45d25f462ff4b15cb51e34484b5c1730af49c6
3
+ size 2472182518
checkpoint-3426/optimizer_0/__3_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b84184914149b14fc5d90e5021a1b98b7ac08960f9da0bf4c8e48a2f7f4f772
3
+ size 2472179190
checkpoint-3426/pytorch_model_fsdp_0/.metadata ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a7fe491a7e31c2a3293024e7e8c6bb3c374a7ae061edb5959cd282a1d48309a
3
+ size 129152
checkpoint-3426/pytorch_model_fsdp_0/__0_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7e46456f9526b1dce3f4498f59cbc2f39ae998e04e5517452b26a8ca91d98dc5
3
+ size 1498714507
checkpoint-3426/pytorch_model_fsdp_0/__1_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d1e8dd555862fcabaca1e467beb02b04255a0bbc3427ffc10a486d84d77baca
3
+ size 1498714507
checkpoint-3426/pytorch_model_fsdp_0/__2_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f07ba245dadc9525c6c7941939f8b0e8a20a18d7ca34c7694c3bf8b42676725e
3
+ size 1498714507
checkpoint-3426/pytorch_model_fsdp_0/__3_0.distcp ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e31a24d8dba9da94fd11387994bba3028c37c6ba3c977afcde97ac7dfbd4f07
3
+ size 1498714507
checkpoint-3426/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b80829a78a755ba0cf5e985f9212f164229ce25f7c02fa877b4cdebf230922b5
3
+ size 15429
checkpoint-3426/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c6f4074cb5241eff8959c1f45b18c23c893119f6c03fa2da27359dbe7dba66c
3
+ size 15429
checkpoint-3426/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1e84026d823a1c04897f710cd119a65b654c08ad443e97c4486fc7a26f6e64b
3
+ size 15429
checkpoint-3426/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9b60a41c989678f542cc75416e61e7a43f81d6221d259490134253d1f38a920
3
+ size 15429
checkpoint-3426/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:932ccc456461adf32cd26dd5fb4d85a4d38f69661d41fe9883bf9a5066a53c7d
3
+ size 1465
checkpoint-3426/trainer_state.json ADDED
@@ -0,0 +1,2442 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.997593524392912,
5
+ "eval_steps": 500,
6
+ "global_step": 3426,
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,
13
+ "eval_loss": 2.1641297340393066,
14
+ "eval_runtime": 4.2967,
15
+ "eval_samples_per_second": 4.655,
16
+ "eval_steps_per_second": 0.465,
17
+ "step": 0
18
+ },
19
+ {
20
+ "epoch": 0.0008750820389411508,
21
+ "grad_norm": 6.37307071685791,
22
+ "learning_rate": 2.915451895043732e-07,
23
+ "loss": 1.6434,
24
+ "step": 1
25
+ },
26
+ {
27
+ "epoch": 0.008750820389411508,
28
+ "grad_norm": 4.0545759201049805,
29
+ "learning_rate": 2.915451895043732e-06,
30
+ "loss": 1.6965,
31
+ "step": 10
32
+ },
33
+ {
34
+ "epoch": 0.017501640778823015,
35
+ "grad_norm": 2.363191604614258,
36
+ "learning_rate": 5.830903790087464e-06,
37
+ "loss": 1.5272,
38
+ "step": 20
39
+ },
40
+ {
41
+ "epoch": 0.02625246116823452,
42
+ "grad_norm": 1.6348848342895508,
43
+ "learning_rate": 8.746355685131195e-06,
44
+ "loss": 1.389,
45
+ "step": 30
46
+ },
47
+ {
48
+ "epoch": 0.03500328155764603,
49
+ "grad_norm": 1.6716902256011963,
50
+ "learning_rate": 1.1661807580174927e-05,
51
+ "loss": 1.348,
52
+ "step": 40
53
+ },
54
+ {
55
+ "epoch": 0.04375410194705753,
56
+ "grad_norm": 1.510228157043457,
57
+ "learning_rate": 1.457725947521866e-05,
58
+ "loss": 1.2997,
59
+ "step": 50
60
+ },
61
+ {
62
+ "epoch": 0.05250492233646904,
63
+ "grad_norm": 1.6153247356414795,
64
+ "learning_rate": 1.749271137026239e-05,
65
+ "loss": 1.2665,
66
+ "step": 60
67
+ },
68
+ {
69
+ "epoch": 0.06125574272588055,
70
+ "grad_norm": 1.446387767791748,
71
+ "learning_rate": 2.0408163265306123e-05,
72
+ "loss": 1.237,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.07000656311529206,
77
+ "grad_norm": 1.3039863109588623,
78
+ "learning_rate": 2.3323615160349855e-05,
79
+ "loss": 1.2341,
80
+ "step": 80
81
+ },
82
+ {
83
+ "epoch": 0.07875738350470357,
84
+ "grad_norm": 1.4866383075714111,
85
+ "learning_rate": 2.6239067055393584e-05,
86
+ "loss": 1.2338,
87
+ "step": 90
88
+ },
89
+ {
90
+ "epoch": 0.08750820389411507,
91
+ "grad_norm": 1.358859658241272,
92
+ "learning_rate": 2.915451895043732e-05,
93
+ "loss": 1.2331,
94
+ "step": 100
95
+ },
96
+ {
97
+ "epoch": 0.09625902428352658,
98
+ "grad_norm": 1.847506046295166,
99
+ "learning_rate": 3.206997084548105e-05,
100
+ "loss": 1.1628,
101
+ "step": 110
102
+ },
103
+ {
104
+ "epoch": 0.10500984467293809,
105
+ "grad_norm": 1.7742356061935425,
106
+ "learning_rate": 3.498542274052478e-05,
107
+ "loss": 1.1846,
108
+ "step": 120
109
+ },
110
+ {
111
+ "epoch": 0.1137606650623496,
112
+ "grad_norm": 1.3415714502334595,
113
+ "learning_rate": 3.790087463556852e-05,
114
+ "loss": 1.1699,
115
+ "step": 130
116
+ },
117
+ {
118
+ "epoch": 0.1225114854517611,
119
+ "grad_norm": 1.5319396257400513,
120
+ "learning_rate": 4.0816326530612245e-05,
121
+ "loss": 1.181,
122
+ "step": 140
123
+ },
124
+ {
125
+ "epoch": 0.1312623058411726,
126
+ "grad_norm": 1.3915642499923706,
127
+ "learning_rate": 4.373177842565598e-05,
128
+ "loss": 1.189,
129
+ "step": 150
130
+ },
131
+ {
132
+ "epoch": 0.14001312623058412,
133
+ "grad_norm": 1.175917625427246,
134
+ "learning_rate": 4.664723032069971e-05,
135
+ "loss": 1.1605,
136
+ "step": 160
137
+ },
138
+ {
139
+ "epoch": 0.14876394661999562,
140
+ "grad_norm": 1.3125519752502441,
141
+ "learning_rate": 4.956268221574344e-05,
142
+ "loss": 1.1734,
143
+ "step": 170
144
+ },
145
+ {
146
+ "epoch": 0.15751476700940714,
147
+ "grad_norm": 1.2966923713684082,
148
+ "learning_rate": 5.247813411078717e-05,
149
+ "loss": 1.2075,
150
+ "step": 180
151
+ },
152
+ {
153
+ "epoch": 0.16626558739881864,
154
+ "grad_norm": 1.2630099058151245,
155
+ "learning_rate": 5.539358600583091e-05,
156
+ "loss": 1.1791,
157
+ "step": 190
158
+ },
159
+ {
160
+ "epoch": 0.17501640778823013,
161
+ "grad_norm": 1.306665301322937,
162
+ "learning_rate": 5.830903790087464e-05,
163
+ "loss": 1.1469,
164
+ "step": 200
165
+ },
166
+ {
167
+ "epoch": 0.18376722817764166,
168
+ "grad_norm": 1.19467031955719,
169
+ "learning_rate": 6.122448979591838e-05,
170
+ "loss": 1.185,
171
+ "step": 210
172
+ },
173
+ {
174
+ "epoch": 0.19251804856705315,
175
+ "grad_norm": 1.416387677192688,
176
+ "learning_rate": 6.41399416909621e-05,
177
+ "loss": 1.2061,
178
+ "step": 220
179
+ },
180
+ {
181
+ "epoch": 0.20126886895646467,
182
+ "grad_norm": 1.3546264171600342,
183
+ "learning_rate": 6.705539358600583e-05,
184
+ "loss": 1.2016,
185
+ "step": 230
186
+ },
187
+ {
188
+ "epoch": 0.21001968934587617,
189
+ "grad_norm": 1.2417850494384766,
190
+ "learning_rate": 6.997084548104956e-05,
191
+ "loss": 1.1797,
192
+ "step": 240
193
+ },
194
+ {
195
+ "epoch": 0.2187705097352877,
196
+ "grad_norm": 1.2703570127487183,
197
+ "learning_rate": 7.28862973760933e-05,
198
+ "loss": 1.2059,
199
+ "step": 250
200
+ },
201
+ {
202
+ "epoch": 0.2275213301246992,
203
+ "grad_norm": 1.2479429244995117,
204
+ "learning_rate": 7.580174927113704e-05,
205
+ "loss": 1.1958,
206
+ "step": 260
207
+ },
208
+ {
209
+ "epoch": 0.23627215051411068,
210
+ "grad_norm": 1.507601261138916,
211
+ "learning_rate": 7.871720116618077e-05,
212
+ "loss": 1.2151,
213
+ "step": 270
214
+ },
215
+ {
216
+ "epoch": 0.2450229709035222,
217
+ "grad_norm": 1.2083854675292969,
218
+ "learning_rate": 8.163265306122449e-05,
219
+ "loss": 1.2202,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2537737912929337,
224
+ "grad_norm": 1.2939566373825073,
225
+ "learning_rate": 8.454810495626822e-05,
226
+ "loss": 1.1916,
227
+ "step": 290
228
+ },
229
+ {
230
+ "epoch": 0.2625246116823452,
231
+ "grad_norm": 1.2871382236480713,
232
+ "learning_rate": 8.746355685131195e-05,
233
+ "loss": 1.222,
234
+ "step": 300
235
+ },
236
+ {
237
+ "epoch": 0.27127543207175675,
238
+ "grad_norm": 1.1558256149291992,
239
+ "learning_rate": 9.037900874635569e-05,
240
+ "loss": 1.2094,
241
+ "step": 310
242
+ },
243
+ {
244
+ "epoch": 0.28002625246116825,
245
+ "grad_norm": 1.0638883113861084,
246
+ "learning_rate": 9.329446064139942e-05,
247
+ "loss": 1.2,
248
+ "step": 320
249
+ },
250
+ {
251
+ "epoch": 0.28877707285057974,
252
+ "grad_norm": 1.1072458028793335,
253
+ "learning_rate": 9.620991253644317e-05,
254
+ "loss": 1.224,
255
+ "step": 330
256
+ },
257
+ {
258
+ "epoch": 0.29752789323999124,
