Hugging Face's logo

Matt300209
/
1B_code_only_1e5

Safetensors
llama
Model card Files
xet
Community
Matt300209 commited on
Commit
f61e12b
·
verified ·
1 Parent(s): 91135e7

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:b9b855d2662954fceb129a79c5e6ac24e443e74407207c79ceca3520588e4534
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 sha256:150617b53af8b6b3aceff6896c597fe647cbc91246269a9cffe1a08d08aa0a3a
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:b1e85ba5f8872d01f6d97181178314da3d6d732e8087dde6ad2a8e4727e5e9e8
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 sha256:391a83cf4e2a7a8f61fdf23f565b714522f91910370c5cdec4a8dedf5a870fd0
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:aebe83ebbbc82821c1d286b4e6f18f643826d5190cc9fe9a574f3a12c5264065
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:ac959a9cd92b4b4ae9a37cd34320315a0c5afe58b7a1effc73abd975dfab605b
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 sha256:8adc9c1295ba6c20e88c04c22fdc77ec8c59424aa962981a9f04a87177537ff0
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 sha256:1968a21f8dbd8e1b9492a99928fa09cb10c8abd69d88a0fdbf81be383897e233
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 sha256:119ab7981418c1d6c974e6fc05dca464e47ea0692a0de3864ea0ad64c56942d1
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 sha256:873ef15b1c38e039f61da33ae5db453fa413afbb7e6b5ac9f0265a6c537b2532
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 sha256:2ddbb07fba7ead8a3ab352dd31315c7a538336f468e247e773577e10d3b797ae
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": 0.8244,
15
+ "eval_samples_per_second": 24.261,
16
+ "eval_steps_per_second": 2.426,
17
+ "step": 0
18
+ },
19
+ {
20
+ "epoch": 0.0008750820389411508,
21
+ "grad_norm": 6.372805118560791,
22
+ "learning_rate": 2.915451895043732e-08,
23
+ "loss": 1.6434,
24
+ "step": 1
25
+ },
26
+ {
27
+ "epoch": 0.008750820389411508,
28
+ "grad_norm": 6.278440475463867,
29
+ "learning_rate": 2.915451895043732e-07,
30
+ "loss": 1.7188,
31
+ "step": 10
32
+ },
33
+ {
34
+ "epoch": 0.017501640778823015,
35
+ "grad_norm": 6.023682594299316,
36
+ "learning_rate": 5.830903790087464e-07,
37
+ "loss": 1.7154,
38
+ "step": 20
39
+ },
40
+ {
41
+ "epoch": 0.02625246116823452,
42
+ "grad_norm": 3.7778327465057373,
43
+ "learning_rate": 8.746355685131196e-07,
44
+ "loss": 1.6563,
45
+ "step": 30
46
+ },
47
+ {
48
+ "epoch": 0.03500328155764603,
49
+ "grad_norm": 3.8672306537628174,
50
+ "learning_rate": 1.1661807580174927e-06,
51
+ "loss": 1.5743,
52
+ "step": 40
53
+ },
54
+ {
55
+ "epoch": 0.04375410194705753,
56
+ "grad_norm": 2.53611159324646,
57
+ "learning_rate": 1.4577259475218661e-06,
58
+ "loss": 1.479,
59
+ "step": 50
60
+ },
61
+ {
62
+ "epoch": 0.05250492233646904,
63
+ "grad_norm": 2.019819974899292,
64
+ "learning_rate": 1.7492711370262391e-06,
65
+ "loss": 1.4242,
66
+ "step": 60
67
+ },
68
+ {
69
+ "epoch": 0.06125574272588055,
70
+ "grad_norm": 1.7553459405899048,
71
+ "learning_rate": 2.0408163265306125e-06,
72
+ "loss": 1.3795,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.07000656311529206,
77
+ "grad_norm": 1.4397658109664917,
78
+ "learning_rate": 2.3323615160349855e-06,
79
+ "loss": 1.3745,
80
+ "step": 80
81
+ },
82
+ {
83
+ "epoch": 0.07875738350470357,
84
+ "grad_norm": 1.5793251991271973,
85
+ "learning_rate": 2.6239067055393585e-06,
86
+ "loss": 1.3529,
87
+ "step": 90
88
+ },
89
+ {
90
+ "epoch": 0.08750820389411507,
91
+ "grad_norm": 1.5069832801818848,
92
+ "learning_rate": 2.9154518950437323e-06,
93
+ "loss": 1.3459,
94
+ "step": 100
95
+ },
96
+ {
97
+ "epoch": 0.09625902428352658,
98
+ "grad_norm": 1.4921239614486694,
99
+ "learning_rate": 3.2069970845481052e-06,
100
+ "loss": 1.2743,
101
+ "step": 110
102
+ },
103
+ {
104
+ "epoch": 0.10500984467293809,
105
+ "grad_norm": 1.4761064052581787,
106
+ "learning_rate": 3.4985422740524782e-06,
107
+ "loss": 1.2959,
108
+ "step": 120
109
+ },
110
+ {
111
+ "epoch": 0.1137606650623496,
112
+ "grad_norm": 1.4724879264831543,
113
+ "learning_rate": 3.790087463556852e-06,
114
+ "loss": 1.2773,
115
+ "step": 130
116
+ },
117
+ {
118
+ "epoch": 0.1225114854517611,
119
+ "grad_norm": 1.4954681396484375,
120
+ "learning_rate": 4.081632653061225e-06,
121
+ "loss": 1.2695,
122
+ "step": 140
123
+ },
124
+ {
125
+ "epoch": 0.1312623058411726,
126
+ "grad_norm": 1.4220881462097168,
127
+ "learning_rate": 4.3731778425655976e-06,
128
+ "loss": 1.2668,
129
+ "step": 150
130
+ },
131
+ {
132
+ "epoch": 0.14001312623058412,
133
+ "grad_norm": 1.4039027690887451,
134
+ "learning_rate": 4.664723032069971e-06,
135
+ "loss": 1.2231,
136
+ "step": 160
137
+ },
138
+ {
139
+ "epoch": 0.14876394661999562,
140
+ "grad_norm": 1.468996524810791,
141
+ "learning_rate": 4.956268221574344e-06,
142
+ "loss": 1.221,
143
+ "step": 170
144
+ },
145
+ {
146
+ "epoch": 0.15751476700940714,
147
+ "grad_norm": 1.5780359506607056,
148
+ "learning_rate": 5.247813411078717e-06,
149
+ "loss": 1.2403,
150
+ "step": 180
151
+ },
152
+ {
153
+ "epoch": 0.16626558739881864,
154
+ "grad_norm": 1.4090375900268555,
155
+ "learning_rate": 5.539358600583091e-06,
156
+ "loss": 1.2021,
157
+ "step": 190
158
+ },
159
+ {
160
+ "epoch": 0.17501640778823013,
161
+ "grad_norm": 1.489652395248413,
162
+ "learning_rate": 5.8309037900874645e-06,
163
+ "loss": 1.1731,
164
+ "step": 200
165
+ },
166
+ {
167
+ "epoch": 0.18376722817764166,
168
+ "grad_norm": 1.3121322393417358,
169
+ "learning_rate": 6.122448979591837e-06,
170
+ "loss": 1.1924,
171
+ "step": 210
172
+ },
173
+ {
174
+ "epoch": 0.19251804856705315,
175
+ "grad_norm": 1.3954280614852905,
176
+ "learning_rate": 6.4139941690962105e-06,
177
+ "loss": 1.2121,
178
+ "step": 220
179
+ },
180
+ {
181
+ "epoch": 0.20126886895646467,
182
+ "grad_norm": 1.4191243648529053,
183
+ "learning_rate": 6.705539358600584e-06,
184
+ "loss": 1.1973,
185
+ "step": 230
186
+ },
187
+ {
188
+ "epoch": 0.21001968934587617,
189
+ "grad_norm": 1.4373400211334229,
190
+ "learning_rate": 6.9970845481049564e-06,
191
+ "loss": 1.1701,
192
+ "step": 240
193
+ },
194
+ {
195
+ "epoch": 0.2187705097352877,
196
+ "grad_norm": 1.4669713973999023,
197
+ "learning_rate": 7.28862973760933e-06,
198
+ "loss": 1.1923,
199
+ "step": 250
200
+ },
201
+ {
202
+ "epoch": 0.2275213301246992,
203
+ "grad_norm": 1.4269640445709229,
204
+ "learning_rate": 7.580174927113704e-06,
205
+ "loss": 1.1713,
206
+ "step": 260
207
+ },
208
+ {
209
+ "epoch": 0.23627215051411068,
210
+ "grad_norm": 1.7535545825958252,
211
+ "learning_rate": 7.871720116618077e-06,
212
+ "loss": 1.1891,
213
+ "step": 270
214
+ },
215
+ {
216
+ "epoch": 0.2450229709035222,
217
+ "grad_norm": 1.599094271659851,
218
+ "learning_rate": 8.16326530612245e-06,
219
+ "loss": 1.1827,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2537737912929337,
224
+ "grad_norm": 1.8278971910476685,
225
+ "learning_rate": 8.454810495626823e-06,
226
+ "loss": 1.1402,
227
+ "step": 290
228
+ },
229
+ {
230
+ "epoch": 0.2625246116823452,
231
+ "grad_norm": 1.5127216577529907,
232
+ "learning_rate": 8.746355685131195e-06,
233
+ "loss": 1.1597,
234
+ "step": 300
235
+ },
236
+ {
237
+ "epoch": 0.27127543207175675,
238
+ "grad_norm": 1.6015137434005737,
239
+ "learning_rate": 9.03790087463557e-06,
240
+ "loss": 1.1398,
241
+ "step": 310
242
+ },
243
+ {
244
+ "epoch": 0.28002625246116825,
245
+ "grad_norm": 1.3944435119628906,
246
+ "learning_rate": 9.329446064139942e-06,
247
+ "loss": 1.1174,
248
+ "step": 320
249
+ },
250
+ {
251
+ "epoch": 0.28877707285057974,
252
+ "grad_norm": 1.2441927194595337,
253
+ "learning_rate": 9.620991253644316e-06,
254
+ "loss": 1.1416,
255
+ "step": 330
256
+ },
257
+ {
258
+ "epoch": 0.29752789323999124,
259
+ "grad_norm": 1.4111624956130981,
260
+ "learning_rate": 9.912536443148689e-06,
261
+ "loss": 1.1352,
262
+ "step": 340
263
+ },
264
+ {
265
+ "epoch": 0.30627871362940273,
266
+ "grad_norm": 1.3007830381393433,
267
+ "learning_rate": 9.999872800049233e-06,
268
+ "loss": 1.1208,
269
+ "step": 350
270
+ },
271
+ {
272
+ "epoch": 0.3150295340188143,
273
+ "grad_norm": 1.3299254179000854,
274
+ "learning_rate": 9.999249795462235e-06,
275
+ "loss": 1.142,
276
+ "step": 360
277
+ },
278
+ {
279
+ "epoch": 0.3237803544082258,
280
+ "grad_norm": 1.2524114847183228,
281
+ "learning_rate": 9.998107687592326e-06,
282
+ "loss": 1.1147,
283
+ "step": 370
284
+ },
285
+ {
286
+ "epoch": 0.3325311747976373,
287
+ "grad_norm": 1.4711254835128784,
288
+ "learning_rate": 9.996446595031691e-06,
289
+ "loss": 1.1182,
290
+ "step": 380
291
+ },
292
+ {
293
+ "epoch": 0.34128199518704877,
294
+ "grad_norm": 1.3600913286209106,
295
+ "learning_rate": 9.994266690261928e-06,
296
+ "loss": 1.1565,
297
+ "step": 390
298
+ },
299
+ {
300
+ "epoch": 0.35003281557646027,
301
+ "grad_norm": 1.3752756118774414,
302
+ "learning_rate": 9.991568199636149e-06,
303
+ "loss": 1.0803,
304
+ "step": 400
305
+ },
306
+ {
307