259
+ "grad_norm": 1.1332110166549683,
260
+ "learning_rate": 9.912536443148688e-05,
261
+ "loss": 1.2266,
262
+ "step": 340
263
+ },
264
+ {
265
+ "epoch": 0.30627871362940273,
266
+ "grad_norm": 1.0537033081054688,
267
+ "learning_rate": 9.999872800049233e-05,
268
+ "loss": 1.2173,
269
+ "step": 350
270
+ },
271
+ {
272
+ "epoch": 0.3150295340188143,
273
+ "grad_norm": 1.090928316116333,
274
+ "learning_rate": 9.999249795462235e-05,
275
+ "loss": 1.2416,
276
+ "step": 360
277
+ },
278
+ {
279
+ "epoch": 0.3237803544082258,
280
+ "grad_norm": 1.0084835290908813,
281
+ "learning_rate": 9.998107687592326e-05,
282
+ "loss": 1.2094,
283
+ "step": 370
284
+ },
285
+ {
286
+ "epoch": 0.3325311747976373,
287
+ "grad_norm": 1.82576584815979,
288
+ "learning_rate": 9.996446595031691e-05,
289
+ "loss": 1.2246,
290
+ "step": 380
291
+ },
292
+ {
293
+ "epoch": 0.34128199518704877,
294
+ "grad_norm": 1.2305560111999512,
295
+ "learning_rate": 9.994266690261928e-05,
296
+ "loss": 1.2647,
297
+ "step": 390
298
+ },
299
+ {
300
+ "epoch": 0.35003281557646027,
301
+ "grad_norm": 1.170067548751831,
302
+ "learning_rate": 9.991568199636148e-05,
303
+ "loss": 1.1829,
304
+ "step": 400
305
+ },
306
+ {
307
+ "epoch": 0.3587836359658718,
308
+ "grad_norm": 1.0299819707870483,
309
+ "learning_rate": 9.988351403355454e-05,
310
+ "loss": 1.2471,
311
+ "step": 410
312
+ },
313
+ {
314
+ "epoch": 0.3675344563552833,
315
+ "grad_norm": 0.9742593765258789,
316
+ "learning_rate": 9.98461663543987e-05,
317
+ "loss": 1.2049,
318
+ "step": 420
319
+ },
320
+ {
321
+ "epoch": 0.3762852767446948,
322
+ "grad_norm": 0.9377045631408691,
323
+ "learning_rate": 9.980364283693638e-05,
324
+ "loss": 1.1963,
325
+ "step": 430
326
+ },
327
+ {
328
+ "epoch": 0.3850360971341063,
329
+ "grad_norm": 1.0425899028778076,
330
+ "learning_rate": 9.97559478966496e-05,
331
+ "loss": 1.2167,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.39378691752351785,
336
+ "grad_norm": 0.9749513864517212,
337
+ "learning_rate": 9.97030864860014e-05,
338
+ "loss": 1.2166,
339
+ "step": 450
340
+ },
341
+ {
342
+ "epoch": 0.40253773791292935,
343
+ "grad_norm": 1.0552780628204346,
344
+ "learning_rate": 9.964506409392176e-05,
345
+ "loss": 1.2228,
346
+ "step": 460
347
+ },
348
+ {
349
+ "epoch": 0.41128855830234085,
350
+ "grad_norm": 1.0261808633804321,
351
+ "learning_rate": 9.958188674523748e-05,
352
+ "loss": 1.2252,
353
+ "step": 470
354
+ },
355
+ {
356
+ "epoch": 0.42003937869175234,
357
+ "grad_norm": 0.9530945420265198,
358
+ "learning_rate": 9.951356100004665e-05,
359
+ "loss": 1.1845,
360
+ "step": 480
361
+ },
362
+ {
363
+ "epoch": 0.42879019908116384,
364
+ "grad_norm": 0.8990481495857239,
365
+ "learning_rate": 9.94400939530375e-05,
366
+ "loss": 1.2283,
367
+ "step": 490
368
+ },
369
+ {
370
+ "epoch": 0.4375410194705754,
371
+ "grad_norm": 0.9713592529296875,
372
+ "learning_rate": 9.936149323275169e-05,
373
+ "loss": 1.1945,
374
+ "step": 500
375
+ },
376
+ {
377
+ "epoch": 0.4462918398599869,
378
+ "grad_norm": 1.003854513168335,
379
+ "learning_rate": 9.927776700079216e-05,
380
+ "loss": 1.2322,
381
+ "step": 510
382
+ },
383
+ {
384
+ "epoch": 0.4550426602493984,
385
+ "grad_norm": 0.935400664806366,
386
+ "learning_rate": 9.918892395097572e-05,
387
+ "loss": 1.2272,
388
+ "step": 520
389
+ },
390
+ {
391
+ "epoch": 0.4637934806388099,
392
+ "grad_norm": 1.075519323348999,
393
+ "learning_rate": 9.90949733084303e-05,
394
+ "loss": 1.2069,
395
+ "step": 530
396
+ },
397
+ {
398
+ "epoch": 0.47254430102822137,
399
+ "grad_norm": 0.9949712157249451,
400
+ "learning_rate": 9.8995924828637e-05,
401
+ "loss": 1.2083,
402
+ "step": 540
403
+ },
404
+ {
405
+ "epoch": 0.4812951214176329,
406
+ "grad_norm": 0.8942195773124695,
407
+ "learning_rate": 9.889178879641717e-05,
408
+ "loss": 1.2137,
409
+ "step": 550
410
+ },
411
+ {
412
+ "epoch": 0.4900459418070444,
413
+ "grad_norm": 0.9525908827781677,
414
+ "learning_rate": 9.878257602486445e-05,
415
+ "loss": 1.2348,
416
+ "step": 560
417
+ },
418
+ {
419
+ "epoch": 0.4987967621964559,
420
+ "grad_norm": 0.895167350769043,
421
+ "learning_rate": 9.8668297854222e-05,
422
+ "loss": 1.2034,
423
+ "step": 570
424
+ },
425
+ {
426
+ "epoch": 0.5075475825858674,
427
+ "grad_norm": 1.0244312286376953,
428
+ "learning_rate": 9.854896615070495e-05,
429
+ "loss": 1.1942,
430
+ "step": 580
431
+ },
432
+ {
433
+ "epoch": 0.516298402975279,
434
+ "grad_norm": 0.9835163950920105,
435
+ "learning_rate": 9.842459330526826e-05,
436
+ "loss": 1.2182,
437
+ "step": 590
438
+ },
439
+ {
440
+ "epoch": 0.5250492233646904,
441
+ "grad_norm": 0.9238565564155579,
442
+ "learning_rate": 9.829519223232006e-05,
443
+ "loss": 1.1992,
444
+ "step": 600
445
+ },
446
+ {
447
+ "epoch": 0.533800043754102,
448
+ "grad_norm": 0.9063233733177185,
449
+ "learning_rate": 9.81607763683808e-05,
450
+ "loss": 1.2135,
451
+ "step": 610
452
+ },
453
+ {
454
+ "epoch": 0.5425508641435135,
455
+ "grad_norm": 0.9810764193534851,
456
+ "learning_rate": 9.802135967068782e-05,
457
+ "loss": 1.1909,
458
+ "step": 620
459
+ },
460
+ {
461
+ "epoch": 0.5513016845329249,
462
+ "grad_norm": 0.9388274550437927,
463
+ "learning_rate": 9.787695661574632e-05,
464
+ "loss": 1.1693,
465
+ "step": 630
466
+ },
467
+ {
468
+ "epoch": 0.5600525049223365,
469
+ "grad_norm": 0.933283805847168,
470
+ "learning_rate": 9.772758219782601e-05,
471
+ "loss": 1.1857,
472
+ "step": 640
473
+ },
474
+ {
475
+ "epoch": 0.5688033253117479,
476
+ "grad_norm": 0.947989284992218,
477
+ "learning_rate": 9.757325192740425e-05,
478
+ "loss": 1.1901,
479
+ "step": 650
480
+ },
481
+ {
482
+ "epoch": 0.5775541457011595,
483
+ "grad_norm": 0.8272686004638672,
484
+ "learning_rate": 9.741398182955548e-05,
485
+ "loss": 1.2087,
486
+ "step": 660
487
+ },
488
+ {
489
+ "epoch": 0.586304966090571,
490
+ "grad_norm": 0.8907813429832458,
491
+ "learning_rate": 9.724978844228717e-05,
492
+ "loss": 1.201,
493
+ "step": 670
494
+ },
495
+ {
496
+ "epoch": 0.5950557864799825,
497
+ "grad_norm": 0.8374966382980347,
498
+ "learning_rate": 9.708068881482271e-05,
499
+ "loss": 1.2028,
500
+ "step": 680
501
+ },
502
+ {
503
+ "epoch": 0.603806606869394,
504
+ "grad_norm": 0.9283761382102966,
505
+ "learning_rate": 9.690670050583095e-05,
506
+ "loss": 1.1995,
507
+ "step": 690
508
+ },
509
+ {
510
+ "epoch": 0.6125574272588055,
511
+ "grad_norm": 0.8573452234268188,
512
+ "learning_rate": 9.672784158160307e-05,
513
+ "loss": 1.1372,
514
+ "step": 700
515
+ },
516
+ {
517
+ "epoch": 0.621308247648217,
518
+ "grad_norm": 0.8831506371498108,
519
+ "learning_rate": 9.654413061417655e-05,
520
+ "loss": 1.1626,
521
+ "step": 710
522
+ },
523
+ {
524
+ "epoch": 0.6300590680376286,
525
+ "grad_norm": 0.9180812239646912,
526
+ "learning_rate": 9.635558667940683e-05,
527
+ "loss": 1.1466,
528
+ "step": 720
529
+ },
530
+ {
531
+ "epoch": 0.63880988842704,
532
+ "grad_norm": 0.8657775521278381,
533
+ "learning_rate": 9.616222935498646e-05,
534
+ "loss": 1.2052,
535
+ "step": 730
536
+ },
537
+ {
538
+ "epoch": 0.6475607088164516,
539
+ "grad_norm": 0.9488115310668945,
540
+ "learning_rate": 9.596407871841229e-05,
541
+ "loss": 1.1651,
542
+ "step": 740
543
+ },
544
+ {
545
+ "epoch": 0.656311529205863,
546
+ "grad_norm": 0.8592752814292908,
547
+ "learning_rate": 9.57611553449006e-05,
548
+ "loss": 1.1953,
549
+ "step": 750
550
+ },
551
+ {
552
+ "epoch": 0.6650623495952745,
553
+ "grad_norm": 0.889445960521698,
554
+ "learning_rate": 9.555348030525076e-05,
555
+ "loss": 1.1647,
556
+ "step": 760
557
+ },
558
+ {
559
+ "epoch": 0.6738131699846861,
560
+ "grad_norm": 0.8997176885604858,
561
+ "learning_rate": 9.534107516365725e-05,
562
+ "loss": 1.167,
563
+ "step": 770
564
+ },
565
+ {
566
+ "epoch": 0.6825639903740975,
567
+ "grad_norm": 0.8208065032958984,
568
+ "learning_rate": 9.512396197547053e-05,
569
+ "loss": 1.1631,
570
+ "step": 780
571
+ },
572
+ {
573
+ "epoch": 0.6913148107635091,
574
+ "grad_norm": 0.8419785499572754,
575
+ "learning_rate": 9.490216328490692e-05,
576
+ "loss": 1.1857,
577
+ "step": 790
578
+ },
579
+ {
580
+ "epoch": 0.7000656311529205,
581
+ "grad_norm": 0.8125676512718201,
582
+ "learning_rate": 9.467570212270767e-05,
583
+ "loss": 1.2426,
584
+ "step": 800
585
+ },
586
+ {
587
+ "epoch": 0.7088164515423321,
588
+ "grad_norm": 0.9222508072853088,
589
+ "learning_rate": 9.444460200374748e-05,
590
+ "loss": 1.1452,
591
+ "step": 810
592
+ },
593
+ {
594
+ "epoch": 0.7175672719317436,
595
+ "grad_norm": 0.8932722210884094,
596
+ "learning_rate": 9.420888692459291e-05,
597
+ "loss": 1.1647,
598
+ "step": 820
599
+ },
600
+ {
601
+ "epoch": 0.7263180923211551,
602
+ "grad_norm": 0.817932665348053,
603
+ "learning_rate": 9.396858136101066e-05,
604
+ "loss": 1.1618,
605
+ "step": 830
606
+ },
607
+ {
608
+ "epoch": 0.7350689127105666,
609
+ "grad_norm": 0.8717408180236816,
610
+ "learning_rate": 9.372371026542595e-05,
611
+ "loss": 1.1517,
612
+ "step": 840
613
+ },
614
+ {
615
+ "epoch": 0.7438197330999782,
616
+ "grad_norm": 0.8578230142593384,
617
+ "learning_rate": 9.347429906433174e-05,
618
+ "loss": 1.2009,
619
+ "step": 850
620
+ },
621
+ {
622
+ "epoch": 0.7525705534893896,
623
+ "grad_norm": 0.8276301622390747,
624
+ "learning_rate": 9.322037365564845e-05,
625
+ "loss": 1.1627,
626
+ "step": 860
627
+ },
628
+ {
629
+ "epoch": 0.7613213738788012,
630
+ "grad_norm": 0.7934173941612244,
631
+ "learning_rate": 9.296196040603475e-05,
632
+ "loss": 1.1735,
633
+ "step": 870
634
+ },
635
+ {
636
+ "epoch": 0.7700721942682126,
637
+ "grad_norm": 0.8270857930183411,
638
+ "learning_rate": 9.269908614814995e-05,
639
+ "loss": 1.149,
640
+ "step": 880
641
+ },
642
+ {
643
+ "epoch": 0.7788230146576242,
644
+ "grad_norm": 1.0632795095443726,
645
+ "learning_rate": 9.243177817786757e-05,
646
+ "loss": 1.1937,
647
+ "step": 890
648
+ },
649
+ {
650
+ "epoch": 0.7875738350470357,
651
+ "grad_norm": 0.8449203968048096,
652
+ "learning_rate": 9.216006425144118e-05,
653
+ "loss": 1.1787,
654
+ "step": 900
655
+ },
656
+ {
657
+ "epoch": 0.7963246554364471,
658
+ "grad_norm": 0.7634311318397522,
659
+ "learning_rate": 9.188397258262221e-05,
660
+ "loss": 1.1445,
661
+ "step": 910
662
+ },
663
+ {
664
+ "epoch": 0.8050754758258587,
665
+ "grad_norm": 0.7856956720352173,
666
+ "learning_rate": 9.160353183973046e-05,
667
+ "loss": 1.1519,
668
+ "step": 920
669
+ },
670
+ {
671
+ "epoch": 0.8138262962152701,
672
+ "grad_norm": 0.8162206411361694,
673
+ "learning_rate": 9.131877114267711e-05,
674
+ "loss": 1.1282,
675
+ "step": 930
676
+ },
677
+ {
678
+ "epoch": 0.8225771166046817,
679
+ "grad_norm": 0.8263719081878662,
680
+ "learning_rate": 9.102972005994122e-05,
681
+ "loss": 1.1127,
682
+ "step": 940
683
+ },
684
+ {
685
+ "epoch": 0.8313279369940932,
686
+ "grad_norm": 0.831969141960144,
687
+ "learning_rate": 9.073640860549931e-05,
688
+ "loss": 1.1446,
689
+ "step": 950
690
+ },
691
+ {
692
+ "epoch": 0.8400787573835047,
693
+ "grad_norm": 0.8127765655517578,
694
+ "learning_rate": 9.043886723570884e-05,
695
+ "loss": 1.1649,
696
+ "step": 960
697
+ },
698
+ {
699
+ "epoch": 0.8488295777729162,
700
+ "grad_norm": 0.7642777562141418,
701
+ "learning_rate": 9.013712684614582e-05,
702
+ "loss": 1.1656,
703
+ "step": 970
704
+ },
705
+ {
706
+ "epoch": 0.8575803981623277,
707
+ "grad_norm": 0.8071634769439697,
708
+ "learning_rate": 8.983121876839657e-05,
709
+ "loss": 1.1414,
710
+ "step": 980
711
+ },
712
+ {
713
+ "epoch": 0.8663312185517392,
714
+ "grad_norm": 0.7789682149887085,
715
+ "learning_rate": 8.95211747668046e-05,
716
+ "loss": 1.1127,
717
+ "step": 990
718
+ },
719
+ {
720
+ "epoch": 0.8750820389411508,
721
+ "grad_norm": 0.7899104356765747,
722
+ "learning_rate": 8.920702703517204e-05,
723
+ "loss": 1.1218,
724
+ "step": 1000
725
+ },
726
+ {
727
+ "epoch": 0.8838328593305622,
728
+ "grad_norm": 0.726874053478241,
729
+ "learning_rate": 8.888880819341702e-05,
730
+ "loss": 1.1345,
731
+ "step": 1010
732
+ },
733
+ {
734
+ "epoch": 0.8925836797199738,
735
+ "grad_norm": 0.8020316958427429,
736
+ "learning_rate": 8.856655128418637e-05,
737
+ "loss": 1.1229,
738
+ "step": 1020
739
+ },
740
+ {
741
+ "epoch": 0.9013345001093852,
742
+ "grad_norm": 0.8466740250587463,
743
+ "learning_rate": 8.824028976942465e-05,
744
+ "loss": 1.1451,
745
+ "step": 1030
746
+ },
747
+ {
748
+ "epoch": 0.9100853204987968,
749
+ "grad_norm": 0.7729008793830872,
750
+ "learning_rate": 8.791005752689958e-05,
751
+ "loss": 1.1204,
752
+ "step": 1040
753
+ },
754
+ {
755
+ "epoch": 0.9188361408882083,
756
+ "grad_norm": 0.8055347204208374,
757
+ "learning_rate": 8.757588884668434e-05,
758
+ "loss": 1.1372,
759
+ "step": 1050
760
+ },
761
+ {
762
+ "epoch": 0.9275869612776197,
763
+ "grad_norm": 0.8180489540100098,
764
+ "learning_rate": 8.723781842759695e-05,
765
+ "loss": 1.1632,
766
+ "step": 1060
767
+ },
768
+ {
769
+ "epoch": 0.9363377816670313,
770
+ "grad_norm": 0.816849946975708,
771
+ "learning_rate": 8.689588137359732e-05,
772
+ "loss": 1.1226,
773
+ "step": 1070
774
+ },
775
+ {
776
+ "epoch": 0.9450886020564427,
777
+ "grad_norm": 0.8028609752655029,
778
+ "learning_rate": 8.655011319014216e-05,
779
+ "loss": 1.1559,
780
+ "step": 1080
781
+ },
782
+ {
783
+ "epoch": 0.9538394224458543,
784
+ "grad_norm": 0.8719303011894226,
785
+ "learning_rate": 8.620054978049829e-05,
786
+ "loss": 1.1522,
787
+ "step": 1090
788
+ },
789
+ {
790
+ "epoch": 0.9625902428352658,
791
+ "grad_norm": 0.766177237033844,
792
+ "learning_rate": 8.584722744201447e-05,
793
+ "loss": 1.1186,
794
+ "step": 1100
795
+ },
796
+ {
797
+ "epoch": 0.9713410632246773,
798
+ "grad_norm": 0.7790910005569458,
799
+ "learning_rate": 8.549018286235255e-05,
800
+ "loss": 1.1227,
801
+ "step": 1110
802
+ },
803
+ {
804
+ "epoch": 0.9800918836140888,
805
+ "grad_norm": 0.7803717255592346,
806
+ "learning_rate": 8.512945311567786e-05,
807
+ "loss": 1.1186,
808
+ "step": 1120
809
+ },
810
+ {
811
+ "epoch": 0.9888427040035004,
812
+ "grad_norm": 0.7084935307502747,
813
+ "learning_rate": 8.476507565880957e-05,
814
+ "loss": 1.1085,
815
+ "step": 1130
816
+ },
817
+ {
818
+ "epoch": 0.9975935243929118,
819
+ "grad_norm": 0.7696670889854431,
820
+ "learning_rate": 8.439708832733142e-05,
821
+ "loss": 1.1123,
822
+ "step": 1140
823
+ },
824
+ {
825
+ "epoch": 1.006125574272588,
826
+ "grad_norm": 0.9948705434799194,
827
+ "learning_rate": 8.402552933166288e-05,
828
+ "loss": 0.9984,
829
+ "step": 1150
830
+ },
831
+ {
832
+ "epoch": 1.0148763946619996,
833
+ "grad_norm": 0.8488800525665283,
834
+ "learning_rate": 8.36504372530916e-05,
835
+ "loss": 0.9603,
836
+ "step": 1160
837
+ },
838
+ {
839
+ "epoch": 1.023627215051411,
840
+ "grad_norm": 0.8748326897621155,
841
+ "learning_rate": 8.32718510397673e-05,
842
+ "loss": 0.9535,
843
+ "step": 1170
844
+ },
845
+ {
846
+ "epoch": 1.0323780354408225,
847
+ "grad_norm": 0.8636799454689026,
848
+ "learning_rate": 8.288981000265748e-05,
849
+ "loss": 0.9705,
850
+ "step": 1180
851
+ },
852
+ {
853
+ "epoch": 1.041128855830234,
854
+ "grad_norm": 0.7722444534301758,
855
+ "learning_rate": 8.250435381146549e-05,
856
+ "loss": 0.9411,
857
+ "step": 1190
858
+ },
859
+ {
860
+ "epoch": 1.0498796762196456,
861
+ "grad_norm": 0.792931318283081,
862
+ "learning_rate": 8.211552249051149e-05,
863
+ "loss": 0.9455,
864
+ "step": 1200
865
+ },
866
+ {
867
+ "epoch": 1.0586304966090572,
868
+ "grad_norm": 0.8179608583450317,
869
+ "learning_rate": 8.17233564145764e-05,
870
+ "loss": 0.9418,
871
+ "step": 1210
872
+ },
873
+ {
874
+ "epoch": 1.0673813169984685,
875
+ "grad_norm": 0.8081513047218323,
876
+ "learning_rate": 8.132789630470952e-05,
877
+ "loss": 0.9451,
878
+ "step": 1220
879
+ },
880
+ {
881
+ "epoch": 1.07613213738788,
882
+ "grad_norm": 0.8941613435745239,
883
+ "learning_rate": 8.092918322400027e-05,
884
+ "loss": 0.9336,
885
+ "step": 1230
886
+ },
887
+ {
888
+ "epoch": 1.0848829577772916,
889
+ "grad_norm": 0.8505945801734924,
890
+ "learning_rate": 8.052725857331429e-05,
891
+ "loss": 0.9432,
892
+ "step": 1240
893
+ },
894
+ {
895
+ "epoch": 1.0936337781667032,
896
+ "grad_norm": 0.7953508496284485,
897
+ "learning_rate": 8.012216408699455e-05,
898
+ "loss": 0.9608,
899
+ "step": 1250
900
+ },
901
+ {
902
+ "epoch": 1.1023845985561147,
903
+ "grad_norm": 0.9065775871276855,
904
+ "learning_rate": 7.971394182852789e-05,
905
+ "loss": 0.9583,
906
+ "step": 1260
907
+ },
908
+ {
909
+ "epoch": 1.1111354189455263,
910