+ "epoch": 0.3587836359658718,
308
+ "grad_norm": 1.2283236980438232,
309
+ "learning_rate": 9.988351403355455e-06,
310
+ "loss": 1.1371,
311
+ "step": 410
312
+ },
313
+ {
314
+ "epoch": 0.3675344563552833,
315
+ "grad_norm": 1.266762614250183,
316
+ "learning_rate": 9.98461663543987e-06,
317
+ "loss": 1.1027,
318
+ "step": 420
319
+ },
320
+ {
321
+ "epoch": 0.3762852767446948,
322
+ "grad_norm": 1.2771412134170532,
323
+ "learning_rate": 9.980364283693638e-06,
324
+ "loss": 1.0891,
325
+ "step": 430
326
+ },
327
+ {
328
+ "epoch": 0.3850360971341063,
329
+ "grad_norm": 1.2226617336273193,
330
+ "learning_rate": 9.97559478966496e-06,
331
+ "loss": 1.1105,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.39378691752351785,
336
+ "grad_norm": 1.2738267183303833,
337
+ "learning_rate": 9.970308648600142e-06,
338
+ "loss": 1.114,
339
+ "step": 450
340
+ },
341
+ {
342
+ "epoch": 0.40253773791292935,
343
+ "grad_norm": 1.3344817161560059,
344
+ "learning_rate": 9.964506409392178e-06,
345
+ "loss": 1.1178,
346
+ "step": 460
347
+ },
348
+ {
349
+ "epoch": 0.41128855830234085,
350
+ "grad_norm": 1.3076369762420654,
351
+ "learning_rate": 9.958188674523748e-06,
352
+ "loss": 1.1205,
353
+ "step": 470
354
+ },
355
+ {
356
+ "epoch": 0.42003937869175234,
357
+ "grad_norm": 1.251315712928772,
358
+ "learning_rate": 9.951356100004666e-06,
359
+ "loss": 1.0795,
360
+ "step": 480
361
+ },
362
+ {
363
+ "epoch": 0.42879019908116384,
364
+ "grad_norm": 1.1784945726394653,
365
+ "learning_rate": 9.94400939530375e-06,
366
+ "loss": 1.1175,
367
+ "step": 490
368
+ },
369
+ {
370
+ "epoch": 0.4375410194705754,
371
+ "grad_norm": 1.219256043434143,
372
+ "learning_rate": 9.93614932327517e-06,
373
+ "loss": 1.0933,
374
+ "step": 500
375
+ },
376
+ {
377
+ "epoch": 0.4462918398599869,
378
+ "grad_norm": 1.398763656616211,
379
+ "learning_rate": 9.927776700079217e-06,
380
+ "loss": 1.1267,
381
+ "step": 510
382
+ },
383
+ {
384
+ "epoch": 0.4550426602493984,
385
+ "grad_norm": 1.25184166431427,
386
+ "learning_rate": 9.918892395097572e-06,
387
+ "loss": 1.1191,
388
+ "step": 520
389
+ },
390
+ {
391
+ "epoch": 0.4637934806388099,
392
+ "grad_norm": 1.3422456979751587,
393
+ "learning_rate": 9.90949733084303e-06,
394
+ "loss": 1.1003,
395
+ "step": 530
396
+ },
397
+ {
398
+ "epoch": 0.47254430102822137,
399
+ "grad_norm": 1.3312543630599976,
400
+ "learning_rate": 9.8995924828637e-06,
401
+ "loss": 1.1004,
402
+ "step": 540
403
+ },
404
+ {
405
+ "epoch": 0.4812951214176329,
406
+ "grad_norm": 1.2708804607391357,
407
+ "learning_rate": 9.889178879641717e-06,
408
+ "loss": 1.1094,
409
+ "step": 550
410
+ },
411
+ {
412
+ "epoch": 0.4900459418070444,
413
+ "grad_norm": 1.3565181493759155,
414
+ "learning_rate": 9.878257602486445e-06,
415
+ "loss": 1.1252,
416
+ "step": 560
417
+ },
418
+ {
419
+ "epoch": 0.4987967621964559,
420
+ "grad_norm": 1.232021450996399,
421
+ "learning_rate": 9.866829785422201e-06,
422
+ "loss": 1.0976,
423
+ "step": 570
424
+ },
425
+ {
426
+ "epoch": 0.5075475825858674,
427
+ "grad_norm": 1.371229887008667,
428
+ "learning_rate": 9.854896615070495e-06,
429
+ "loss": 1.0941,
430
+ "step": 580
431
+ },
432
+ {
433
+ "epoch": 0.516298402975279,
434
+ "grad_norm": 1.35117769241333,
435
+ "learning_rate": 9.842459330526826e-06,
436
+ "loss": 1.1127,
437
+ "step": 590
438
+ },
439
+ {
440
+ "epoch": 0.5250492233646904,
441
+ "grad_norm": 1.234413504600525,
442
+ "learning_rate": 9.829519223232006e-06,
443
+ "loss": 1.09,
444
+ "step": 600
445
+ },
446
+ {
447
+ "epoch": 0.533800043754102,
448
+ "grad_norm": 1.2576864957809448,
449
+ "learning_rate": 9.816077636838079e-06,
450
+ "loss": 1.1094,
451
+ "step": 610
452
+ },
453
+ {
454
+ "epoch": 0.5425508641435135,
455
+ "grad_norm": 1.340950608253479,
456
+ "learning_rate": 9.802135967068783e-06,
457
+ "loss": 1.0889,
458
+ "step": 620
459
+ },
460
+ {
461
+ "epoch": 0.5513016845329249,
462
+ "grad_norm": 1.350095272064209,
463
+ "learning_rate": 9.787695661574632e-06,
464
+ "loss": 1.0661,
465
+ "step": 630
466
+ },
467
+ {
468
+ "epoch": 0.5600525049223365,
469
+ "grad_norm": 1.2629624605178833,
470
+ "learning_rate": 9.772758219782601e-06,
471
+ "loss": 1.0823,
472
+ "step": 640
473
+ },
474
+ {
475
+ "epoch": 0.5688033253117479,
476
+ "grad_norm": 1.234574556350708,
477
+ "learning_rate": 9.757325192740425e-06,
478
+ "loss": 1.0937,
479
+ "step": 650
480
+ },
481
+ {
482
+ "epoch": 0.5775541457011595,
483
+ "grad_norm": 1.1775031089782715,
484
+ "learning_rate": 9.741398182955548e-06,
485
+ "loss": 1.1066,
486
+ "step": 660
487
+ },
488
+ {
489
+ "epoch": 0.586304966090571,
490
+ "grad_norm": 1.2653073072433472,
491
+ "learning_rate": 9.724978844228717e-06,
492
+ "loss": 1.0964,
493
+ "step": 670
494
+ },
495
+ {
496
+ "epoch": 0.5950557864799825,
497
+ "grad_norm": 1.2665278911590576,
498
+ "learning_rate": 9.70806888148227e-06,
499
+ "loss": 1.1007,
500
+ "step": 680
501
+ },
502
+ {
503
+ "epoch": 0.603806606869394,
504
+ "grad_norm": 1.2909613847732544,
505
+ "learning_rate": 9.690670050583095e-06,
506
+ "loss": 1.0973,
507
+ "step": 690
508
+ },
509
+ {
510
+ "epoch": 0.6125574272588055,
511
+ "grad_norm": 1.2396572828292847,
512
+ "learning_rate": 9.672784158160307e-06,
513
+ "loss": 1.0423,
514
+ "step": 700
515
+ },
516
+ {
517
+ "epoch": 0.621308247648217,
518
+ "grad_norm": 1.2569255828857422,
519
+ "learning_rate": 9.654413061417655e-06,
520
+ "loss": 1.0764,
521
+ "step": 710
522
+ },
523
+ {
524
+ "epoch": 0.6300590680376286,
525
+ "grad_norm": 1.3747756481170654,
526
+ "learning_rate": 9.635558667940683e-06,
527
+ "loss": 1.0563,
528
+ "step": 720
529
+ },
530
+ {
531
+ "epoch": 0.63880988842704,
532
+ "grad_norm": 1.2752923965454102,
533
+ "learning_rate": 9.616222935498647e-06,
534
+ "loss": 1.1084,
535
+ "step": 730
536
+ },
537
+ {
538
+ "epoch": 0.6475607088164516,
539
+ "grad_norm": 1.218320369720459,
540
+ "learning_rate": 9.596407871841229e-06,
541
+ "loss": 1.0674,
542
+ "step": 740
543
+ },
544
+ {
545
+ "epoch": 0.656311529205863,
546
+ "grad_norm": 1.2466013431549072,
547
+ "learning_rate": 9.57611553449006e-06,
548
+ "loss": 1.1048,
549
+ "step": 750
550
+ },
551
+ {
552
+ "epoch": 0.6650623495952745,
553
+ "grad_norm": 1.2956342697143555,
554
+ "learning_rate": 9.555348030525076e-06,
555
+ "loss": 1.0723,
556
+ "step": 760
557
+ },
558
+ {
559
+ "epoch": 0.6738131699846861,
560
+ "grad_norm": 1.29911208152771,
561
+ "learning_rate": 9.534107516365725e-06,
562
+ "loss": 1.0743,
563
+ "step": 770
564
+ },
565
+ {
566
+ "epoch": 0.6825639903740975,
567
+ "grad_norm": 1.2063359022140503,
568
+ "learning_rate": 9.512396197547053e-06,
569
+ "loss": 1.0717,
570
+ "step": 780
571
+ },
572
+ {
573
+ "epoch": 0.6913148107635091,
574
+ "grad_norm": 1.184462547302246,
575
+ "learning_rate": 9.490216328490693e-06,
576
+ "loss": 1.0895,
577
+ "step": 790
578
+ },
579
+ {
580
+ "epoch": 0.7000656311529205,
581
+ "grad_norm": 1.186874270439148,
582
+ "learning_rate": 9.467570212270767e-06,
583
+ "loss": 1.1476,
584
+ "step": 800
585
+ },
586
+ {
587
+ "epoch": 0.7088164515423321,
588
+ "grad_norm": 1.2901315689086914,
589
+ "learning_rate": 9.444460200374747e-06,
590
+ "loss": 1.0538,
591
+ "step": 810
592
+ },
593
+ {
594
+ "epoch": 0.7175672719317436,
595
+ "grad_norm": 1.2175313234329224,
596
+ "learning_rate": 9.42088869245929e-06,
597
+ "loss": 1.079,
598
+ "step": 820
599
+ },
600
+ {
601
+ "epoch": 0.7263180923211551,
602
+ "grad_norm": 1.1771786212921143,
603
+ "learning_rate": 9.396858136101066e-06,
604
+ "loss": 1.0756,
605
+ "step": 830
606
+ },
607
+ {
608
+ "epoch": 0.7350689127105666,
609
+ "grad_norm": 1.2814205884933472,
610
+ "learning_rate": 9.372371026542595e-06,
611
+ "loss": 1.0697,
612
+ "step": 840
613
+ },
614
+ {
615
+ "epoch": 0.7438197330999782,
616
+ "grad_norm": 1.2603764533996582,
617
+ "learning_rate": 9.347429906433175e-06,
618
+ "loss": 1.1079,
619
+ "step": 850
620
+ },
621
+ {
622
+ "epoch": 0.7525705534893896,
623
+ "grad_norm": 1.2228299379348755,
624
+ "learning_rate": 9.322037365564844e-06,
625
+ "loss": 1.0746,
626
+ "step": 860
627
+ },
628
+ {
629
+ "epoch": 0.7613213738788012,
630
+ "grad_norm": 1.1895763874053955,
631
+ "learning_rate": 9.296196040603476e-06,
632
+ "loss": 1.0831,
633
+ "step": 870
634
+ },
635
+ {
636
+ "epoch": 0.7700721942682126,
637
+ "grad_norm": 1.2288453578948975,
638
+ "learning_rate": 9.269908614814995e-06,
639
+ "loss": 1.0649,
640
+ "step": 880
641
+ },
642
+ {
643
+ "epoch": 0.7788230146576242,
644
+ "grad_norm": 1.1580345630645752,
645
+ "learning_rate": 9.243177817786757e-06,
646
+ "loss": 1.1053,
647
+ "step": 890
648
+ },
649
+ {
650
+ "epoch": 0.7875738350470357,
651
+ "grad_norm": 1.1613861322402954,
652
+ "learning_rate": 9.216006425144118e-06,
653
+ "loss": 1.0854,
654
+ "step": 900
655
+ },
656
+ {
657
+ "epoch": 0.7963246554364471,
658
+ "grad_norm": 1.0809310674667358,
659
+ "learning_rate": 9.188397258262221e-06,
660
+ "loss": 1.059,
661
+ "step": 910
662
+ },
663
+ {
664
+ "epoch": 0.8050754758258587,
665
+ "grad_norm": 1.2021125555038452,
666
+ "learning_rate": 9.160353183973045e-06,
667
+ "loss": 1.064,
668
+ "step": 920
669
+ },
670
+ {
671
+ "epoch": 0.8138262962152701,
672
+ "grad_norm": 1.286584496498108,
673
+ "learning_rate": 9.131877114267711e-06,
674
+ "loss": 1.0516,
675
+ "step": 930
676
+ },
677
+ {
678
+ "epoch": 0.8225771166046817,
679
+ "grad_norm": 1.177673578262329,
680
+ "learning_rate": 9.102972005994123e-06,
681
+ "loss": 1.0358,
682
+ "step": 940
683
+ },
684
+ {
685
+ "epoch": 0.8313279369940932,
686
+ "grad_norm": 1.2117091417312622,
687
+ "learning_rate": 9.073640860549931e-06,
688
+ "loss": 1.0701,
689
+ "step": 950
690
+ },
691
+ {
692
+ "epoch": 0.8400787573835047,
693
+ "grad_norm": 1.2018595933914185,
694
+ "learning_rate": 9.043886723570885e-06,
695
+ "loss": 1.0833,
696
+ "step": 960
697
+ },
698
+ {
699
+ "epoch": 0.8488295777729162,
700
+ "grad_norm": 1.1337984800338745,
701
+ "learning_rate": 9.013712684614582e-06,
702
+ "loss": 1.0785,
703
+ "step": 970
704
+ },
705
+ {
706
+ "epoch": 0.8575803981623277,
707
+ "grad_norm": 1.2217832803726196,
708
+ "learning_rate": 8.983121876839658e-06,
709
+ "loss": 1.0623,
710
+ "step": 980
711
+ },
712
+ {
713
+ "epoch": 0.8663312185517392,
714
+ "grad_norm": 1.2249287366867065,
715
+ "learning_rate": 8.95211747668046e-06,
716
+ "loss": 1.0285,
717
+ "step": 990
718
+ },
719
+ {
720
+ "epoch": 0.8750820389411508,
721
+ "grad_norm": 1.1673548221588135,
722
+ "learning_rate": 8.920702703517205e-06,
723
+ "loss": 1.0354,
724
+ "step": 1000
725
+ },
726
+ {
727
+ "epoch": 0.8838328593305622,
728
+ "grad_norm": 1.1298819780349731,
729
+ "learning_rate": 8.888880819341703e-06,
730
+ "loss": 1.0538,
731
+ "step": 1010
732
+ },
733
+ {
734
+ "epoch": 0.8925836797199738,
735
+ "grad_norm": 1.1600337028503418,
736
+ "learning_rate": 8.856655128418638e-06,
737
+ "loss": 1.0527,
738
+ "step": 1020
739
+ },
740
+ {
741
+ "epoch": 0.9013345001093852,
742
+ "grad_norm": 1.2292169332504272,
743
+ "learning_rate": 8.824028976942465e-06,
744
+ "loss": 1.0613,
745
+ "step": 1030
746
+ },
747
+ {
748
+ "epoch": 0.9100853204987968,
749
+ "grad_norm": 1.1878681182861328,
750
+ "learning_rate": 8.791005752689958e-06,
751
+ "loss": 1.0489,
752
+ "step": 1040
753
+ },
754
+ {
755
+ "epoch": 0.9188361408882083,
756
+ "grad_norm": 1.2165477275848389,
757
+ "learning_rate": 8.757588884668435e-06,
758
+ "loss": 1.0675,
759
+ "step": 1050
760
+ },
761
+ {
762
+ "epoch": 0.9275869612776197,
763
+ "grad_norm": 1.1682305335998535,
764
+ "learning_rate": 8.723781842759696e-06,
765
+ "loss": 1.084,
766
+ "step": 1060
767
+ },
768
+ {
769
+ "epoch": 0.9363377816670313,
770
+ "grad_norm": 1.3387882709503174,
771
+ "learning_rate": 8.689588137359731e-06,
772
+ "loss": 1.0564,
773
+ "step": 1070
774
+ },
775
+ {
776
+ "epoch": 0.9450886020564427,
777
+ "grad_norm": 1.2144585847854614,
778
+ "learning_rate": 8.655011319014217e-06,
779
+ "loss": 1.08,
780
+ "step": 1080
781
+ },
782
+ {
783
+ "epoch": 0.9538394224458543,
784
+ "grad_norm": 1.2637611627578735,
785
+ "learning_rate": 8.620054978049828e-06,
786
+ "loss": 1.0716,
787
+ "step": 1090
788
+ },
789
+ {
790
+ "epoch": 0.9625902428352658,
791
+ "grad_norm": 1.230635404586792,
792
+ "learning_rate": 8.584722744201448e-06,
793
+ "loss": 1.0429,
794
+ "step": 1100
795
+ },
796
+ {
797
+ "epoch": 0.9713410632246773,
798
+ "grad_norm": 1.2495629787445068,
799
+ "learning_rate": 8.549018286235255e-06,
800
+ "loss": 1.0464,
801
+ "step": 1110
802
+ },
803
+ {
804
+ "epoch": 0.9800918836140888,
805
+ "grad_norm": 1.1702367067337036,
806
+ "learning_rate": 8.512945311567786e-06,
807
+ "loss": 1.0476,
808
+ "step": 1120
809
+ },
810
+ {
811
+ "epoch": 0.9888427040035004,
812
+ "grad_norm": 1.1367738246917725,
813
+ "learning_rate": 8.476507565880959e-06,
814
+ "loss": 1.0392,
815
+ "step": 1130
816
+ },
817
+ {
818
+ "epoch": 0.9975935243929118,
819
+ "grad_norm": 1.1903005838394165,
820
+ "learning_rate": 8.439708832733142e-06,
821
+ "loss": 1.0414,
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 sha256:832596a9b4a95dc8d19b2429cc7735a9b11a9131d7452ae2298aef2b316bccb9
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 sha256:60c893da735e17a501e6be9aaeb5336723cc530fb430317d6d3552bbefc35d44
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 sha256:91d5f86d0e6b7aeaa34bd79d483ed1223ee8c6de98c9f813595f88257d4acd7b
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 sha256:3c5e39e82106cbd99b06cb0418b3d33aa1c4f4599d180a56ac89c756e75d2ffa
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 sha256:64175613ee851c54fb1e76fa75c85e3b89d192f625045a951c58b182d9099b3a
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 sha256:909e4a9349539cfe0c0eb9f9789a02d1be64fcdae99a5f750c52f82df3fcc175
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:ff0231e8ab7a33b95f87025fea9c359ccb0b935e46aad0604c48b1ecca178860
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 sha256:56f346fe28c1f50d92c00a48fb8f14d29a9bb47fa40371708b89eca2b1475d49
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:bedbf4535d816fa164640c49955149ac0cd0e01f2eeb5eb8f186cc67784f8528
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:b872ce25c1f25ccab23ed7ee5da4ea1a78d5bcd0709990eec6ea43a5b5c9ee7f
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 sha256:135d81a4bd51a021ff9c597843c18ead1042037e41e977db16b5209695d08db1
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": 0.8244,
15
+ "eval_samples_per_second": 24.261,
16
+ "eval_steps_per_second": 2.426,
17
+ "step": 0
18
+ },
19
+ {
20
+ "epoch": 0.0008750820389411508,
21
+ "grad_norm": 6.372805118560791,
22
+ "learning_rate": 2.915451895043732e-08,
23
+ "loss": 1.6434,
24
+ "step": 1
25
+ },
26
+ {
27
+ "epoch": 0.008750820389411508,
28
+ "grad_norm": 6.278440475463867,
29
+ "learning_rate": 2.915451895043732e-07,
30
+ "loss": 1.7188,
31
+ "step": 10
32
+ },
33
+ {
34
+ "epoch": 0.017501640778823015,
35
+ "grad_norm": 6.023682594299316,
36
+ "learning_rate": 5.830903790087464e-07,
37
+ "loss": 1.7154,
38
+ "step": 20
39
+ },
40
+ {
41
+ "epoch": 0.02625246116823452,
42
+ "grad_norm": 3.7778327465057373,
43
+ "learning_rate": 8.746355685131196e-07,
44
+ "loss": 1.6563,
45
+ "step": 30
46
+ },
47
+ {
48
+ "epoch": 0.03500328155764603,
49
+ "grad_norm": 3.8672306537628174,
50
+ "learning_rate": 1.1661807580174927e-06,
51
+ "loss": 1.5743,
52
+ "step": 40
53
+ },
54
+ {
55
+ "epoch": 0.04375410194705753,
56
+ "grad_norm": 2.53611159324646,
57
+ "learning_rate": 1.4577259475218661e-06,
58
+ "loss": 1.479,
59
+ "step": 50
60
+ },
61
+ {
62
+ "epoch": 0.05250492233646904,
63
+ "grad_norm": 2.019819974899292,
64
+ "learning_rate": 1.7492711370262391e-06,
65
+ "loss": 1.4242,
66
+ "step": 60
67
+ },
68
+ {
69
+ "epoch": 0.06125574272588055,
70
+ "grad_norm": 1.7553459405899048,
71
+ "learning_rate": 2.0408163265306125e-06,
72
+ "loss": 1.3795,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.07000656311529206,
77
+ "grad_norm": 1.4397658109664917,
78
+ "learning_rate": 2.3323615160349855e-06,
79
+ "loss": 1.3745,
80
+ "step": 80
81
+ },
82
+ {
83
+ "epoch": 0.07875738350470357,
84
+ "grad_norm": 1.5793251991271973,
85
+ "learning_rate": 2.6239067055393585e-06,
86
+ "loss": 1.3529,
87
+ "step": 90
88
+ },
89
+ {
90
+ "epoch": 0.08750820389411507,
91
+ "grad_norm": 1.5069832801818848,
92
+ "learning_rate": 2.9154518950437323e-06,
93
+ "loss": 1.3459,
94
+ "step": 100
95
+ },
96
+ {
97
+ "epoch": 0.09625902428352658,
98
+ "grad_norm": 1.4921239614486694,
99
+ "learning_rate": 3.2069970845481052e-06,
100
+ "loss": 1.2743,
101
+ "step": 110
102
+ },
103
+ {
104
+ "epoch": 0.10500984467293809,
105
+ "grad_norm": 1.4761064052581787,
106
+ "learning_rate": 3.4985422740524782e-06,
107
+ "loss": 1.2959,
108
+ "step": 120
109
+ },
110
+ {
111
+ "epoch": 0.1137606650623496,
112
+ "grad_norm": 1.4724879264831543,
113
+ "learning_rate": 3.790087463556852e-06,
114
+ "loss": 1.2773,
115
+ "step": 130
116
+ },
117
+ {
118
+ "epoch": 0.1225114854517611,
119
+ "grad_norm": 1.4954681396484375,
120
+ "learning_rate": 4.081632653061225e-06,
121
+ "loss": 1.2695,
122
+ "step": 140
123
+ },
124
+ {
125
+ "epoch": 0.1312623058411726,
126
+ "grad_norm": 1.4220881462097168,
127
+ "learning_rate": 4.3731778425655976e-06,
128
+ "loss": 1.2668,
129
+ "step": 150
130
+ },
131
+ {
132
+ "epoch": 0.14001312623058412,
133
+ "grad_norm": 1.4039027690887451,
134
+ "learning_rate": 4.664723032069971e-06,
135
+ "loss": 1.2231,
136
+ "step": 160
137
+ },
138
+ {
139
+ "epoch": 0.14876394661999562,
140
+ "grad_norm": 1.468996524810791,
141
+ "learning_rate": 4.956268221574344e-06,
142
+ "loss": 1.221,
143
+ "step": 170
144
+ },
145
+ {
146
+ "epoch": 0.15751476700940714,
147
+ "grad_norm": 1.5780359506607056,
148
+ "learning_rate": 5.247813411078717e-06,
149
+ "loss": 1.2403,
150
+ "step": 180
151