+ "grad_norm": 0.8752979040145874,
911
+ "learning_rate": 7.930263418617713e-05,
912
+ "loss": 0.9752,
913
+ "step": 1270
914
+ },
915
+ {
916
+ "epoch": 1.1198862393349376,
917
+ "grad_norm": 0.8173210024833679,
918
+ "learning_rate": 7.888828386857985e-05,
919
+ "loss": 0.9505,
920
+ "step": 1280
921
+ },
922
+ {
923
+ "epoch": 1.1286370597243491,
924
+ "grad_norm": 0.7964232563972473,
925
+ "learning_rate": 7.84709339003135e-05,
926
+ "loss": 0.9346,
927
+ "step": 1290
928
+ },
929
+ {
930
+ "epoch": 1.1373878801137607,
931
+ "grad_norm": 0.8364039063453674,
932
+ "learning_rate": 7.805062761742799e-05,
933
+ "loss": 0.9606,
934
+ "step": 1300
935
+ },
936
+ {
937
+ "epoch": 1.1461387005031722,
938
+ "grad_norm": 0.8122743964195251,
939
+ "learning_rate": 7.76274086629458e-05,
940
+ "loss": 0.9455,
941
+ "step": 1310
942
+ },
943
+ {
944
+ "epoch": 1.1548895208925836,
945
+ "grad_norm": 0.8229530453681946,
946
+ "learning_rate": 7.720132098233031e-05,
947
+ "loss": 0.9465,
948
+ "step": 1320
949
+ },
950
+ {
951
+ "epoch": 1.1636403412819951,
952
+ "grad_norm": 0.7859545946121216,
953
+ "learning_rate": 7.677240881892258e-05,
954
+ "loss": 0.9421,
955
+ "step": 1330
956
+ },
957
+ {
958
+ "epoch": 1.1723911616714067,
959
+ "grad_norm": 0.8277194499969482,
960
+ "learning_rate": 7.634071670934734e-05,
961
+ "loss": 0.9541,
962
+ "step": 1340
963
+ },
964
+ {
965
+ "epoch": 1.1811419820608182,
966
+ "grad_norm": 0.7695391774177551,
967
+ "learning_rate": 7.590628947888847e-05,
968
+ "loss": 0.9433,
969
+ "step": 1350
970
+ },
971
+ {
972
+ "epoch": 1.1898928024502298,
973
+ "grad_norm": 0.82956862449646,
974
+ "learning_rate": 7.546917223683453e-05,
975
+ "loss": 0.9395,
976
+ "step": 1360
977
+ },
978
+ {
979
+ "epoch": 1.1986436228396413,
980
+ "grad_norm": 0.8398957252502441,
981
+ "learning_rate": 7.502941037179474e-05,
982
+ "loss": 0.9432,
983
+ "step": 1370
984
+ },
985
+ {
986
+ "epoch": 1.2073944432290526,
987
+ "grad_norm": 0.8232219219207764,
988
+ "learning_rate": 7.4587049546986e-05,
989
+ "loss": 0.9887,
990
+ "step": 1380
991
+ },
992
+ {
993
+ "epoch": 1.2161452636184642,
994
+ "grad_norm": 0.7886401414871216,
995
+ "learning_rate": 7.414213569549145e-05,
996
+ "loss": 0.959,
997
+ "step": 1390
998
+ },
999
+ {
1000
+ "epoch": 1.2248960840078758,
1001
+ "grad_norm": 0.7588061690330505,
1002
+ "learning_rate": 7.369471501549087e-05,
1003
+ "loss": 0.9651,
1004
+ "step": 1400
1005
+ },
1006
+ {
1007
+ "epoch": 1.2336469043972873,
1008
+ "grad_norm": 0.8234547972679138,
1009
+ "learning_rate": 7.324483396546371e-05,
1010
+ "loss": 0.9935,
1011
+ "step": 1410
1012
+ },
1013
+ {
1014
+ "epoch": 1.2423977247866986,
1015
+ "grad_norm": 0.7623049020767212,
1016
+ "learning_rate": 7.279253925936498e-05,
1017
+ "loss": 0.95,
1018
+ "step": 1420
1019
+ },
1020
+ {
1021
+ "epoch": 1.2511485451761102,
1022
+ "grad_norm": 0.7172533869743347,
1023
+ "learning_rate": 7.233787786177464e-05,
1024
+ "loss": 0.9441,
1025
+ "step": 1430
1026
+ },
1027
+ {
1028
+ "epoch": 1.2598993655655217,
1029
+ "grad_norm": 0.7973024249076843,
1030
+ "learning_rate": 7.188089698302098e-05,
1031
+ "loss": 0.9682,
1032
+ "step": 1440
1033
+ },
1034
+ {
1035
+ "epoch": 1.2686501859549333,
1036
+ "grad_norm": 0.7969169020652771,
1037
+ "learning_rate": 7.14216440742785e-05,
1038
+ "loss": 0.9541,
1039
+ "step": 1450
1040
+ },
1041
+ {
1042
+ "epoch": 1.2774010063443448,
1043
+ "grad_norm": 0.8296008110046387,
1044
+ "learning_rate": 7.096016682264074e-05,
1045
+ "loss": 0.9597,
1046
+ "step": 1460
1047
+ },
1048
+ {
1049
+ "epoch": 1.2861518267337564,
1050
+ "grad_norm": 0.8538773655891418,
1051
+ "learning_rate": 7.049651314616862e-05,
1052
+ "loss": 0.9225,
1053
+ "step": 1470
1054
+ },
1055
+ {
1056
+ "epoch": 1.2949026471231677,
1057
+ "grad_norm": 0.8068443536758423,
1058
+ "learning_rate": 7.003073118891487e-05,
1059
+ "loss": 0.9282,
1060
+ "step": 1480
1061
+ },
1062
+ {
1063
+ "epoch": 1.3036534675125793,
1064
+ "grad_norm": 0.8286350965499878,
1065
+ "learning_rate": 6.956286931592481e-05,
1066
+ "loss": 0.9645,
1067
+ "step": 1490
1068
+ },
1069
+ {
1070
+ "epoch": 1.3124042879019908,
1071
+ "grad_norm": 0.796650767326355,
1072
+ "learning_rate": 6.909297610821443e-05,
1073
+ "loss": 0.934,
1074
+ "step": 1500
1075
+ },
1076
+ {
1077
+ "epoch": 1.3211551082914024,
1078
+ "grad_norm": 0.7479636073112488,
1079
+ "learning_rate": 6.862110035772589e-05,
1080
+ "loss": 0.9319,
1081
+ "step": 1510
1082
+ },
1083
+ {
1084
+ "epoch": 1.3299059286808137,
1085
+ "grad_norm": 0.7734546065330505,
1086
+ "learning_rate": 6.81472910622611e-05,
1087
+ "loss": 0.9577,
1088
+ "step": 1520
1089
+ },
1090
+ {
1091
+ "epoch": 1.3386567490702252,
1092
+ "grad_norm": 0.821553647518158,
1093
+ "learning_rate": 6.7671597420394e-05,
1094
+ "loss": 0.966,
1095
+ "step": 1530
1096
+ },
1097
+ {
1098
+ "epoch": 1.3474075694596368,
1099
+ "grad_norm": 0.8141497373580933,
1100
+ "learning_rate": 6.719406882636196e-05,
1101
+ "loss": 0.9593,
1102
+ "step": 1540
1103
+ },
1104
+ {
1105
+ "epoch": 1.3561583898490484,
1106
+ "grad_norm": 0.7738481163978577,
1107
+ "learning_rate": 6.671475486493691e-05,
1108
+ "loss": 0.9474,
1109
+ "step": 1550
1110
+ },
1111
+ {
1112
+ "epoch": 1.36490921023846,
1113
+ "grad_norm": 0.7933306694030762,
1114
+ "learning_rate": 6.62337053062766e-05,
1115
+ "loss": 0.9297,
1116
+ "step": 1560
1117
+ },
1118
+ {
1119
+ "epoch": 1.3736600306278715,
1120
+ "grad_norm": 0.7921035885810852,
1121
+ "learning_rate": 6.575097010075664e-05,
1122
+ "loss": 0.9283,
1123
+ "step": 1570
1124
+ },
1125
+ {
1126
+ "epoch": 1.382410851017283,
1127
+ "grad_norm": 0.7895304560661316,
1128
+ "learning_rate": 6.526659937378392e-05,
1129
+ "loss": 0.9559,
1130
+ "step": 1580
1131
+ },
1132
+ {
1133
+ "epoch": 1.3911616714066943,
1134
+ "grad_norm": 0.8518770933151245,
1135
+ "learning_rate": 6.478064342059164e-05,
1136
+ "loss": 0.9571,
1137
+ "step": 1590
1138
+ },
1139
+ {
1140
+ "epoch": 1.3999124917961059,
1141
+ "grad_norm": 0.8479484915733337,
1142
+ "learning_rate": 6.429315270101701e-05,
1143
+ "loss": 0.9578,
1144
+ "step": 1600
1145
+ },
1146
+ {
1147
+ "epoch": 1.4086633121855174,
1148
+ "grad_norm": 0.763520359992981,
1149
+ "learning_rate": 6.380417783426153e-05,
1150
+ "loss": 0.9304,
1151
+ "step": 1610
1152
+ },
1153
+ {
1154
+ "epoch": 1.4174141325749288,
1155
+ "grad_norm": 0.7657537460327148,
1156
+ "learning_rate": 6.3313769593635e-05,
1157
+ "loss": 0.9541,
1158
+ "step": 1620
1159
+ },
1160
+ {
1161
+ "epoch": 1.4261649529643403,
1162
+ "grad_norm": 0.7461577653884888,
1163
+ "learning_rate": 6.282197890128338e-05,
1164
+ "loss": 0.9331,
1165
+ "step": 1630
1166
+ },
1167
+ {
1168
+ "epoch": 1.4349157733537519,
1169
+ "grad_norm": 0.7814643383026123,
1170
+ "learning_rate": 6.23288568229012e-05,
1171
+ "loss": 0.9431,
1172
+ "step": 1640
1173
+ },
1174
+ {
1175
+ "epoch": 1.4436665937431634,
1176
+ "grad_norm": 0.7600048184394836,
1177
+ "learning_rate": 6.183445456242904e-05,
1178
+ "loss": 0.9549,
1179
+ "step": 1650
1180
+ },
1181
+ {
1182
+ "epoch": 1.452417414132575,
1183
+ "grad_norm": 0.7684770226478577,
1184
+ "learning_rate": 6.13388234567369e-05,
1185
+ "loss": 0.9074,
1186
+ "step": 1660
1187
+ },
1188
+ {
1189
+ "epoch": 1.4611682345219865,
1190
+ "grad_norm": 0.7696465849876404,
1191
+ "learning_rate": 6.084201497029333e-05,
1192
+ "loss": 0.9449,
1193
+ "step": 1670
1194
+ },
1195
+ {
1196
+ "epoch": 1.469919054911398,
1197
+ "grad_norm": 0.818397045135498,
1198
+ "learning_rate": 6.034408068982172e-05,
1199
+ "loss": 0.9396,
1200
+ "step": 1680
1201
+ },
1202
+ {
1203
+ "epoch": 1.4786698753008094,
1204
+ "grad_norm": 0.7443113923072815,
1205
+ "learning_rate": 5.984507231894364e-05,
1206
+ "loss": 0.928,
1207
+ "step": 1690
1208
+ },
1209
+ {
1210
+ "epoch": 1.487420695690221,
1211
+ "grad_norm": 0.8166981339454651,
1212
+ "learning_rate": 5.934504167281022e-05,
1213
+ "loss": 0.9323,
1214
+ "step": 1700
1215
+ },
1216
+ {
1217
+ "epoch": 1.4961715160796325,
1218
+ "grad_norm": 0.8331442475318909,
1219
+ "learning_rate": 5.884404067272174e-05,
1220
+ "loss": 0.9458,
1221
+ "step": 1710
1222
+ },
1223
+ {
1224
+ "epoch": 1.5049223364690438,
1225
+ "grad_norm": 0.7990359663963318,
1226
+ "learning_rate": 5.834212134073644e-05,
1227
+ "loss": 0.9094,
1228
+ "step": 1720
1229
+ },
1230
+ {
1231
+ "epoch": 1.5136731568584554,
1232
+ "grad_norm": 0.768055260181427,
1233
+ "learning_rate": 5.7839335794268666e-05,
1234
+ "loss": 0.9385,
1235
+ "step": 1730
1236
+ },
1237
+ {
1238
+ "epoch": 1.522423977247867,
1239
+ "grad_norm": 0.7425386309623718,
1240
+ "learning_rate": 5.7335736240677174e-05,
1241
+ "loss": 0.9446,
1242
+ "step": 1740
1243
+ },
1244
+ {
1245
+ "epoch": 1.5311747976372785,
1246
+ "grad_norm": 0.7698354721069336,
1247
+ "learning_rate": 5.683137497184424e-05,
1248
+ "loss": 0.9057,
1249
+ "step": 1750
1250
+ },
1251
+ {
1252
+ "epoch": 1.53992561802669,
1253
+ "grad_norm": 0.8186885118484497,
1254
+ "learning_rate": 5.6326304358745694e-05,
1255
+ "loss": 0.9758,
1256
+ "step": 1760
1257
+ },
1258
+ {
1259
+ "epoch": 1.5486764384161016,
1260
+ "grad_norm": 0.8006178736686707,
1261
+ "learning_rate": 5.5820576846013086e-05,
1262
+ "loss": 0.9424,
1263
+ "step": 1770
1264
+ },
1265
+ {
1266
+ "epoch": 1.5574272588055131,
1267
+ "grad_norm": 0.7572160363197327,
1268
+ "learning_rate": 5.531424494648789e-05,
1269
+ "loss": 0.945,
1270
+ "step": 1780
1271
+ },
1272
+ {
1273
+ "epoch": 1.5661780791949247,
1274
+ "grad_norm": 0.8019212484359741,
1275
+ "learning_rate": 5.480736123576886e-05,
1276
+ "loss": 0.9528,
1277
+ "step": 1790
1278
+ },
1279
+ {
1280
+ "epoch": 1.574928899584336,
1281
+ "grad_norm": 0.7867367267608643,
1282
+ "learning_rate": 5.4299978346752756e-05,
1283
+ "loss": 0.9499,
1284
+ "step": 1800
1285
+ },
1286
+ {
1287
+ "epoch": 1.5836797199737476,
1288
+ "grad_norm": 0.7404739856719971,
1289
+ "learning_rate": 5.379214896416907e-05,
1290
+ "loss": 0.9177,
1291
+ "step": 1810
1292
+ },
1293
+ {
1294
+ "epoch": 1.592430540363159,
1295
+ "grad_norm": 0.712374746799469,
1296
+ "learning_rate": 5.328392581910961e-05,
1297
+ "loss": 0.9212,
1298
+ "step": 1820
1299
+ },
1300
+ {
1301
+ "epoch": 1.6011813607525704,
1302
+ "grad_norm": 0.7724602222442627,
1303
+ "learning_rate": 5.277536168355293e-05,
1304
+ "loss": 0.9402,
1305
+ "step": 1830
1306
+ },
1307
+ {
1308
+ "epoch": 1.609932181141982,
1309
+ "grad_norm": 0.8452624678611755,
1310
+ "learning_rate": 5.226650936488474e-05,
1311
+ "loss": 0.9193,
1312
+ "step": 1840
1313
+ },
1314
+ {
1315
+ "epoch": 1.6186830015313936,
1316
+ "grad_norm": 0.7993892431259155,
1317
+ "learning_rate": 5.1757421700414645e-05,
1318
+ "loss": 0.9328,
1319
+ "step": 1850
1320
+ },
1321
+ {
1322
+ "epoch": 1.627433821920805,
1323
+ "grad_norm": 0.7931053042411804,
1324
+ "learning_rate": 5.124815155188966e-05,
1325
+ "loss": 0.9759,
1326
+ "step": 1860
1327
+ },
1328
+ {
1329
+ "epoch": 1.6361846423102167,
1330
+ "grad_norm": 0.7516879439353943,
1331
+ "learning_rate": 5.0738751800005226e-05,
1332
+ "loss": 0.9229,
1333
+ "step": 1870
1334
+ },
1335
+ {
1336
+ "epoch": 1.6449354626996282,
1337
+ "grad_norm": 0.6935803890228271,
1338
+ "learning_rate": 5.022927533891434e-05,
1339
+ "loss": 0.9234,
1340
+ "step": 1880
1341
+ },
1342
+ {
1343
+ "epoch": 1.6536862830890398,
1344
+ "grad_norm": 0.8120819926261902,
1345
+ "learning_rate": 4.9719775070735196e-05,
1346
+ "loss": 0.9582,
1347
+ "step": 1890
1348
+ },
1349
+ {
1350
+ "epoch": 1.662437103478451,
1351
+ "grad_norm": 0.781491219997406,
1352
+ "learning_rate": 4.921030390005802e-05,
1353
+ "loss": 0.9177,
1354
+ "step": 1900
1355
+ },
1356
+ {
1357
+ "epoch": 1.6711879238678626,
1358
+ "grad_norm": 0.7723197340965271,
1359
+ "learning_rate": 4.870091472845166e-05,
1360
+ "loss": 0.9382,
1361
+ "step": 1910
1362
+ },
1363
+ {
1364
+ "epoch": 1.679938744257274,
1365
+ "grad_norm": 0.751848578453064,
1366
+ "learning_rate": 4.819166044897047e-05,
1367
+ "loss": 0.9148,
1368
+ "step": 1920
1369
+ },
1370
+ {
1371
+ "epoch": 1.6886895646466855,
1372
+ "grad_norm": 0.8265687823295593,
1373
+ "learning_rate": 4.768259394066216e-05,
1374
+ "loss": 0.9386,
1375
+ "step": 1930
1376
+ },
1377
+ {
1378
+ "epoch": 1.697440385036097,
1379
+ "grad_norm": 0.7489150762557983,
1380
+ "learning_rate": 4.7173768063076956e-05,
1381
+ "loss": 0.9337,
1382
+ "step": 1940
1383
+ },
1384
+ {
1385
+ "epoch": 1.7061912054255086,
1386
+ "grad_norm": 0.8112419247627258,
1387
+ "learning_rate": 4.666523565077889e-05,
1388
+ "loss": 0.917,
1389
+ "step": 1950
1390
+ },
1391
+ {
1392
+ "epoch": 1.7149420258149202,
1393
+ "grad_norm": 0.7611903548240662,
1394
+ "learning_rate": 4.615704950785965e-05,
1395
+ "loss": 0.9338,
1396
+ "step": 1960
1397
+ },
1398
+ {
1399
+ "epoch": 1.7236928462043317,
1400
+ "grad_norm": 0.7738552093505859,
1401
+ "learning_rate": 4.5649262402455704e-05,
1402
+ "loss": 0.9359,
1403
+ "step": 1970
1404
+ },
1405
+ {
1406
+ "epoch": 1.7324436665937433,
1407
+ "grad_norm": 0.7741129994392395,
1408
+ "learning_rate": 4.5141927061268844e-05,
1409
+ "loss": 0.9209,
1410
+ "step": 1980
1411
+ },
1412
+ {
1413
+ "epoch": 1.7411944869831548,
1414
+ "grad_norm": 0.7815123200416565,
1415
+ "learning_rate": 4.463509616409144e-05,
1416
+ "loss": 0.916,
1417
+ "step": 1990
1418
+ },
1419
+ {
1420
+ "epoch": 1.7499453073725662,
1421
+ "grad_norm": 0.6793907284736633,
1422
+ "learning_rate": 4.412882233833629e-05,
1423
+ "loss": 0.9042,
1424
+ "step": 2000
1425
+ },
1426
+ {
1427
+ "epoch": 1.7586961277619777,
1428
+ "grad_norm": 0.711877167224884,
1429
+ "learning_rate": 4.362315815357197e-05,
1430
+ "loss": 0.9072,
1431
+ "step": 2010
1432
+ },
1433
+ {
1434
+ "epoch": 1.767446948151389,
1435
+ "grad_norm": 0.7671152353286743,
1436
+ "learning_rate": 4.311815611606423e-05,
1437
+ "loss": 0.9226,
1438
+ "step": 2020
1439
+ },
1440
+ {
1441
+ "epoch": 1.7761977685408006,
1442
+ "grad_norm": 0.7221760153770447,
1443
+ "learning_rate": 4.261386866332397e-05,
1444
+ "loss": 0.9077,
1445
+ "step": 2030
1446
+ },
1447
+ {
1448
+ "epoch": 1.7849485889302121,
1449
+ "grad_norm": 0.7419185042381287,
1450
+ "learning_rate": 4.211034815866228e-05,
1451
+ "loss": 0.9017,
1452
+ "step": 2040
1453
+ },
1454
+ {
1455
+ "epoch": 1.7936994093196237,
1456
+ "grad_norm": 0.740652859210968,
1457
+ "learning_rate": 4.1607646885753206e-05,
1458
+ "loss": 0.9159,
1459
+ "step": 2050
1460
+ },
1461
+ {
1462
+ "epoch": 1.8024502297090352,
1463
+ "grad_norm": 0.7923645377159119,
1464
+ "learning_rate": 4.110581704320495e-05,
1465
+ "loss": 0.9135,
1466
+ "step": 2060
1467
+ },
1468
+ {
1469
+ "epoch": 1.8112010500984468,
1470
+ "grad_norm": 0.7528581023216248,
1471
+ "learning_rate": 4.060491073913957e-05,
1472
+ "loss": 0.9291,
1473
+ "step": 2070
1474
+ },
1475
+ {
1476
+ "epoch": 1.8199518704878583,
1477
+ "grad_norm": 0.7974929809570312,
1478
+ "learning_rate": 4.010497998578239e-05,
1479
+ "loss": 0.9208,
1480
+ "step": 2080
1481
+ },
1482
+ {
1483
+ "epoch": 1.8287026908772699,
1484
+ "grad_norm": 0.7388021349906921,
1485
+ "learning_rate": 3.960607669406129e-05,
1486
+ "loss": 0.9155,
1487
+ "step": 2090
1488
+ },
1489
+ {
1490
+ "epoch": 1.8374535112666812,
1491
+ "grad_norm": 0.7927631735801697,
1492
+ "learning_rate": 3.910825266821629e-05,
1493
+ "loss": 0.9156,
1494
+ "step": 2100
1495
+ },
1496
+ {
1497
+ "epoch": 1.8462043316560928,
1498
+ "grad_norm": 0.7545862793922424,
1499
+ "learning_rate": 3.861155960042057e-05,
1500
+ "loss": 0.877,
1501
+ "step": 2110
1502
+ },
1503
+ {
1504
+ "epoch": 1.8549551520455043,
1505
+ "grad_norm": 0.8433863520622253,
1506
+ "learning_rate": 3.8116049065412864e-05,
1507
+ "loss": 0.9107,
1508
+ "step": 2120
1509
+ },
1510
+ {
1511
+ "epoch": 1.8637059724349156,
1512
+ "grad_norm": 0.7610001564025879,
1513
+ "learning_rate": 3.762177251514218e-05,
1514
+ "loss": 0.9068,
1515
+ "step": 2130
1516
+ },
1517
+ {
1518
+ "epoch": 1.8724567928243272,
1519
+ "grad_norm": 0.7452635765075684,
1520
+ "learning_rate": 3.712878127342515e-05,
1521
+ "loss": 0.9113,
1522
+ "step": 2140
1523
+ },
1524
+ {
1525
+ "epoch": 1.8812076132137387,
1526
+ "grad_norm": 0.7530847787857056,
1527
+ "learning_rate": 3.663712653061682e-05,
1528
+ "loss": 0.8881,
1529
+ "step": 2150
1530
+ },
1531
+ {
1532
+ "epoch": 1.8899584336031503,
1533
+ "grad_norm": 0.7940835356712341,
1534
+ "learning_rate": 3.614685933829525e-05,
1535
+ "loss": 0.8807,
1536
+ "step": 2160
1537
+ },
1538
+ {
1539