+ },
152
+ {
153
+ "epoch": 0.16626558739881864,
154
+ "grad_norm": 1.4090375900268555,
155
+ "learning_rate": 5.539358600583091e-06,
156
+ "loss": 1.2021,
157
+ "step": 190
158
+ },
159
+ {
160
+ "epoch": 0.17501640778823013,
161
+ "grad_norm": 1.489652395248413,
162
+ "learning_rate": 5.8309037900874645e-06,
163
+ "loss": 1.1731,
164
+ "step": 200
165
+ },
166
+ {
167
+ "epoch": 0.18376722817764166,
168
+ "grad_norm": 1.3121322393417358,
169
+ "learning_rate": 6.122448979591837e-06,
170
+ "loss": 1.1924,
171
+ "step": 210
172
+ },
173
+ {
174
+ "epoch": 0.19251804856705315,
175
+ "grad_norm": 1.3954280614852905,
176
+ "learning_rate": 6.4139941690962105e-06,
177
+ "loss": 1.2121,
178
+ "step": 220
179
+ },
180
+ {
181
+ "epoch": 0.20126886895646467,
182
+ "grad_norm": 1.4191243648529053,
183
+ "learning_rate": 6.705539358600584e-06,
184
+ "loss": 1.1973,
185
+ "step": 230
186
+ },
187
+ {
188
+ "epoch": 0.21001968934587617,
189
+ "grad_norm": 1.4373400211334229,
190
+ "learning_rate": 6.9970845481049564e-06,
191
+ "loss": 1.1701,
192
+ "step": 240
193
+ },
194
+ {
195
+ "epoch": 0.2187705097352877,
196
+ "grad_norm": 1.4669713973999023,
197
+ "learning_rate": 7.28862973760933e-06,
198
+ "loss": 1.1923,
199
+ "step": 250
200
+ },
201
+ {
202
+ "epoch": 0.2275213301246992,
203
+ "grad_norm": 1.4269640445709229,
204
+ "learning_rate": 7.580174927113704e-06,
205
+ "loss": 1.1713,
206
+ "step": 260
207
+ },
208
+ {
209
+ "epoch": 0.23627215051411068,
210
+ "grad_norm": 1.7535545825958252,
211
+ "learning_rate": 7.871720116618077e-06,
212
+ "loss": 1.1891,
213
+ "step": 270
214
+ },
215
+ {
216
+ "epoch": 0.2450229709035222,
217
+ "grad_norm": 1.599094271659851,
218
+ "learning_rate": 8.16326530612245e-06,
219
+ "loss": 1.1827,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2537737912929337,
224
+ "grad_norm": 1.8278971910476685,
225
+ "learning_rate": 8.454810495626823e-06,
226
+ "loss": 1.1402,
227
+ "step": 290
228
+ },
229
+ {
230
+ "epoch": 0.2625246116823452,
231
+ "grad_norm": 1.5127216577529907,
232
+ "learning_rate": 8.746355685131195e-06,
233
+ "loss": 1.1597,
234
+ "step": 300
235
+ },
236
+ {
237
+ "epoch": 0.27127543207175675,
238
+ "grad_norm": 1.6015137434005737,
239
+ "learning_rate": 9.03790087463557e-06,
240
+ "loss": 1.1398,
241
+ "step": 310
242
+ },
243
+ {
244
+ "epoch": 0.28002625246116825,
245
+ "grad_norm": 1.3944435119628906,
246
+ "learning_rate": 9.329446064139942e-06,
247
+ "loss": 1.1174,
248
+ "step": 320
249
+ },
250
+ {
251
+ "epoch": 0.28877707285057974,
252
+ "grad_norm": 1.2441927194595337,
253
+ "learning_rate": 9.620991253644316e-06,
254
+ "loss": 1.1416,
255
+ "step": 330
256
+ },
257
+ {
258
+ "epoch": 0.29752789323999124,
259
+ "grad_norm": 1.4111624956130981,
260
+ "learning_rate": 9.912536443148689e-06,
261
+ "loss": 1.1352,
262
+ "step": 340
263
+ },
264
+ {
265
+ "epoch": 0.30627871362940273,
266
+ "grad_norm": 1.3007830381393433,
267
+ "learning_rate": 9.999872800049233e-06,
268
+ "loss": 1.1208,
269
+ "step": 350
270
+ },
271
+ {
272
+ "epoch": 0.3150295340188143,
273
+ "grad_norm": 1.3299254179000854,
274
+ "learning_rate": 9.999249795462235e-06,
275
+ "loss": 1.142,
276
+ "step": 360
277
+ },
278
+ {
279
+ "epoch": 0.3237803544082258,
280
+ "grad_norm": 1.2524114847183228,
281
+ "learning_rate": 9.998107687592326e-06,
282
+ "loss": 1.1147,
283
+ "step": 370
284
+ },
285
+ {
286
+ "epoch": 0.3325311747976373,
287
+ "grad_norm": 1.4711254835128784,
288
+ "learning_rate": 9.996446595031691e-06,
289
+ "loss": 1.1182,
290
+ "step": 380
291
+ },
292
+ {
293
+ "epoch": 0.34128199518704877,
294
+ "grad_norm": 1.3600913286209106,
295
+ "learning_rate": 9.994266690261928e-06,
296
+ "loss": 1.1565,
297
+ "step": 390
298
+ },
299
+ {
300
+ "epoch": 0.35003281557646027,
301
+ "grad_norm": 1.3752756118774414,
302
+ "learning_rate": 9.991568199636149e-06,
303
+ "loss": 1.0803,
304
+ "step": 400
305
+ },
306
+ {
307
+ "epoch": 0.3587836359658718,
308
+ "grad_norm": 1.2283236980438232,
309
+ "learning_rate": 9.988351403355455e-06,
310
+ "loss": 1.1371,
311
+ "step": 410
312
+ },
313
+ {
314
+ "epoch": 0.3675344563552833,
315
+ "grad_norm": 1.266762614250183,
316
+ "learning_rate": 9.98461663543987e-06,
317
+ "loss": 1.1027,
318
+ "step": 420
319
+ },
320
+ {
321
+ "epoch": 0.3762852767446948,
322
+ "grad_norm": 1.2771412134170532,
323
+ "learning_rate": 9.980364283693638e-06,
324
+ "loss": 1.0891,
325
+ "step": 430
326
+ },
327
+ {
328
+ "epoch": 0.3850360971341063,
329
+ "grad_norm": 1.2226617336273193,
330
+ "learning_rate": 9.97559478966496e-06,
331
+ "loss": 1.1105,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.39378691752351785,
336
+ "grad_norm": 1.2738267183303833,
337
+ "learning_rate": 9.970308648600142e-06,
338
+ "loss": 1.114,
339
+ "step": 450
340
+ },
341
+ {
342
+ "epoch": 0.40253773791292935,
343
+ "grad_norm": 1.3344817161560059,
344
+ "learning_rate": 9.964506409392178e-06,
345
+ "loss": 1.1178,
346
+ "step": 460
347
+ },
348
+ {
349
+ "epoch": 0.41128855830234085,
350
+ "grad_norm": 1.3076369762420654,
351
+ "learning_rate": 9.958188674523748e-06,
352
+ "loss": 1.1205,
353
+ "step": 470
354
+ },
355
+ {
356
+ "epoch": 0.42003937869175234,
357
+ "grad_norm": 1.251315712928772,
358
+ "learning_rate": 9.951356100004666e-06,
359
+ "loss": 1.0795,
360
+ "step": 480
361
+ },
362
+ {
363
+ "epoch": 0.42879019908116384,
364
+ "grad_norm": 1.1784945726394653,
365
+ "learning_rate": 9.94400939530375e-06,
366
+ "loss": 1.1175,
367
+ "step": 490
368
+ },
369
+ {
370
+ "epoch": 0.4375410194705754,
371
+ "grad_norm": 1.219256043434143,
372
+ "learning_rate": 9.93614932327517e-06,
373
+ "loss": 1.0933,
374
+ "step": 500
375
+ },
376
+ {
377
+ "epoch": 0.4462918398599869,
378
+ "grad_norm": 1.398763656616211,
379
+ "learning_rate": 9.927776700079217e-06,
380
+ "loss": 1.1267,
381
+ "step": 510
382
+ },
383
+ {
384
+ "epoch": 0.4550426602493984,
385
+ "grad_norm": 1.25184166431427,
386
+ "learning_rate": 9.918892395097572e-06,
387
+ "loss": 1.1191,
388
+ "step": 520
389
+ },
390
+ {
391
+ "epoch": 0.4637934806388099,
392
+ "grad_norm": 1.3422456979751587,
393
+ "learning_rate": 9.90949733084303e-06,
394
+ "loss": 1.1003,
395
+ "step": 530
396
+ },
397
+ {
398
+ "epoch": 0.47254430102822137,
399
+ "grad_norm": 1.3312543630599976,
400
+ "learning_rate": 9.8995924828637e-06,
401
+ "loss": 1.1004,
402
+ "step": 540
403
+ },
404
+ {
405
+ "epoch": 0.4812951214176329,
406
+ "grad_norm": 1.2708804607391357,
407
+ "learning_rate": 9.889178879641717e-06,
408
+ "loss": 1.1094,
409
+ "step": 550
410
+ },
411
+ {
412
+ "epoch": 0.4900459418070444,
413
+ "grad_norm": 1.3565181493759155,
414
+ "learning_rate": 9.878257602486445e-06,
415
+ "loss": 1.1252,
416
+ "step": 560
417
+ },
418
+ {
419
+ "epoch": 0.4987967621964559,
420
+ "grad_norm": 1.232021450996399,
421
+ "learning_rate": 9.866829785422201e-06,
422
+ "loss": 1.0976,
423
+ "step": 570
424
+ },
425
+ {
426
+ "epoch": 0.5075475825858674,
427
+ "grad_norm": 1.371229887008667,
428
+ "learning_rate": 9.854896615070495e-06,
429
+ "loss": 1.0941,
430
+ "step": 580
431
+ },
432
+ {
433
+ "epoch": 0.516298402975279,
434
+ "grad_norm": 1.35117769241333,
435
+ "learning_rate": 9.842459330526826e-06,
436
+ "loss": 1.1127,
437
+ "step": 590
438
+ },
439
+ {
440
+ "epoch": 0.5250492233646904,
441
+ "grad_norm": 1.234413504600525,
442
+ "learning_rate": 9.829519223232006e-06,
443
+ "loss": 1.09,
444
+ "step": 600
445
+ },
446
+ {
447
+ "epoch": 0.533800043754102,
448
+ "grad_norm": 1.2576864957809448,
449
+ "learning_rate": 9.816077636838079e-06,
450
+ "loss": 1.1094,
451
+ "step": 610
452
+ },
453
+ {
454
+ "epoch": 0.5425508641435135,
455
+ "grad_norm": 1.340950608253479,
456
+ "learning_rate": 9.802135967068783e-06,
457
+ "loss": 1.0889,
458
+ "step": 620
459
+ },
460
+ {
461
+ "epoch": 0.5513016845329249,
462
+ "grad_norm": 1.350095272064209,
463
+ "learning_rate": 9.787695661574632e-06,
464
+ "loss": 1.0661,
465
+ "step": 630
466
+ },
467
+ {
468
+ "epoch": 0.5600525049223365,
469
+ "grad_norm": 1.2629624605178833,
470
+ "learning_rate": 9.772758219782601e-06,
471
+ "loss": 1.0823,
472
+ "step": 640
473
+ },
474
+ {
475
+ "epoch": 0.5688033253117479,
476
+ "grad_norm": 1.234574556350708,
477
+ "learning_rate": 9.757325192740425e-06,
478
+ "loss": 1.0937,
479
+ "step": 650
480
+ },
481
+ {
482
+ "epoch": 0.5775541457011595,
483
+ "grad_norm": 1.1775031089782715,
484
+ "learning_rate": 9.741398182955548e-06,
485
+ "loss": 1.1066,
486
+ "step": 660
487
+ },
488
+ {
489
+ "epoch": 0.586304966090571,
490
+ "grad_norm": 1.2653073072433472,
491
+ "learning_rate": 9.724978844228717e-06,
492
+ "loss": 1.0964,
493
+ "step": 670
494
+ },
495
+ {
496
+ "epoch": 0.5950557864799825,
497
+ "grad_norm": 1.2665278911590576,
498
+ "learning_rate": 9.70806888148227e-06,
499
+ "loss": 1.1007,
500
+ "step": 680
501
+ },
502
+ {
503
+ "epoch": 0.603806606869394,
504
+ "grad_norm": 1.2909613847732544,