+ "epoch": 1.8987092539925619,
1540
+ "grad_norm": 0.7291942834854126,
1541
+ "learning_rate": 3.5658030603960436e-05,
1542
+ "loss": 0.9112,
1543
+ "step": 2170
1544
+ },
1545
+ {
1546
+ "epoch": 1.9074600743819734,
1547
+ "grad_norm": 0.7321741580963135,
1548
+ "learning_rate": 3.5170691085748324e-05,
1549
+ "loss": 0.8835,
1550
+ "step": 2180
1551
+ },
1552
+ {
1553
+ "epoch": 1.916210894771385,
1554
+ "grad_norm": 0.7166718244552612,
1555
+ "learning_rate": 3.468489138716029e-05,
1556
+ "loss": 0.9115,
1557
+ "step": 2190
1558
+ },
1559
+ {
1560
+ "epoch": 1.9249617151607963,
1561
+ "grad_norm": 0.741908848285675,
1562
+ "learning_rate": 3.42006819518086e-05,
1563
+ "loss": 0.9198,
1564
+ "step": 2200
1565
+ },
1566
+ {
1567
+ "epoch": 1.9337125355502078,
1568
+ "grad_norm": 0.8034552931785583,
1569
+ "learning_rate": 3.371811305817858e-05,
1570
+ "loss": 0.8952,
1571
+ "step": 2210
1572
+ },
1573
+ {
1574
+ "epoch": 1.9424633559396194,
1575
+ "grad_norm": 0.6716712713241577,
1576
+ "learning_rate": 3.32372348144079e-05,
1577
+ "loss": 0.9197,
1578
+ "step": 2220
1579
+ },
1580
+ {
1581
+ "epoch": 1.9512141763290307,
1582
+ "grad_norm": 0.6731935739517212,
1583
+ "learning_rate": 3.275809715308349e-05,
1584
+ "loss": 0.8949,
1585
+ "step": 2230
1586
+ },
1587
+ {
1588
+ "epoch": 1.9599649967184423,
1589
+ "grad_norm": 0.7521949410438538,
1590
+ "learning_rate": 3.228074982605673e-05,
1591
+ "loss": 0.8949,
1592
+ "step": 2240
1593
+ },
1594
+ {
1595
+ "epoch": 1.9687158171078538,
1596
+ "grad_norm": 0.6889147162437439,
1597
+ "learning_rate": 3.180524239927749e-05,
1598
+ "loss": 0.8941,
1599
+ "step": 2250
1600
+ },
1601
+ {
1602
+ "epoch": 1.9774666374972654,
1603
+ "grad_norm": 0.7169541716575623,
1604
+ "learning_rate": 3.1331624247647276e-05,
1605
+ "loss": 0.8922,
1606
+ "step": 2260
1607
+ },
1608
+ {
1609
+ "epoch": 1.986217457886677,
1610
+ "grad_norm": 0.7812347412109375,
1611
+ "learning_rate": 3.0859944549892325e-05,
1612
+ "loss": 0.8915,
1613
+ "step": 2270
1614
+ },
1615
+ {
1616
+ "epoch": 1.9949682782760885,
1617
+ "grad_norm": 0.7609400153160095,
1618
+ "learning_rate": 3.039025228345725e-05,
1619
+ "loss": 0.9227,
1620
+ "step": 2280
1621
+ },
1622
+ {
1623
+ "epoch": 2.0035003281557646,
1624
+ "grad_norm": 0.7317432165145874,
1625
+ "learning_rate": 2.9922596219419158e-05,
1626
+ "loss": 0.8076,
1627
+ "step": 2290
1628
+ },
1629
+ {
1630
+ "epoch": 2.012251148545176,
1631
+ "grad_norm": 0.8811588883399963,
1632
+ "learning_rate": 2.9457024917423593e-05,
1633
+ "loss": 0.6705,
1634
+ "step": 2300
1635
+ },
1636
+ {
1637
+ "epoch": 2.0210019689345877,
1638
+ "grad_norm": 0.856757640838623,
1639
+ "learning_rate": 2.8993586720642307e-05,
1640
+ "loss": 0.6602,
1641
+ "step": 2310
1642
+ },
1643
+ {
1644
+ "epoch": 2.0297527893239993,
1645
+ "grad_norm": 0.791458249092102,
1646
+ "learning_rate": 2.85323297507534e-05,
1647
+ "loss": 0.6417,
1648
+ "step": 2320
1649
+ },
1650
+ {
1651
+ "epoch": 2.038503609713411,
1652
+ "grad_norm": 0.8533851504325867,
1653
+ "learning_rate": 2.8073301902944515e-05,
1654
+ "loss": 0.6673,
1655
+ "step": 2330
1656
+ },
1657
+ {
1658
+ "epoch": 2.047254430102822,
1659
+ "grad_norm": 0.8373380899429321,
1660
+ "learning_rate": 2.7616550840939757e-05,
1661
+ "loss": 0.6674,
1662
+ "step": 2340
1663
+ },
1664
+ {
1665
+ "epoch": 2.0560052504922335,
1666
+ "grad_norm": 0.8470285534858704,
1667
+ "learning_rate": 2.7162123992050316e-05,
1668
+ "loss": 0.6611,
1669
+ "step": 2350
1670
+ },
1671
+ {
1672
+ "epoch": 2.064756070881645,
1673
+ "grad_norm": 0.8354354500770569,
1674
+ "learning_rate": 2.671006854224982e-05,
1675
+ "loss": 0.6449,
1676
+ "step": 2360
1677
+ },
1678
+ {
1679
+ "epoch": 2.0735068912710566,
1680
+ "grad_norm": 0.8518164157867432,
1681
+ "learning_rate": 2.626043143127477e-05,
1682
+ "loss": 0.6387,
1683
+ "step": 2370
1684
+ },
1685
+ {
1686
+ "epoch": 2.082257711660468,
1687
+ "grad_norm": 0.9259853959083557,
1688
+ "learning_rate": 2.5813259347750447e-05,
1689
+ "loss": 0.6864,
1690
+ "step": 2380
1691
+ },
1692
+ {
1693
+ "epoch": 2.0910085320498797,
1694
+ "grad_norm": 0.8142094612121582,
1695
+ "learning_rate": 2.5368598724342974e-05,
1696
+ "loss": 0.6668,
1697
+ "step": 2390
1698
+ },
1699
+ {
1700
+ "epoch": 2.0997593524392912,
1701
+ "grad_norm": 0.8513460755348206,
1702
+ "learning_rate": 2.4926495732937876e-05,
1703
+ "loss": 0.6588,
1704
+ "step": 2400
1705
+ },
1706
+ {
1707
+ "epoch": 2.108510172828703,
1708
+ "grad_norm": 0.8699485063552856,
1709
+ "learning_rate": 2.4486996279845862e-05,
1710
+ "loss": 0.681,
1711
+ "step": 2410
1712
+ },
1713
+ {
1714
+ "epoch": 2.1172609932181143,
1715
+ "grad_norm": 0.8531279563903809,
1716
+ "learning_rate": 2.4050146001035983e-05,
1717
+ "loss": 0.6548,
1718
+ "step": 2420
1719
+ },
1720
+ {
1721
+ "epoch": 2.126011813607526,
1722
+ "grad_norm": 0.8686200976371765,
1723
+ "learning_rate": 2.3615990257396946e-05,
1724
+ "loss": 0.6546,
1725
+ "step": 2430
1726
+ },
1727
+ {
1728
+ "epoch": 2.134762633996937,
1729
+ "grad_norm": 0.8861080408096313,
1730
+ "learning_rate": 2.318457413002718e-05,
1731
+ "loss": 0.655,
1732
+ "step": 2440
1733
+ },
1734
+ {
1735
+ "epoch": 2.1435134543863485,
1736
+ "grad_norm": 0.8589378595352173,
1737
+ "learning_rate": 2.2755942415553593e-05,
1738
+ "loss": 0.6359,
1739
+ "step": 2450
1740
+ },
1741
+ {
1742
+ "epoch": 2.15226427477576,
1743
+ "grad_norm": 0.784140944480896,
1744
+ "learning_rate": 2.233013962148016e-05,
1745
+ "loss": 0.6206,
1746
+ "step": 2460
1747
+ },
1748
+ {
1749
+ "epoch": 2.1610150951651717,
1750
+ "grad_norm": 0.9516744613647461,
1751
+ "learning_rate": 2.1907209961566415e-05,
1752
+ "loss": 0.671,
1753
+ "step": 2470
1754
+ },
1755
+ {
1756
+ "epoch": 2.169765915554583,
1757
+ "grad_norm": 0.8857921957969666,
1758
+ "learning_rate": 2.1487197351236443e-05,
1759
+ "loss": 0.6435,
1760
+ "step": 2480
1761
+ },
1762
+ {
1763
+ "epoch": 2.1785167359439948,
1764
+ "grad_norm": 0.8351215124130249,
1765
+ "learning_rate": 2.107014540301886e-05,
1766
+ "loss": 0.6536,
1767
+ "step": 2490
1768
+ },
1769
+ {
1770
+ "epoch": 2.1872675563334063,
1771
+ "grad_norm": 0.8609489798545837,
1772
+ "learning_rate": 2.0656097422018346e-05,
1773
+ "loss": 0.652,
1774
+ "step": 2500
1775
+ },
1776
+ {
1777
+ "epoch": 2.196018376722818,
1778
+ "grad_norm": 0.8463150858879089,
1779
+ "learning_rate": 2.024509640141885e-05,
1780
+ "loss": 0.6601,
1781
+ "step": 2510
1782
+ },
1783
+ {
1784
+ "epoch": 2.2047691971122294,
1785
+ "grad_norm": 0.8469570279121399,
1786
+ "learning_rate": 1.9837185018019467e-05,
1787
+ "loss": 0.6646,
1788
+ "step": 2520
1789
+ },
1790
+ {
1791
+ "epoch": 2.213520017501641,
1792
+ "grad_norm": 0.870857298374176,
1793
+ "learning_rate": 1.9432405627802963e-05,
1794
+ "loss": 0.6429,
1795
+ "step": 2530
1796
+ },
1797
+ {
1798
+ "epoch": 2.2222708378910525,
1799
+ "grad_norm": 0.865593671798706,
1800
+ "learning_rate": 1.903080026153774e-05,
1801
+ "loss": 0.649,
1802
+ "step": 2540
1803
+ },
1804
+ {
1805
+ "epoch": 2.2310216582804636,
1806
+ "grad_norm": 0.8608151078224182,
1807
+ "learning_rate": 1.8632410620413503e-05,
1808
+ "loss": 0.6564,
1809
+ "step": 2550
1810
+ },
1811
+ {
1812
+ "epoch": 2.239772478669875,
1813
+ "grad_norm": 0.8044232726097107,
1814
+ "learning_rate": 1.823727807171115e-05,
1815
+ "loss": 0.6786,
1816
+ "step": 2560
1817
+ },
1818
+ {
1819
+ "epoch": 2.2485232990592867,
1820
+ "grad_norm": 0.8390374779701233,
1821
+ "learning_rate": 1.7845443644507354e-05,
1822
+ "loss": 0.6385,
1823
+ "step": 2570
1824
+ },
1825
+ {
1826
+ "epoch": 2.2572741194486983,
1827
+ "grad_norm": 0.8899725079536438,
1828
+ "learning_rate": 1.745694802541429e-05,
1829
+ "loss": 0.6674,
1830
+ "step": 2580
1831
+ },
1832
+ {
1833
+ "epoch": 2.26602493983811,
1834
+ "grad_norm": 0.8680664300918579,
1835
+ "learning_rate": 1.7071831554354823e-05,
1836
+ "loss": 0.6337,
1837
+ "step": 2590
1838
+ },
1839
+ {
1840
+ "epoch": 2.2747757602275214,
1841
+ "grad_norm": 0.843752384185791,
1842
+ "learning_rate": 1.6690134220373864e-05,
1843
+ "loss": 0.6582,
1844
+ "step": 2600
1845
+ },
1846
+ {
1847
+ "epoch": 2.283526580616933,
1848
+ "grad_norm": 0.8479620814323425,
1849
+ "learning_rate": 1.631189565748596e-05,
1850