505
+ "learning_rate": 9.690670050583095e-06,
506
+ "loss": 1.0973,
507
+ "step": 690
508
+ },
509
+ {
510
+ "epoch": 0.6125574272588055,
511
+ "grad_norm": 1.2396572828292847,
512
+ "learning_rate": 9.672784158160307e-06,
513
+ "loss": 1.0423,
514
+ "step": 700
515
+ },
516
+ {
517
+ "epoch": 0.621308247648217,
518
+ "grad_norm": 1.2569255828857422,
519
+ "learning_rate": 9.654413061417655e-06,
520
+ "loss": 1.0764,
521
+ "step": 710
522
+ },
523
+ {
524
+ "epoch": 0.6300590680376286,
525
+ "grad_norm": 1.3747756481170654,
526
+ "learning_rate": 9.635558667940683e-06,
527
+ "loss": 1.0563,
528
+ "step": 720
529
+ },
530
+ {
531
+ "epoch": 0.63880988842704,
532
+ "grad_norm": 1.2752923965454102,
533
+ "learning_rate": 9.616222935498647e-06,
534
+ "loss": 1.1084,
535
+ "step": 730
536
+ },
537
+ {
538
+ "epoch": 0.6475607088164516,
539
+ "grad_norm": 1.218320369720459,
540
+ "learning_rate": 9.596407871841229e-06,
541
+ "loss": 1.0674,
542
+ "step": 740
543
+ },
544
+ {
545
+ "epoch": 0.656311529205863,
546
+ "grad_norm": 1.2466013431549072,
547
+ "learning_rate": 9.57611553449006e-06,
548
+ "loss": 1.1048,
549
+ "step": 750
550
+ },
551
+ {
552
+ "epoch": 0.6650623495952745,
553
+ "grad_norm": 1.2956342697143555,
554
+ "learning_rate": 9.555348030525076e-06,
555
+ "loss": 1.0723,
556
+ "step": 760
557
+ },
558
+ {
559
+ "epoch": 0.6738131699846861,
560
+ "grad_norm": 1.29911208152771,
561
+ "learning_rate": 9.534107516365725e-06,
562
+ "loss": 1.0743,
563
+ "step": 770
564
+ },
565
+ {
566
+ "epoch": 0.6825639903740975,
567
+ "grad_norm": 1.2063359022140503,
568
+ "learning_rate": 9.512396197547053e-06,
569
+ "loss": 1.0717,
570
+ "step": 780
571
+ },
572
+ {
573
+ "epoch": 0.6913148107635091,
574
+ "grad_norm": 1.184462547302246,
575
+ "learning_rate": 9.490216328490693e-06,
576
+ "loss": 1.0895,
577
+ "step": 790
578
+ },
579
+ {
580
+ "epoch": 0.7000656311529205,
581
+ "grad_norm": 1.186874270439148,
582
+ "learning_rate": 9.467570212270767e-06,
583
+ "loss": 1.1476,
584
+ "step": 800
585
+ },
586
+ {
587
+ "epoch": 0.7088164515423321,
588
+ "grad_norm": 1.2901315689086914,
589
+ "learning_rate": 9.444460200374747e-06,
590
+ "loss": 1.0538,
591
+ "step": 810
592
+ },
593
+ {
594
+ "epoch": 0.7175672719317436,
595
+ "grad_norm": 1.2175313234329224,
596
+ "learning_rate": 9.42088869245929e-06,
597
+ "loss": 1.079,
598
+ "step": 820
599
+ },
600
+ {
601
+ "epoch": 0.7263180923211551,
602
+ "grad_norm": 1.1771786212921143,
603
+ "learning_rate": 9.396858136101066e-06,
604
+ "loss": 1.0756,
605
+ "step": 830
606
+ },
607
+ {
608
+ "epoch": 0.7350689127105666,
609
+ "grad_norm": 1.2814205884933472,
610
+ "learning_rate": 9.372371026542595e-06,
611
+ "loss": 1.0697,
612
+ "step": 840
613
+ },
614
+ {
615
+ "epoch": 0.7438197330999782,
616
+ "grad_norm": 1.2603764533996582,
617
+ "learning_rate": 9.347429906433175e-06,
618
+ "loss": 1.1079,
619
+ "step": 850
620
+ },
621
+ {
622
+ "epoch": 0.7525705534893896,
623
+ "grad_norm": 1.2228299379348755,
624
+ "learning_rate": 9.322037365564844e-06,
625
+ "loss": 1.0746,
626
+ "step": 860
627
+ },
628
+ {
629
+ "epoch": 0.7613213738788012,
630
+ "grad_norm": 1.1895763874053955,
631
+ "learning_rate": 9.296196040603476e-06,
632
+ "loss": 1.0831,
633
+ "step": 870
634
+ },
635
+ {
636
+ "epoch": 0.7700721942682126,
637
+ "grad_norm": 1.2288453578948975,
638
+ "learning_rate": 9.269908614814995e-06,
639
+ "loss": 1.0649,
640
+ "step": 880
641
+ },
642
+ {
643
+ "epoch": 0.7788230146576242,
644
+ "grad_norm": 1.1580345630645752,
645
+ "learning_rate": 9.243177817786757e-06,
646
+ "loss": 1.1053,
647
+ "step": 890
648
+ },
649
+ {
650
+ "epoch": 0.7875738350470357,
651
+ "grad_norm": 1.1613861322402954,
652
+ "learning_rate": 9.216006425144118e-06,
653
+ "loss": 1.0854,
654
+ "step": 900
655
+ },
656
+ {
657
+ "epoch": 0.7963246554364471,
658
+ "grad_norm": 1.0809310674667358,
659
+ "learning_rate": 9.188397258262221e-06,
660
+ "loss": 1.059,
661
+ "step": 910
662
+ },
663
+ {
664
+ "epoch": 0.8050754758258587,
665
+ "grad_norm": 1.2021125555038452,
666
+ "learning_rate": 9.160353183973045e-06,
667
+ "loss": 1.064,
668
+ "step": 920
669
+ },
670
+ {
671
+ "epoch": 0.8138262962152701,
672
+ "grad_norm": 1.286584496498108,
673
+ "learning_rate": 9.131877114267711e-06,
674
+ "loss": 1.0516,
675
+ "step": 930
676
+ },
677
+ {
678
+ "epoch": 0.8225771166046817,
679
+ "grad_norm": 1.177673578262329,
680
+ "learning_rate": 9.102972005994123e-06,
681
+ "loss": 1.0358,
682
+ "step": 940
683
+ },
684
+ {
685
+ "epoch": 0.8313279369940932,
686
+ "grad_norm": 1.2117091417312622,
687
+ "learning_rate": 9.073640860549931e-06,
688
+ "loss": 1.0701,
689
+ "step": 950
690
+ },
691
+ {
692
+ "epoch": 0.8400787573835047,
693
+ "grad_norm": 1.2018595933914185,
694
+ "learning_rate": 9.043886723570885e-06,
695
+ "loss": 1.0833,
696
+ "step": 960
697
+ },
698
+ {
699
+ "epoch": 0.8488295777729162,
700
+ "grad_norm": 1.1337984800338745,
701
+ "learning_rate": 9.013712684614582e-06,
702
+ "loss": 1.0785,
703
+ "step": 970
704
+ },
705
+ {
706
+ "epoch": 0.8575803981623277,
707
+ "grad_norm": 1.2217832803726196,
708
+ "learning_rate": 8.983121876839658e-06,
709
+ "loss": 1.0623,
710
+ "step": 980
711
+ },
712
+ {
713
+ "epoch": 0.8663312185517392,
714
+ "grad_norm": 1.2249287366867065,
715
+ "learning_rate": 8.95211747668046e-06,
716
+ "loss": 1.0285,
717
+ "step": 990
718
+ },
719
+ {
720
+ "epoch": 0.8750820389411508,
721
+ "grad_norm": 1.1673548221588135,
722
+ "learning_rate": 8.920702703517205e-06,
723
+ "loss": 1.0354,
724
+ "step": 1000
725
+ },
726
+ {
727
+ "epoch": 0.8838328593305622,
728
+ "grad_norm": 1.1298819780349731,
729
+ "learning_rate": 8.888880819341703e-06,
730
+ "loss": 1.0538,
731
+ "step": 1010
732
+ },
733
+ {
734
+ "epoch": 0.8925836797199738,
735
+ "grad_norm": 1.1600337028503418,
736
+ "learning_rate": 8.856655128418638e-06,
737
+ "loss": 1.0527,
738
+ "step": 1020
739
+ },
740
+ {
741
+ "epoch": 0.9013345001093852,
742
+ "grad_norm": 1.2292169332504272,
743
+ "learning_rate": 8.824028976942465e-06,
744
+ "loss": 1.0613,
745
+ "step": 1030
746
+ },
747
+ {
748
+ "epoch": 0.9100853204987968,
749
+ "grad_norm": 1.1878681182861328,
750
+ "learning_rate": 8.791005752689958e-06,
751
+ "loss": 1.0489,
752
+ "step": 1040
753
+ },
754
+ {
755
+ "epoch": 0.9188361408882083,
756
+ "grad_norm": 1.2165477275848389,
757
+ "learning_rate": 8.757588884668435e-06,
758
+ "loss": 1.0675,
759
+ "step": 1050
760
+ },
761
+ {
762
+ "epoch": 0.9275869612776197,
763
+ "grad_norm": 1.1682305335998535,
764
+ "learning_rate": 8.723781842759696e-06,
765
+ "loss": 1.084,
766
+ "step": 1060
767
+ },
768
+ {
769
+ "epoch": 0.9363377816670313,
770
+ "grad_norm": 1.3387882709503174,
771
+ "learning_rate": 8.689588137359731e-06,
772
+ "loss": 1.0564,
773
+ "step": 1070
774
+ },
775
+ {
776
+ "epoch": 0.9450886020564427,
777
+ "grad_norm": 1.2144585847854614,
778
+ "learning_rate": 8.655011319014217e-06,
779
+ "loss": 1.08,
780
+ "step": 1080
781
+ },
782
+ {
783
+ "epoch": 0.9538394224458543,
784
+ "grad_norm": 1.2637611627578735,
785
+ "learning_rate": 8.620054978049828e-06,
786
+ "loss": 1.0716,
787
+ "step": 1090
788
+ },
789
+ {
790
+ "epoch": 0.9625902428352658,
791
+ "grad_norm": 1.230635404586792,
792
+ "learning_rate": 8.584722744201448e-06,
793
+ "loss": 1.0429,
794
+ "step": 1100
795
+ },
796
+ {
797
+ "epoch": 0.9713410632246773,
798
+ "grad_norm": 1.2495629787445068,
799
+ "learning_rate": 8.549018286235255e-06,
800
+ "loss": 1.0464,
801
+ "step": 1110
802
+ },
803
+ {
804
+ "epoch": 0.9800918836140888,
805
+ "grad_norm": 1.1702367067337036,
806
+ "learning_rate": 8.512945311567786e-06,
807
+ "loss": 1.0476,
808
+ "step": 1120
809
+ },
810
+ {
811
+ "epoch": 0.9888427040035004,
812
+ "grad_norm": 1.1367738246917725,
813
+ "learning_rate": 8.476507565880959e-06,
814
+ "loss": 1.0392,
815
+ "step": 1130
816
+ },
817
+ {
818
+ "epoch": 0.9975935243929118,
819
+ "grad_norm": 1.1903005838394165,
820
+ "learning_rate": 8.439708832733142e-06,
821
+ "loss": 1.0414,
822
+ "step": 1140
823
+ },
824
+ {
825
+ "epoch": 1.006125574272588,
826
+ "grad_norm": 1.310563325881958,
827
+ "learning_rate": 8.402552933166288e-06,
828
+ "loss": 0.9966,
829
+ "step": 1150
830
+ },
831
+ {
832
+ "epoch": 1.0148763946619996,
833
+ "grad_norm": 1.2823761701583862,
834
+ "learning_rate": 8.36504372530916e-06,
835
+ "loss": 0.9961,
836
+ "step": 1160
837
+ },
838
+ {
839
+ "epoch": 1.023627215051411,
840
+ "grad_norm": 1.340758204460144,
841
+ "learning_rate": 8.327185103976731e-06,
842
+ "loss": 0.9904,
843
+ "step": 1170
844
+ },
845
+ {
846
+ "epoch": 1.0323780354408225,
847
+ "grad_norm": 1.2751260995864868,
848
+ "learning_rate": 8.288981000265748e-06,
849
+ "loss": 1.004,
850
+ "step": 1180
851
+ },
852
+ {
853
+ "epoch": 1.041128855830234,
854
+ "grad_norm": 1.1516119241714478,
855
+ "learning_rate": 8.25043538114655e-06,
856
+ "loss": 0.9757,
857
+ "step": 1190
858
+ },
859
+ {
860
+ "epoch": 1.0498796762196456,
861
+ "grad_norm": 1.206026315689087,
862
+ "learning_rate": 8.211552249051149e-06,
863
+ "loss": 0.9799,
864
+ "step": 1200
865
+ },
866
+ {
867
+ "epoch": 1.0586304966090572,
868
+ "grad_norm": 1.2714760303497314,
869
+ "learning_rate": 8.17233564145764e-06,
870
+ "loss": 0.9731,
871
+ "step": 1210
872
+ },
873
+ {
874
+ "epoch": 1.0673813169984685,
875
+ "grad_norm": 1.2047169208526611,
876
+ "learning_rate": 8.132789630470952e-06,
877
+ "loss": 0.971,
878
+ "step": 1220
879
+ },
880
+ {
881
+ "epoch": 1.07613213738788,
882
+ "grad_norm": 1.282832384109497,
883
+ "learning_rate": 8.092918322400028e-06,
884
+ "loss": 0.9695,
885
+ "step": 1230
886
+ },
887
+ {
888
+ "epoch": 1.0848829577772916,
889
+ "grad_norm": 1.2463730573654175,
890
+ "learning_rate": 8.05272585733143e-06,
891
+ "loss": 0.9727,
892
+ "step": 1240
893
+ },
894
+ {
895
+ "epoch": 1.0936337781667032,
896
+ "grad_norm": 1.230220079421997,
897
+ "learning_rate": 8.012216408699456e-06,
898
+ "loss": 0.9909,
899
+ "step": 1250
900
+ },
901
+ {
902
+ "epoch": 1.1023845985561147,
903
+ "grad_norm": 1.3110296726226807,
904
+ "learning_rate": 7.97139418285279e-06,
905
+ "loss": 0.9887,
906
+ "step": 1260
907
+ },
908
+ {
909
+ "epoch": 1.1111354189455263,
910
+ "grad_norm": 1.234381914138794,
911
+ "learning_rate": 7.930263418617714e-06,
912
+ "loss": 1.0014,
913
+ "step": 1270
914
+ },
915
+ {
916
+ "epoch": 1.1198862393349376,
917
+ "grad_norm": 1.2342872619628906,
918
+ "learning_rate": 7.888828386857985e-06,
919
+ "loss": 0.9795,
920
+ "step": 1280
921
+ },
922
+ {
923
+ "epoch": 1.1286370597243491,
924
+ "grad_norm": 1.279505729675293,
925
+ "learning_rate": 7.847093390031351e-06,
926
+ "loss": 0.9647,
927
+ "step": 1290
928
+ },
929
+ {
930
+ "epoch": 1.1373878801137607,
931
+ "grad_norm": 1.3248056173324585,
932
+ "learning_rate": 7.8050627617428e-06,
933
+ "loss": 0.9881,
934
+ "step": 1300
935
+ },
936
+ {
937
+ "epoch": 1.1461387005031722,
938
+ "grad_norm": 1.234870433807373,
939
+ "learning_rate": 7.76274086629458e-06,
940
+ "loss": 0.9728,
941
+ "step": 1310
942
+ },
943
+ {
944
+ "epoch": 1.1548895208925836,
945
+ "grad_norm": 1.2661969661712646,
946
+ "learning_rate": 7.72013209823303e-06,
947
+ "loss": 0.9716,
948
+ "step": 1320
949
+ },
950
+ {
951
+ "epoch": 1.1636403412819951,
952
+ "grad_norm": 1.180408239364624,
953
+ "learning_rate": 7.677240881892259e-06,
954
+ "loss": 0.9641,
955
+ "step": 1330
956
+ },
957
+ {
958
+ "epoch": 1.1723911616714067,
959
+ "grad_norm": 1.2939667701721191,
960
+ "learning_rate": 7.634071670934734e-06,
961
+ "loss": 0.9832,
962
+ "step": 1340
963
+ },
964
+ {
965
+ "epoch": 1.1811419820608182,
966
+ "grad_norm": 1.1289607286453247,
967
+ "learning_rate": 7.5906289478888475e-06,
968
+ "loss": 0.9733,
969
+ "step": 1350
970
+ },
971
+ {
972
+ "epoch": 1.1898928024502298,
973
+ "grad_norm": 1.2281969785690308,
974
+ "learning_rate": 7.546917223683453e-06,
975
+ "loss": 0.9645,
976
+ "step": 1360
977
+ },
978
+ {
979
+ "epoch": 1.1986436228396413,
980
+ "grad_norm": 1.226364254951477,
981
+ "learning_rate": 7.502941037179474e-06,
982
+ "loss": 0.9668,
983
+ "step": 1370
984
+ },
985
+ {
986
+ "epoch": 1.2073944432290526,
987
+ "grad_norm": 1.2580183744430542,
988
+ "learning_rate": 7.4587049546986e-06,
989
+ "loss": 1.0167,
990
+ "step": 1380
991
+ },
992
+ {
993
+ "epoch": 1.2161452636184642,
994
+ "grad_norm": 1.261915683746338,
995
+ "learning_rate": 7.414213569549145e-06,
996
+ "loss": 0.9896,
997
+ "step": 1390
998
+ },
999
+ {
1000
+ "epoch": 1.2248960840078758,
1001
+ "grad_norm": 1.2158194780349731,
1002
+ "learning_rate": 7.369471501549087e-06,
1003
+ "loss": 0.9893,
1004
+ "step": 1400
1005
+ },
1006
+ {
1007
+ "epoch": 1.2336469043972873,
1008
+ "grad_norm": 1.2699130773544312,
1009
+ "learning_rate": 7.324483396546371e-06,
1010
+ "loss": 1.0171,
1011
+ "step": 1410
1012
+ },
1013
+ {
1014
+ "epoch": 1.2423977247866986,
1015
+ "grad_norm": 1.1923282146453857,
1016
+ "learning_rate": 7.279253925936498e-06,
1017
+ "loss": 0.9802,
1018
+ "step": 1420
1019
+ },
1020
+ {
1021
+ "epoch": 1.2511485451761102,
1022
+ "grad_norm": 1.1498876810073853,
1023
+ "learning_rate": 7.233787786177464e-06,
1024
+ "loss": 0.9683,
1025
+ "step": 1430
1026
+ },
1027
+ {
1028
+ "epoch": 1.2598993655655217,
1029
+ "grad_norm": 1.2153609991073608,
1030
+ "learning_rate": 7.188089698302097e-06,
1031
+ "loss": 0.9916,
1032
+ "step": 1440
1033
+ },
1034
+ {
1035
+ "epoch": 1.2686501859549333,
1036
+ "grad_norm": 1.1934350728988647,
1037
+ "learning_rate": 7.142164407427849e-06,
1038
+ "loss": 0.9788,
1039
+ "step": 1450
1040
+ },
1041
+ {
1042
+ "epoch": 1.2774010063443448,
1043
+ "grad_norm": 1.2987148761749268,
1044
+ "learning_rate": 7.096016682264074e-06,
1045
+ "loss": 0.9865,
1046
+ "step": 1460
1047
+ },
1048
+ {
1049
+ "epoch": 1.2861518267337564,
1050
+ "grad_norm": 1.2581541538238525,
1051
+ "learning_rate": 7.049651314616862e-06,
1052
+ "loss": 0.9494,
1053
+ "step": 1470
1054
+ },
1055
+ {
1056
+ "epoch": 1.2949026471231677,
1057
+ "grad_norm": 1.257711410522461,
1058
+ "learning_rate": 7.003073118891487e-06,
1059
+ "loss": 0.9534,
1060
+ "step": 1480
1061
+ },
1062
+ {
1063
+ "epoch": 1.3036534675125793,
1064
+ "grad_norm": 1.3140227794647217,
1065
+ "learning_rate": 6.956286931592481e-06,
1066
+ "loss": 0.9888,
1067
+ "step": 1490
1068
+ },
1069
+ {
1070
+ "epoch": 1.3124042879019908,
1071
+ "grad_norm": 1.2232003211975098,
1072
+ "learning_rate": 6.909297610821444e-06,
1073
+ "loss": 0.9632,
1074
+ "step": 1500
1075
+ },
1076
+ {
1077
+ "epoch": 1.3211551082914024,
1078
+ "grad_norm": 1.190757393836975,
1079
+ "learning_rate": 6.86211003577259e-06,
1080
+ "loss": 0.9598,
1081
+ "step": 1510
1082
+ },
1083
+ {
1084
+ "epoch": 1.3299059286808137,
1085
+ "grad_norm": 1.180031657218933,
1086
+ "learning_rate": 6.81472910622611e-06,
1087
+ "loss": 0.9804,
1088
+ "step": 1520
1089
+ },
1090
+ {
1091
+ "epoch": 1.3386567490702252,
1092
+ "grad_norm": 1.2404285669326782,
1093
+ "learning_rate": 6.7671597420394e-06,
1094
+ "loss": 0.9915,
1095
+ "step": 1530
1096
+ },
1097
+ {
1098
+ "epoch": 1.3474075694596368,
1099
+ "grad_norm": 1.2322285175323486,
1100
+ "learning_rate": 6.719406882636196e-06,
1101
+ "loss": 0.9908,
1102
+ "step": 1540
1103
+ },
1104
+ {
1105
+ "epoch": 1.3561583898490484,
1106
+ "grad_norm": 1.1908921003341675,
1107
+ "learning_rate": 6.671475486493691e-06,
1108
+ "loss": 0.9716,
1109
+ "step": 1550
1110
+ },
1111
+ {
1112
+ "epoch": 1.36490921023846,
1113
+ "grad_norm": 1.2217786312103271,
1114
+ "learning_rate": 6.623370530627661e-06,
1115
+ "loss": 0.9602,
1116
+ "step": 1560
1117
+ },
1118
+ {
1119
+ "epoch": 1.3736600306278715,
1120
+ "grad_norm": 1.2622575759887695,
1121
+ "learning_rate": 6.575097010075665e-06,
1122
+ "loss": 0.9509,
1123
+ "step": 1570
1124
+ },
1125
+ {
1126
+ "epoch": 1.382410851017283,
1127
+ "grad_norm": 1.2419332265853882,
1128
+ "learning_rate": 6.526659937378392e-06,
1129
+ "loss": 0.9821,
1130
+ "step": 1580
1131
+ },
1132
+ {
1133
+ "epoch": 1.3911616714066943,
1134
+ "grad_norm": 1.3119240999221802,
1135
+ "learning_rate": 6.478064342059165e-06,
1136
+ "loss": 0.9823,
1137
+ "step": 1590
1138
+ },
1139
+ {
1140
+ "epoch": 1.3999124917961059,
1141
+ "grad_norm": 1.3174080848693848,
1142
+ "learning_rate": 6.429315270101702e-06,
1143
+ "loss": 0.9844,
1144
+ "step": 1600
1145
+ },
1146
+ {
1147
+ "epoch": 1.4086633121855174,
1148
+ "grad_norm": 1.2387707233428955,
1149
+ "learning_rate": 6.380417783426154e-06,
1150
+ "loss": 0.9545,
1151
+ "step": 1610
1152
+ },
1153
+ {
1154
+ "epoch": 1.4174141325749288,
1155
+ "grad_norm": 1.1914652585983276,
1156
+ "learning_rate": 6.331376959363501e-06,
1157
+ "loss": 0.9792,
1158
+ "step": 1620
1159
+ },
1160
+ {
1161
+ "epoch": 1.4261649529643403,
1162
+ "grad_norm": 1.2240160703659058,
1163
+ "learning_rate": 6.2821978901283386e-06,
1164
+ "loss": 0.9573,
1165
+ "step": 1630
1166
+ },
1167
+ {
1168
+ "epoch": 1.4349157733537519,
1169
+ "grad_norm": 1.2401772737503052,
1170
+ "learning_rate": 6.232885682290119e-06,
1171
+ "loss": 0.9669,
1172
+ "step": 1640
1173
+ },
1174
+ {
1175
+ "epoch": 1.4436665937431634,
1176
+ "grad_norm": 1.2027502059936523,
1177
+ "learning_rate": 6.183445456242905e-06,
1178
+ "loss": 0.9815,
1179
+ "step": 1650
1180
+ },
1181
+ {
1182
+ "epoch": 1.452417414132575,
1183
+ "grad_norm": 1.2344744205474854,
1184
+ "learning_rate": 6.13388234567369e-06,
1185
+ "loss": 0.9361,
1186
+ "step": 1660
1187
+ },
1188
+ {
1189
+ "epoch": 1.4611682345219865,
1190
+ "grad_norm": 1.2103825807571411,
1191
+ "learning_rate": 6.084201497029333e-06,
1192
+ "loss": 0.9744,
1193
+ "step": 1670
1194
+ },
1195
+ {
1196
+ "epoch": 1.469919054911398,
1197
+ "grad_norm": 1.3063782453536987,
1198
+ "learning_rate": 6.034408068982171e-06,
1199
+ "loss": 0.9681,
1200
+ "step": 1680
1201
+ },
1202
+ {
1203