+ "loss": 0.6531,
1851
+ "step": 2610
1852
+ },
1853
+ {
1854
+ "epoch": 2.2922774010063445,
1855
+ "grad_norm": 0.8540963530540466,
1856
+ "learning_rate": 1.5937155140559884e-05,
1857
+ "loss": 0.6386,
1858
+ "step": 2620
1859
+ },
1860
+ {
1861
+ "epoch": 2.301028221395756,
1862
+ "grad_norm": 0.8910626769065857,
1863
+ "learning_rate": 1.556595158124049e-05,
1864
+ "loss": 0.6533,
1865
+ "step": 2630
1866
+ },
1867
+ {
1868
+ "epoch": 2.309779041785167,
1869
+ "grad_norm": 0.8980609178543091,
1870
+ "learning_rate": 1.5198323523908265e-05,
1871
+ "loss": 0.6487,
1872
+ "step": 2640
1873
+ },
1874
+ {
1875
+ "epoch": 2.3185298621745787,
1876
+ "grad_norm": 0.8651220202445984,
1877
+ "learning_rate": 1.483430914167701e-05,
1878
+ "loss": 0.6598,
1879
+ "step": 2650
1880
+ },
1881
+ {
1882
+ "epoch": 2.3272806825639902,
1883
+ "grad_norm": 0.9363301396369934,
1884
+ "learning_rate": 1.447394623243018e-05,
1885
+ "loss": 0.678,
1886
+ "step": 2660
1887
+ },
1888
+ {
1889
+ "epoch": 2.336031502953402,
1890
+ "grad_norm": 0.8213711977005005,
1891
+ "learning_rate": 1.4117272214895899e-05,
1892
+ "loss": 0.6359,
1893
+ "step": 2670
1894
+ },
1895
+ {
1896
+ "epoch": 2.3447823233428133,
1897
+ "grad_norm": 0.8319981694221497,
1898
+ "learning_rate": 1.3764324124761735e-05,
1899
+ "loss": 0.6698,
1900
+ "step": 2680
1901
+ },
1902
+ {
1903
+ "epoch": 2.353533143732225,
1904
+ "grad_norm": 0.8644684553146362,
1905
+ "learning_rate": 1.3415138610828988e-05,
1906
+ "loss": 0.6499,
1907
+ "step": 2690
1908
+ },
1909
+ {
1910
+ "epoch": 2.3622839641216364,
1911
+ "grad_norm": 0.8030415773391724,
1912
+ "learning_rate": 1.3069751931207164e-05,
1913
+ "loss": 0.6219,
1914
+ "step": 2700
1915
+ },
1916
+ {
1917
+ "epoch": 2.371034784511048,
1918
+ "grad_norm": 0.717867374420166,
1919
+ "learning_rate": 1.2728199949549142e-05,
1920
+ "loss": 0.6315,
1921
+ "step": 2710
1922
+ },
1923
+ {
1924
+ "epoch": 2.3797856049004595,
1925
+ "grad_norm": 0.790641188621521,
1926
+ "learning_rate": 1.2390518131327172e-05,
1927
+ "loss": 0.6273,
1928
+ "step": 2720
1929
+ },
1930
+ {
1931
+ "epoch": 2.388536425289871,
1932
+ "grad_norm": 0.7971424460411072,
1933
+ "learning_rate": 1.2056741540150328e-05,
1934
+ "loss": 0.6415,
1935
+ "step": 2730
1936
+ },
1937
+ {
1938
+ "epoch": 2.3972872456792826,
1939
+ "grad_norm": 0.8711332678794861,
1940
+ "learning_rate": 1.1726904834123581e-05,
1941
+ "loss": 0.6474,
1942
+ "step": 2740
1943
+ },
1944
+ {
1945
+ "epoch": 2.4060380660686937,
1946
+ "grad_norm": 0.9122375249862671,
1947
+ "learning_rate": 1.1401042262249057e-05,
1948
+ "loss": 0.656,
1949
+ "step": 2750
1950
+ },
1951
+ {
1952
+ "epoch": 2.4147888864581053,
1953
+ "grad_norm": 0.9008005261421204,
1954
+ "learning_rate": 1.10791876608698e-05,
1955
+ "loss": 0.6204,
1956
+ "step": 2760
1957
+ },
1958
+ {
1959
+ "epoch": 2.423539706847517,
1960
+ "grad_norm": 0.7929515838623047,
1961
+ "learning_rate": 1.0761374450156202e-05,
1962
+ "loss": 0.6271,
1963
+ "step": 2770
1964
+ },
1965
+ {
1966
+ "epoch": 2.4322905272369284,
1967
+ "grad_norm": 0.8115386962890625,
1968
+ "learning_rate": 1.04476356306359e-05,
1969
+ "loss": 0.63,
1970
+ "step": 2780
1971
+ },
1972
+ {
1973
+ "epoch": 2.44104134762634,
1974
+ "grad_norm": 0.8349243402481079,
1975
+ "learning_rate": 1.0138003779767103e-05,
1976
+ "loss": 0.6557,
1977
+ "step": 2790
1978
+ },
1979
+ {
1980
+ "epoch": 2.4497921680157515,
1981
+ "grad_norm": 0.8012370467185974,
1982
+ "learning_rate": 9.832511048555826e-06,
1983
+ "loss": 0.6668,
1984
+ "step": 2800
1985
+ },
1986
+ {
1987
+ "epoch": 2.458542988405163,
1988
+ "grad_norm": 0.885390043258667,
1989
+ "learning_rate": 9.53118915821748e-06,
1990
+ "loss": 0.6331,
1991
+ "step": 2810
1992
+ },
1993
+ {
1994
+ "epoch": 2.4672938087945746,
1995
+ "grad_norm": 0.8025451302528381,
1996
+ "learning_rate": 9.234069396883066e-06,
1997
+ "loss": 0.6539,
1998
+ "step": 2820
1999
+ },
2000
+ {
2001
+ "epoch": 2.476044629183986,
2002
+ "grad_norm": 0.8048675060272217,
2003
+ "learning_rate": 8.941182616350318e-06,
2004
+ "loss": 0.6396,
2005
+ "step": 2830
2006
+ },
2007
+ {
2008
+ "epoch": 2.4847954495733973,
2009
+ "grad_norm": 0.8225316405296326,
2010
+ "learning_rate": 8.65255922888017e-06,
2011
+ "loss": 0.6491,
2012
+ "step": 2840
2013
+ },
2014
+ {
2015
+ "epoch": 2.493546269962809,
2016
+ "grad_norm": 0.8796355128288269,
2017
+ "learning_rate": 8.368229204038897e-06,
2018
+ "loss": 0.665,
2019
+ "step": 2850
2020
+ },
2021
+ {
2022
+ "epoch": 2.5022970903522204,
2023
+ "grad_norm": 0.8243616223335266,
2024
+ "learning_rate": 8.088222065586132e-06,
2025
+ "loss": 0.6448,
2026
+ "step": 2860
2027
+ },
2028
+ {
2029
+ "epoch": 2.511047910741632,
2030
+ "grad_norm": 0.873279333114624,
2031
+ "learning_rate": 7.812566888409228e-06,
2032
+ "loss": 0.6534,
2033
+ "step": 2870
2034
+ },
2035
+ {
2036
+ "epoch": 2.5197987311310435,
2037
+ "grad_norm": 0.9333387017250061,
2038
+ "learning_rate": 7.541292295504327e-06,
2039
+ "loss": 0.6393,
2040
+ "step": 2880
2041
+ },
2042
+ {
2043
+ "epoch": 2.528549551520455,
2044
+ "grad_norm": 0.800175666809082,
2045
+ "learning_rate": 7.274426455004129e-06,
2046
+ "loss": 0.6389,
2047
+ "step": 2890
2048
+ },
2049
+ {
2050
+ "epoch": 2.5373003719098666,
2051
+ "grad_norm": 0.8389214873313904,
2052
+ "learning_rate": 7.011997077253113e-06,
2053
+ "loss": 0.6337,
2054
+ "step": 2900
2055
+ },
2056
+ {
2057
+ "epoch": 2.546051192299278,
2058
+ "grad_norm": 0.8694761395454407,
2059
+ "learning_rate": 6.754031411930139e-06,
2060
+ "loss": 0.6512,
2061
+ "step": 2910
2062
+ },
2063
+ {
2064
+ "epoch": 2.5548020126886897,
2065
+ "grad_norm": 0.8349895477294922,
2066
+ "learning_rate": 6.5005562452189865e-06,
2067
+ "loss": 0.6452,
2068
+ "step": 2920
2069
+ },
2070
+ {
2071
+ "epoch": 2.5635528330781012,
2072
+ "grad_norm": 0.8583366274833679,
2073
+ "learning_rate": 6.251597897026951e-06,
2074
+ "loss": 0.6462,
2075
+ "step": 2930
2076
+ },
2077
+ {
2078
+ "epoch": 2.5723036534675128,
2079
+ "grad_norm": 0.7806413173675537,
2080
+ "learning_rate": 6.007182218251883e-06,
2081
+ "loss": 0.6168,
2082
+ "step": 2940
2083
+ },
2084
+ {
2085
+ "epoch": 2.5810544738569243,
2086
+ "grad_norm": 0.8689520955085754,
2087
+ "learning_rate": 5.7673345880979775e-06,
2088
+ "loss": 0.6336,
2089
+ "step": 2950
2090
+ },
2091
+ {
2092
+ "epoch": 2.5898052942463354,
2093
+ "grad_norm": 0.7911117672920227,
2094
+ "learning_rate": 5.532079911440419e-06,
2095
+ "loss": 0.6244,
2096
+ "step": 2960
2097
+ },
2098
+ {
2099
+ "epoch": 2.598556114635747,
2100
+ "grad_norm": 0.8084582686424255,
2101
+ "learning_rate": 5.301442616239372e-06,
2102
+ "loss": 0.6187,
2103
+ "step": 2970
2104
+ },
2105
+ {
2106
+ "epoch": 2.6073069350251585,
2107
+ "grad_norm": 0.8228966593742371,
2108
+ "learning_rate": 5.075446651003535e-06,
2109
+ "loss": 0.6187,
2110
+ "step": 2980
2111
+ },
2112
+ {
2113
+ "epoch": 2.61605775541457,
2114
+ "grad_norm": 0.8767485022544861,
2115
+ "learning_rate": 4.854115482303345e-06,
2116
+ "loss": 0.617,
2117
+ "step": 2990
2118
+ },
2119
+ {
2120
+ "epoch": 2.6248085758039816,
2121
+ "grad_norm": 0.8234701752662659,
2122
+ "learning_rate": 4.637472092334327e-06,
2123
+ "loss": 0.653,
2124
+ "step": 3000
2125
+ },
2126
+ {
2127
+ "epoch": 2.633559396193393,
2128
+ "grad_norm": 0.8716809749603271,
2129
+ "learning_rate": 4.425538976530708e-06,
2130
+ "loss": 0.6382,
2131
+ "step": 3010
2132
+ },
2133
+ {
2134
+ "epoch": 2.6423102165828047,
2135
+ "grad_norm": 0.8220137357711792,
2136
+ "learning_rate": 4.218338141229566e-06,
2137
+ "loss": 0.6273,
2138
+ "step": 3020
2139
+ },
2140
+ {
2141
+ "epoch": 2.6510610369722163,
2142
+ "grad_norm": 0.7726868987083435,
2143
+ "learning_rate": 4.015891101385794e-06,
2144
+ "loss": 0.631,
2145
+ "step": 3030
2146
+ },
2147
+ {
2148
+ "epoch": 2.6598118573616274,
2149
+ "grad_norm": 0.8567168116569519,
2150
+ "learning_rate": 3.818218878338054e-06,
2151
+ "loss": 0.6436,
2152
+ "step": 3040
2153
+ },
2154
+ {
2155
+ "epoch": 2.668562677751039,
2156
+ "grad_norm": 0.8137094378471375,
2157
+ "learning_rate": 3.6253419976260027e-06,
2158
+ "loss": 0.6439,
2159
+ "step": 3050
2160
+ },
2161
+ {
2162
+ "epoch": 2.6773134981404505,
2163
+ "grad_norm": 0.835558295249939,