+ "epoch": 1.4786698753008094,
1204
+ "grad_norm": 1.2130053043365479,
1205
+ "learning_rate": 5.984507231894364e-06,
1206
+ "loss": 0.9586,
1207
+ "step": 1690
1208
+ },
1209
+ {
1210
+ "epoch": 1.487420695690221,
1211
+ "grad_norm": 1.2866191864013672,
1212
+ "learning_rate": 5.934504167281022e-06,
1213
+ "loss": 0.9681,
1214
+ "step": 1700
1215
+ },
1216
+ {
1217
+ "epoch": 1.4961715160796325,
1218
+ "grad_norm": 1.3027279376983643,
1219
+ "learning_rate": 5.884404067272175e-06,
1220
+ "loss": 0.975,
1221
+ "step": 1710
1222
+ },
1223
+ {
1224
+ "epoch": 1.5049223364690438,
1225
+ "grad_norm": 1.2601852416992188,
1226
+ "learning_rate": 5.834212134073645e-06,
1227
+ "loss": 0.9388,
1228
+ "step": 1720
1229
+ },
1230
+ {
1231
+ "epoch": 1.5136731568584554,
1232
+ "grad_norm": 1.2413978576660156,
1233
+ "learning_rate": 5.783933579426867e-06,
1234
+ "loss": 0.9723,
1235
+ "step": 1730
1236
+ },
1237
+ {
1238
+ "epoch": 1.522423977247867,
1239
+ "grad_norm": 1.2116490602493286,
1240
+ "learning_rate": 5.733573624067717e-06,
1241
+ "loss": 0.973,
1242
+ "step": 1740
1243
+ },
1244
+ {
1245
+ "epoch": 1.5311747976372785,
1246
+ "grad_norm": 1.2060950994491577,
1247
+ "learning_rate": 5.683137497184424e-06,
1248
+ "loss": 0.9375,
1249
+ "step": 1750
1250
+ },
1251
+ {
1252
+ "epoch": 1.53992561802669,
1253
+ "grad_norm": 1.2472314834594727,
1254
+ "learning_rate": 5.63263043587457e-06,
1255
+ "loss": 1.0112,
1256
+ "step": 1760
1257
+ },
1258
+ {
1259
+ "epoch": 1.5486764384161016,
1260
+ "grad_norm": 1.2653073072433472,
1261
+ "learning_rate": 5.582057684601308e-06,
1262
+ "loss": 0.9778,
1263
+ "step": 1770
1264
+ },
1265
+ {
1266
+ "epoch": 1.5574272588055131,
1267
+ "grad_norm": 1.2270764112472534,
1268
+ "learning_rate": 5.531424494648789e-06,
1269
+ "loss": 0.9807,
1270
+ "step": 1780
1271
+ },
1272
+ {
1273
+ "epoch": 1.5661780791949247,
1274
+ "grad_norm": 1.2816975116729736,
1275
+ "learning_rate": 5.480736123576886e-06,
1276
+ "loss": 0.9847,
1277
+ "step": 1790
1278
+ },
1279
+ {
1280
+ "epoch": 1.574928899584336,
1281
+ "grad_norm": 1.2429735660552979,
1282
+ "learning_rate": 5.429997834675276e-06,
1283
+ "loss": 0.9886,
1284
+ "step": 1800
1285
+ },
1286
+ {
1287
+ "epoch": 1.5836797199737476,
1288
+ "grad_norm": 1.3056929111480713,
1289
+ "learning_rate": 5.379214896416908e-06,
1290
+ "loss": 0.9514,
1291
+ "step": 1810
1292
+ },
1293
+ {
1294
+ "epoch": 1.592430540363159,
1295
+ "grad_norm": 1.155527949333191,
1296
+ "learning_rate": 5.328392581910962e-06,
1297
+ "loss": 0.9593,
1298
+ "step": 1820
1299
+ },
1300
+ {
1301
+ "epoch": 1.6011813607525704,
1302
+ "grad_norm": 1.2867212295532227,
1303
+ "learning_rate": 5.277536168355293e-06,
1304
+ "loss": 0.9722,
1305
+ "step": 1830
1306
+ },
1307
+ {
1308
+ "epoch": 1.609932181141982,
1309
+ "grad_norm": 1.3213430643081665,
1310
+ "learning_rate": 5.226650936488474e-06,
1311
+ "loss": 0.9588,
1312
+ "step": 1840
1313
+ },
1314
+ {
1315
+ "epoch": 1.6186830015313936,
1316
+ "grad_norm": 1.2772150039672852,
1317
+ "learning_rate": 5.1757421700414645e-06,
1318
+ "loss": 0.9729,
1319
+ "step": 1850
1320
+ },
1321
+ {
1322
+ "epoch": 1.627433821920805,
1323
+ "grad_norm": 1.2589421272277832,
1324
+ "learning_rate": 5.124815155188966e-06,
1325
+ "loss": 1.0171,
1326
+ "step": 1860
1327
+ },
1328
+ {
1329
+ "epoch": 1.6361846423102167,
1330
+ "grad_norm": 1.2326509952545166,
1331
+ "learning_rate": 5.073875180000523e-06,
1332
+ "loss": 0.9622,
1333
+ "step": 1870
1334
+ },
1335
+ {
1336
+ "epoch": 1.6449354626996282,
1337
+ "grad_norm": 1.1593009233474731,
1338
+ "learning_rate": 5.022927533891434e-06,
1339
+ "loss": 0.9584,
1340
+ "step": 1880
1341
+ },
1342
+ {
1343
+ "epoch": 1.6536862830890398,
1344
+ "grad_norm": 1.303597331047058,
1345
+ "learning_rate": 4.97197750707352e-06,
1346
+ "loss": 0.9994,
1347
+ "step": 1890
1348
+ },
1349
+ {
1350
+ "epoch": 1.662437103478451,
1351
+ "grad_norm": 1.2183244228363037,
1352
+ "learning_rate": 4.9210303900058025e-06,
1353
+ "loss": 0.9582,
1354
+ "step": 1900
1355
+ },
1356
+ {
1357
+ "epoch": 1.6711879238678626,
1358
+ "grad_norm": 1.2271572351455688,
1359
+ "learning_rate": 4.870091472845166e-06,
1360
+ "loss": 0.9775,
1361
+ "step": 1910
1362
+ },
1363
+ {
1364
+ "epoch": 1.679938744257274,
1365
+ "grad_norm": 1.222212791442871,
1366
+ "learning_rate": 4.819166044897047e-06,
1367
+ "loss": 0.9578,
1368
+ "step": 1920
1369
+ },
1370
+ {
1371
+ "epoch": 1.6886895646466855,
1372
+ "grad_norm": 1.2827479839324951,
1373
+ "learning_rate": 4.7682593940662156e-06,
1374
+ "loss": 0.982,
1375
+ "step": 1930
1376
+ },
1377
+ {
1378
+ "epoch": 1.697440385036097,
1379
+ "grad_norm": 1.2533340454101562,
1380
+ "learning_rate": 4.717376806307696e-06,
1381
+ "loss": 0.9735,
1382
+ "step": 1940
1383
+ },
1384
+ {
1385
+ "epoch": 1.7061912054255086,
1386
+ "grad_norm": 1.267432451248169,
1387
+ "learning_rate": 4.666523565077889e-06,
1388
+ "loss": 0.9598,
1389
+ "step": 1950
1390
+ },
1391
+ {
1392
+ "epoch": 1.7149420258149202,
1393
+ "grad_norm": 1.2450517416000366,
1394
+ "learning_rate": 4.615704950785965e-06,
1395
+ "loss": 0.9751,
1396
+ "step": 1960
1397
+ },
1398
+ {
1399
+ "epoch": 1.7236928462043317,
1400
+ "grad_norm": 1.2049565315246582,
1401
+ "learning_rate": 4.5649262402455705e-06,
1402
+ "loss": 0.9737,
1403
+ "step": 1970
1404
+ },
1405
+ {
1406
+ "epoch": 1.7324436665937433,
1407
+ "grad_norm": 1.2524805068969727,
1408
+ "learning_rate": 4.514192706126885e-06,
1409
+ "loss": 0.9662,
1410
+ "step": 1980
1411
+ },
1412
+ {
1413
+ "epoch": 1.7411944869831548,
1414
+ "grad_norm": 1.2566184997558594,
1415
+ "learning_rate": 4.463509616409144e-06,
1416
+ "loss": 0.9618,
1417
+ "step": 1990
1418
+ },
1419
+ {
1420
+ "epoch": 1.7499453073725662,
1421
+ "grad_norm": 1.176069736480713,
1422
+ "learning_rate": 4.4128822338336295e-06,
1423
+ "loss": 0.9521,
1424
+ "step": 2000
1425
+ },
1426
+ {
1427
+ "epoch": 1.7586961277619777,
1428
+ "grad_norm": 1.1588139533996582,
1429
+ "learning_rate": 4.362315815357197e-06,
1430
+ "loss": 0.9488,
1431
+ "step": 2010
1432
+ },
1433
+ {
1434
+ "epoch": 1.767446948151389,
1435
+ "grad_norm": 1.1984119415283203,
1436
+ "learning_rate": 4.311815611606423e-06,
1437
+ "loss": 0.9702,
1438
+ "step": 2020
1439
+ },
1440
+ {
1441
+ "epoch": 1.7761977685408006,
1442
+ "grad_norm": 1.1387180089950562,
1443
+ "learning_rate": 4.261386866332397e-06,
1444
+ "loss": 0.9528,
1445
+ "step": 2030
1446
+ },
1447
+ {
1448
+ "epoch": 1.7849485889302121,
1449
+ "grad_norm": 1.1617263555526733,
1450
+ "learning_rate": 4.211034815866228e-06,
1451
+ "loss": 0.951,
1452
+ "step": 2040
1453
+ },
1454
+ {
1455
+ "epoch": 1.7936994093196237,
1456
+ "grad_norm": 1.1776400804519653,
1457
+ "learning_rate": 4.160764688575321e-06,
1458
+ "loss": 0.9634,
1459
+ "step": 2050
1460
+ },
1461
+ {
1462
+ "epoch": 1.8024502297090352,
1463
+ "grad_norm": 1.25057852268219,
1464
+ "learning_rate": 4.110581704320495e-06,
1465
+ "loss": 0.9633,
1466
+ "step": 2060
1467
+ },
1468
+ {
1469
+ "epoch": 1.8112010500984468,
1470
+ "grad_norm": 1.2121225595474243,
1471
+ "learning_rate": 4.0604910739139575e-06,
1472
+ "loss": 0.9749,
1473
+ "step": 2070
1474
+ },
1475
+ {
1476
+ "epoch": 1.8199518704878583,
1477
+ "grad_norm": 1.3024873733520508,
1478
+ "learning_rate": 4.0104979985782394e-06,
1479
+ "loss": 0.9744,
1480
+ "step": 2080
1481
+ },
1482
+ {
1483
+ "epoch": 1.8287026908772699,
1484
+ "grad_norm": 1.1998348236083984,
1485
+ "learning_rate": 3.960607669406129e-06,
1486
+ "loss": 0.9622,
1487
+ "step": 2090
1488
+ },
1489
+ {
1490
+ "epoch": 1.8374535112666812,
1491
+ "grad_norm": 1.2691677808761597,
1492
+ "learning_rate": 3.9108252668216294e-06,
1493
+ "loss": 0.9702,
1494
+ "step": 2100
1495
+ },
1496
+ {
1497
+ "epoch": 1.8462043316560928,
1498
+ "grad_norm": 1.2797294855117798,
1499
+ "learning_rate": 3.861155960042057e-06,
1500
+ "loss": 0.9294,
1501
+ "step": 2110
1502
+ },
1503
+ {
1504
+ "epoch": 1.8549551520455043,
1505
+ "grad_norm": 1.2609057426452637,
1506
+ "learning_rate": 3.811604906541287e-06,
1507
+ "loss": 0.9624,
1508
+ "step": 2120
1509
+ },
1510
+ {
1511
+ "epoch": 1.8637059724349156,
1512
+ "grad_norm": 1.2273874282836914,
1513
+ "learning_rate": 3.762177251514218e-06,
1514
+ "loss": 0.9563,
1515
+ "step": 2130
1516
+ },
1517
+ {
1518
+ "epoch": 1.8724567928243272,
1519
+ "grad_norm": 1.225128173828125,
1520
+ "learning_rate": 3.7128781273425156e-06,
1521
+ "loss": 0.9587,
1522
+ "step": 2140
1523
+ },
1524
+ {
1525
+ "epoch": 1.8812076132137387,
1526
+ "grad_norm": 1.2284705638885498,
1527
+ "learning_rate": 3.6637126530616816e-06,
1528
+ "loss": 0.9401,
1529
+ "step": 2150
1530
+ },
1531
+ {
1532
+ "epoch": 1.8899584336031503,
1533
+ "grad_norm": 1.2802456617355347,
1534
+ "learning_rate": 3.6146859338295248e-06,
1535
+ "loss": 0.9386,
1536
+ "step": 2160
1537
+ },
1538
+ {
1539
+ "epoch": 1.8987092539925619,
1540
+ "grad_norm": 1.22531259059906,