2164
+ "learning_rate": 3.437280486858996e-06,
2165
+ "loss": 0.6324,
2166
+ "step": 3060
2167
+ },
2168
+ {
2169
+ "epoch": 2.686064318529862,
2170
+ "grad_norm": 0.8575820326805115,
2171
+ "learning_rate": 3.254053873636498e-06,
2172
+ "loss": 0.6339,
2173
+ "step": 3070
2174
+ },
2175
+ {
2176
+ "epoch": 2.6948151389192736,
2177
+ "grad_norm": 0.8318324685096741,
2178
+ "learning_rate": 3.075681183520407e-06,
2179
+ "loss": 0.6386,
2180
+ "step": 3080
2181
+ },
2182
+ {
2183
+ "epoch": 2.703565959308685,
2184
+ "grad_norm": 0.8100464940071106,
2185
+ "learning_rate": 2.9021809380595066e-06,
2186
+ "loss": 0.6362,
2187
+ "step": 3090
2188
+ },
2189
+ {
2190
+ "epoch": 2.7123167796980967,
2191
+ "grad_norm": 0.7721264362335205,
2192
+ "learning_rate": 2.7335711528663046e-06,
2193
+ "loss": 0.639,
2194
+ "step": 3100
2195
+ },
2196
+ {
2197
+ "epoch": 2.7210676000875083,
2198
+ "grad_norm": 0.8360996842384338,
2199
+ "learning_rate": 2.5698693357462665e-06,
2200
+ "loss": 0.625,
2201
+ "step": 3110
2202
+ },
2203
+ {
2204
+ "epoch": 2.72981842047692,
2205
+ "grad_norm": 0.859639048576355,
2206
+ "learning_rate": 2.4110924848799566e-06,
2207
+ "loss": 0.6159,
2208
+ "step": 3120
2209
+ },
2210
+ {
2211
+ "epoch": 2.7385692408663314,
2212
+ "grad_norm": 0.9522019624710083,
2213
+ "learning_rate": 2.2572570870579848e-06,
2214
+ "loss": 0.6437,
2215
+ "step": 3130
2216
+ },
2217
+ {
2218
+ "epoch": 2.747320061255743,
2219
+ "grad_norm": 0.8836742639541626,
2220
+ "learning_rate": 2.1083791159690613e-06,
2221
+ "loss": 0.6294,
2222
+ "step": 3140
2223
+ },
2224
+ {
2225
+ "epoch": 2.7560708816451545,
2226
+ "grad_norm": 0.901431143283844,
2227
+ "learning_rate": 1.9644740305413633e-06,
2228
+ "loss": 0.6387,
2229
+ "step": 3150
2230
+ },
2231
+ {
2232
+ "epoch": 2.764821702034566,
2233
+ "grad_norm": 0.8418229818344116,
2234
+ "learning_rate": 1.825556773337339e-06,
2235
+ "loss": 0.6355,
2236
+ "step": 3160
2237
+ },
2238
+ {
2239
+ "epoch": 2.773572522423977,
2240
+ "grad_norm": 0.854288637638092,
2241
+ "learning_rate": 1.691641769002139e-06,
2242
+ "loss": 0.6357,
2243
+ "step": 3170
2244
+ },
2245
+ {
2246
+ "epoch": 2.7823233428133887,
2247
+ "grad_norm": 0.897578239440918,
2248
+ "learning_rate": 1.5627429227657863e-06,
2249
+ "loss": 0.6584,
2250
+ "step": 3180
2251
+ },
2252
+ {
2253
+ "epoch": 2.7910741632028,
2254
+ "grad_norm": 0.94583660364151,
2255
+ "learning_rate": 1.4388736189993258e-06,
2256
+ "loss": 0.6333,
2257
+ "step": 3190
2258
+ },
2259
+ {
2260
+ "epoch": 2.7998249835922118,
2261
+ "grad_norm": 0.8198401927947998,
2262
+ "learning_rate": 1.3200467198250688e-06,
2263
+ "loss": 0.6343,
2264
+ "step": 3200
2265
+ },
2266
+ {
2267
+ "epoch": 2.8085758039816233,
2268
+ "grad_norm": 0.8177739381790161,
2269
+ "learning_rate": 1.2062745637809736e-06,
2270
+ "loss": 0.6351,
2271
+ "step": 3210
2272
+ },
2273
+ {
2274
+ "epoch": 2.817326624371035,
2275
+ "grad_norm": 0.8060468435287476,
2276
+ "learning_rate": 1.0975689645395193e-06,
2277
+ "loss": 0.616,
2278
+ "step": 3220
2279
+ },
2280
+ {
2281
+ "epoch": 2.8260774447604464,
2282
+ "grad_norm": 0.8761562705039978,
2283
+ "learning_rate": 9.939412096809986e-07,
2284
+ "loss": 0.6426,
2285
+ "step": 3230
2286
+ },
2287
+ {
2288
+ "epoch": 2.8348282651498575,
2289
+ "grad_norm": 0.8060517311096191,
2290
+ "learning_rate": 8.954020595214441e-07,
2291
+ "loss": 0.6432,
2292
+ "step": 3240
2293
+ },
2294
+ {
2295
+ "epoch": 2.843579085539269,
2296
+ "grad_norm": 0.8677119016647339,
2297
+ "learning_rate": 8.019617459953332e-07,
2298
+ "loss": 0.6246,
2299
+ "step": 3250
2300
+ },
2301
+ {
2302
+ "epoch": 2.8523299059286806,
2303
+ "grad_norm": 0.8648504018783569,
2304
+ "learning_rate": 7.136299715931438e-07,
2305
+ "loss": 0.6246,
2306
+ "step": 3260
2307
+ },
2308
+ {
2309
+ "epoch": 2.861080726318092,
2310
+ "grad_norm": 0.8173497319221497,
2311
+ "learning_rate": 6.304159083538652e-07,
2312
+ "loss": 0.6188,
2313
+ "step": 3270
2314
+ },
2315
+ {
2316
+ "epoch": 2.8698315467075037,
2317
+ "grad_norm": 0.8254984617233276,
2318
+ "learning_rate": 5.523281969126271e-07,
2319
+ "loss": 0.6365,
2320
+ "step": 3280
2321
+ },
2322
+ {
2323
+ "epoch": 2.8785823670969153,
2324
+ "grad_norm": 0.8288102746009827,
2325
+ "learning_rate": 4.793749456034946e-07,
2326
+ "loss": 0.6199,
2327
+ "step": 3290
2328
+ },
2329
+ {
2330
+ "epoch": 2.887333187486327,
2331
+ "grad_norm": 0.8244780898094177,
2332
+ "learning_rate": 4.115637296174979e-07,
2333
+ "loss": 0.6438,
2334
+ "step": 3300
2335
+ },
2336
+ {
2337
+ "epoch": 2.8960840078757384,
2338
+ "grad_norm": 0.8321549296379089,
2339
+ "learning_rate": 3.4890159021608303e-07,
2340
+ "loss": 0.623,
2341
+ "step": 3310
2342
+ },
2343
+ {
2344
+ "epoch": 2.90483482826515,
2345
+ "grad_norm": 0.748741865158081,
2346
+ "learning_rate": 2.913950339999638e-07,
2347
+ "loss": 0.6297,
2348
+ "step": 3320
2349
+ },
2350
+ {
2351
+ "epoch": 2.9135856486545615,
2352
+ "grad_norm": 0.8655825853347778,
2353
+ "learning_rate": 2.390500322334954e-07,
2354
+ "loss": 0.6257,
2355
+ "step": 3330
2356
+ },
2357
+ {
2358
+ "epoch": 2.922336469043973,
2359
+ "grad_norm": 0.8270664215087891,
2360
+ "learning_rate": 1.9187202022464823e-07,
2361
+ "loss": 0.6356,
2362
+ "step": 3340
2363
+ },
2364
+ {
2365
+ "epoch": 2.9310872894333846,
2366
+ "grad_norm": 0.892652153968811,
2367
+ "learning_rate": 1.4986589676063145e-07,
2368
+ "loss": 0.6228,
2369
+ "step": 3350
2370
+ },
2371
+ {
2372
+ "epoch": 2.939838109822796,
2373
+ "grad_norm": 0.9109422564506531,
2374
+ "learning_rate": 1.1303602359919452e-07,
2375
+ "loss": 0.6493,
2376
+ "step": 3360
2377
+ },
2378
+ {
2379
+ "epoch": 2.9485889302122072,
2380
+ "grad_norm": 0.820085883140564,
2381
+ "learning_rate": 8.138622501575598e-08,
2382
+ "loss": 0.5908,
2383
+ "step": 3370
2384
+ },
2385
+ {
2386
+ "epoch": 2.957339750601619,
2387
+ "grad_norm": 0.8783090114593506,
2388
+ "learning_rate": 5.491978740627679e-08,
2389
+ "loss": 0.6305,
2390
+ "step": 3380
2391
+ },
2392
+ {
2393
+ "epoch": 2.9660905709910304,
2394
+ "grad_norm": 0.8442906141281128,
2395
+ "learning_rate": 3.36394589460276e-08,
2396
+ "loss": 0.6265,
2397
+ "step": 3390
2398
+ },
2399
+ {
2400
+ "epoch": 2.974841391380442,
2401
+ "grad_norm": 0.8543626666069031,
2402
+ "learning_rate": 1.7547449304222695e-08,
2403
+ "loss": 0.627,
2404
+ "step": 3400
2405
+ },
2406
+ {
2407
+ "epoch": 2.9835922117698535,
2408
+ "grad_norm": 0.8204004168510437,
2409
+ "learning_rate": 6.645429414570137e-09,
2410
+ "loss": 0.6266,
2411
+ "step": 3410
2412
+ },
2413
+ {
2414
+ "epoch": 2.992343032159265,
2415
+ "grad_norm": 0.804568350315094,
2416
+ "learning_rate": 9.34531301793884e-10,
2417
+ "loss": 0.6332,
2418
+ "step": 3420
2419
+ }
2420
+ ],
2421
+ "logging_steps": 10,
2422
+ "max_steps": 3426,
2423
+ "num_input_tokens_seen": 0,
2424
+ "num_train_epochs": 3,
2425
+ "save_steps": 500.0,
2426
+ "stateful_callbacks": {
2427
+ "TrainerControl": {
2428
+ "args": {
2429
+ "should_epoch_stop": false,
2430
+ "should_evaluate": false,
2431
+ "should_log": false,
2432
+ "should_save": true,
2433
+ "should_training_stop": true
2434
+ },
2435
+ "attributes": {}
2436
+ }
2437
+ },
2438
+ "total_flos": 5.3232347449707725e+17,
2439
+ "train_batch_size": 4,
2440
+ "trial_name": null,
2441
+ "trial_params": null
2442
+ }
config.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "meta-llama/Llama-3.2-1B",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 128000,
9
+ "eos_token_id": 128001,
10
+ "head_dim": 64,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 2048,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 8192,
15
+ "max_position_embeddings": 131072,
16
+ "mlp_bias": false,
17
+ "model_type": "llama",
18
+ "num_attention_heads": 32,
19
+ "num_hidden_layers": 16,
20
+ "num_key_value_heads": 8,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": {
24
+ "factor": 32.0,
25
+ "high_freq_factor": 4.0,
26
+ "low_freq_factor": 1.0,
27
+ "original_max_position_embeddings": 8192,
28
+ "rope_type": "llama3"
29
+ },
30
+ "rope_theta": 500000.0,
31
+ "tie_word_embeddings": true,
32
+ "torch_dtype": "float32",
33
+ "transformers_version": "4.48.1",
34
+ "use_cache": false,
35
+ "vocab_size": 128256
36
+ }