1541
+ "learning_rate": 3.5658030603960437e-06,
1542
+ "loss": 0.9652,
1543
+ "step": 2170
1544
+ },
1545
+ {
1546
+ "epoch": 1.9074600743819734,
1547
+ "grad_norm": 1.2205225229263306,
1548
+ "learning_rate": 3.5170691085748325e-06,
1549
+ "loss": 0.9418,
1550
+ "step": 2180
1551
+ },
1552
+ {
1553
+ "epoch": 1.916210894771385,
1554
+ "grad_norm": 1.1588237285614014,
1555
+ "learning_rate": 3.4684891387160292e-06,
1556
+ "loss": 0.9624,
1557
+ "step": 2190
1558
+ },
1559
+ {
1560
+ "epoch": 1.9249617151607963,
1561
+ "grad_norm": 1.2422109842300415,
1562
+ "learning_rate": 3.42006819518086e-06,
1563
+ "loss": 0.9714,
1564
+ "step": 2200
1565
+ },
1566
+ {
1567
+ "epoch": 1.9337125355502078,
1568
+ "grad_norm": 1.2960257530212402,
1569
+ "learning_rate": 3.371811305817858e-06,
1570
+ "loss": 0.9521,
1571
+ "step": 2210
1572
+ },
1573
+ {
1574
+ "epoch": 1.9424633559396194,
1575
+ "grad_norm": 1.1022367477416992,
1576
+ "learning_rate": 3.3237234814407904e-06,
1577
+ "loss": 0.9775,
1578
+ "step": 2220
1579
+ },
1580
+ {
1581
+ "epoch": 1.9512141763290307,
1582
+ "grad_norm": 1.160735011100769,
1583
+ "learning_rate": 3.275809715308349e-06,
1584
+ "loss": 0.9497,
1585
+ "step": 2230
1586
+ },
1587
+ {
1588
+ "epoch": 1.9599649967184423,
1589
+ "grad_norm": 1.2306522130966187,
1590
+ "learning_rate": 3.2280749826056736e-06,
1591
+ "loss": 0.9544,
1592
+ "step": 2240
1593
+ },
1594
+ {
1595
+ "epoch": 1.9687158171078538,
1596
+ "grad_norm": 1.1739164590835571,
1597
+ "learning_rate": 3.180524239927749e-06,
1598
+ "loss": 0.9526,
1599
+ "step": 2250
1600
+ },
1601
+ {
1602
+ "epoch": 1.9774666374972654,
1603
+ "grad_norm": 1.2096645832061768,
1604
+ "learning_rate": 3.1331624247647276e-06,
1605
+ "loss": 0.9503,
1606
+ "step": 2260
1607
+ },
1608
+ {
1609
+ "epoch": 1.986217457886677,
1610
+ "grad_norm": 1.2347972393035889,
1611
+ "learning_rate": 3.085994454989232e-06,
1612
+ "loss": 0.9473,
1613
+ "step": 2270
1614
+ },
1615
+ {
1616
+ "epoch": 1.9949682782760885,
1617
+ "grad_norm": 1.28371000289917,
1618
+ "learning_rate": 3.039025228345725e-06,
1619
+ "loss": 0.9787,
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:bc217d8087950d9fd3494f1be8c7d71105ef0db7962bc2fbe0ffa2af67311f0a
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 sha256:7c3b54aecafadb646fe43709d0b21be2b2c95bebeedff7db139e9110a4e59286
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 sha256:2c9487d7f0d45d7aa0fa387e90ef4d800019a6c9b6a85c8afccac576c3216848
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:a8f7bdd934df86e4ed787c672d8e800952066f83c1bb81956ba58ef8edb9e9a3
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 sha256:544c5ca4086cf3fee147d0bd7e7aa450ddec0a2e6edc47805b9414fa803adbc9
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 sha256:9fb748955dc46c10f37c98e83382abe618e4e38fe2a97531a64de73bde4f2b93
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 sha256:6193561445270385149bb372e15ca4a608b1a8b9f7c8fcb69585a622a7193b8d
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:f7f27a6999461b877a0f4fab2e274cd5b739b0602342058edede1198a9292405
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:bdca2f25501a0c69f21b417a89e61656d810066ae4fdedcb4dadd8abc6be4aab
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 sha256:55f47b3eea82387e1558781ef41b16160137853deab53c717ef6cd47dfbfddb9
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 sha256:39fdabeef7466290e9470312293aadb5b1aafae0225a32b7df79e89fa94d0826
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": 0.8244,
15
+ "eval_samples_per_second": 24.261,
16
+ "eval_steps_per_second": 2.426,
17
+ "step": 0
18
+ },
19
+ {
20
+ "epoch": 0.0008750820389411508,
21
+ "grad_norm": 6.372805118560791,
22
+ "learning_rate": 2.915451895043732e-08,
23
+ "loss": 1.6434,
24
+ "step": 1
25
+ },
26
+ {
27
+ "epoch": 0.008750820389411508,
28
+ "grad_norm": 6.278440475463867,
29
+ "learning_rate": 2.915451895043732e-07,
30
+ "loss": 1.7188,
31
+ "step": 10
32
+ },
33
+ {
34
+ "epoch": 0.017501640778823015,
35
+ "grad_norm": 6.023682594299316,
36
+ "learning_rate": 5.830903790087464e-07,
37
+ "loss": 1.7154,
38
+ "step": 20
39
+ },
40
+ {
41
+ "epoch": 0.02625246116823452,
42
+ "grad_norm": 3.7778327465057373,
43
+ "learning_rate": 8.746355685131196e-07,
44
+ "loss": 1.6563,
45
+ "step": 30
46
+ },
47
+ {
48
+ "epoch": 0.03500328155764603,
49
+ "grad_norm": 3.8672306537628174,
50
+ "learning_rate": 1.1661807580174927e-06,
51
+ "loss": 1.5743,
52
+ "step": 40
53
+ },
54
+ {
55
+ "epoch": 0.04375410194705753,
56
+ "grad_norm": 2.53611159324646,
57
+ "learning_rate": 1.4577259475218661e-06,
58
+ "loss": 1.479,
59
+ "step": 50
60
+ },
61
+ {
62
+ "epoch": 0.05250492233646904,
63
+ "grad_norm": 2.019819974899292,
64
+ "learning_rate": 1.7492711370262391e-06,
65
+ "loss": 1.4242,
66
+ "step": 60
67
+ },
68
+ {
69
+ "epoch": 0.06125574272588055,
70
+ "grad_norm": 1.7553459405899048,
71
+ "learning_rate": 2.0408163265306125e-06,
72
+ "loss": 1.3795,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.07000656311529206,
77
+ "grad_norm": 1.4397658109664917,
78
+ "learning_rate": 2.3323615160349855e-06,
79
+ "loss": 1.3745,
80
+ "step": 80
81
+ },
82
+ {
83
+ "epoch": 0.07875738350470357,
84
+ "grad_norm": 1.5793251991271973,
85
+ "learning_rate": 2.6239067055393585e-06,
86
+ "loss": 1.3529,
87
+ "step": 90
88
+ },
89
+ {
90
+ "epoch": 0.08750820389411507,
91
+ "grad_norm": 1.5069832801818848,
92
+ "learning_rate": 2.9154518950437323e-06,
93
+ "loss": 1.3459,
94
+ "step": 100
95
+ },
96
+ {
97
+ "epoch": 0.09625902428352658,
98
+ "grad_norm": 1.4921239614486694,
99
+ "learning_rate": 3.2069970845481052e-06,
100
+ "loss": 1.2743,
101
+ "step": 110
102
+ },
103
+ {
104
+ "epoch": 0.10500984467293809,
105
+ "grad_norm": 1.4761064052581787,
106
+ "learning_rate": 3.4985422740524782e-06,
107
+ "loss": 1.2959,
108
+ "step": 120
109
+ },
110
+ {
111
+ "epoch": 0.1137606650623496,
112
+ "grad_norm": 1.4724879264831543,
113
+ "learning_rate": 3.790087463556852e-06,
114
+ "loss": 1.2773,
115
+ "step": 130
116
+ },
117
+ {
118
+ "epoch": 0.1225114854517611,
119
+ "grad_norm": 1.4954681396484375,
120
+ "learning_rate": 4.081632653061225e-06,
121
+ "loss": 1.2695,
122
+ "step": 140
123
+ },
124
+ {
125
+ "epoch": 0.1312623058411726,
126
+ "grad_norm": 1.4220881462097168,
127
+ "learning_rate": 4.3731778425655976e-06,
128
+ "loss": 1.2668,
129
+ "step": 150
130
+ },
131
+ {
132
+ "epoch": 0.14001312623058412,
133
+ "grad_norm": 1.4039027690887451,
134
+ "learning_rate": 4.664723032069971e-06,
135
+ "loss": 1.2231,
136
+ "step": 160
137
+ },
138
+ {
139
+ "epoch": 0.14876394661999562,
140
+ "grad_norm": 1.468996524810791,
141
+ "learning_rate": 4.956268221574344e-06,
142
+ "loss": 1.221,
143
+ "step": 170
144
+ },
145
+ {
146
+ "epoch": 0.15751476700940714,
147
+ "grad_norm": 1.5780359506607056,
148
+ "learning_rate": 5.247813411078717e-06,
149
+ "loss": 1.2403,
150
+ "step": 180
151
+ },
152
+ {
153
+ "epoch": 0.16626558739881864,
154
+ "grad_norm": 1.4090375900268555,
155
+ "learning_rate": 5.539358600583091e-06,
156
+ "loss": 1.2021,
157
+ "step": 190
158
+ },
159
+ {
160
+ "epoch": 0.17501640778823013,
161
+ "grad_norm": 1.489652395248413,
162
+ "learning_rate": 5.8309037900874645e-06,
163
+ "loss": 1.1731,
164
+ "step": 200
165
+ },
166
+ {
167
+ "epoch": 0.18376722817764166,
168
+ "grad_norm": 1.3121322393417358,
169
+ "learning_rate": 6.122448979591837e-06,
170
+ "loss": 1.1924,
171
+ "step": 210
172
+ },
173
+ {
174
+ "epoch": 0.19251804856705315,
175
+ "grad_norm": 1.3954280614852905,
176
+ "learning_rate": 6.4139941690962105e-06,
177
+ "loss": 1.2121,
178
+ "step": 220
179
+ },
180
+ {
181
+ "epoch": 0.20126886895646467,
182
+ "grad_norm": 1.4191243648529053,
183
+ "learning_rate": 6.705539358600584e-06,
184
+ "loss": 1.1973,
185
+ "step": 230
186
+ },
187
+ {
188
+ "epoch": 0.21001968934587617,
189
+ "grad_norm": 1.4373400211334229,
190
+ "learning_rate": 6.9970845481049564e-06,
191
+ "loss": 1.1701,
192
+ "step": 240
193
+ },
194
+ {
195
+ "epoch": 0.2187705097352877,
196
+ "grad_norm": 1.4669713973999023,
197
+ "learning_rate": 7.28862973760933e-06,
198
+ "loss": 1.1923,
199
+ "step": 250
200
+ },
201
+ {
202
+ "epoch": 0.2275213301246992,
203
+ "grad_norm": 1.4269640445709229,
204
+ "learning_rate": 7.580174927113704e-06,
205
+ "loss": 1.1713,
206
+ "step": 260
207
+ },
208
+ {
209
+ "epoch": 0.23627215051411068,
210
+ "grad_norm": 1.7535545825958252,
211
+ "learning_rate": 7.871720116618077e-06,
212
+ "loss": 1.1891,
213
+ "step": 270
214
+ },
215
+ {
216
+ "epoch": 0.2450229709035222,
217
+ "grad_norm": 1.599094271659851,
218
+ "learning_rate": 8.16326530612245e-06,
219
+ "loss": 1.1827,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2537737912929337,
224
+ "grad_norm": 1.8278971910476685,
225
+ "learning_rate": 8.454810495626823e-06,
226
+ "loss": 1.1402,
227
+ "step": 290
228
+ },
229
+ {
230
+ "epoch": 0.2625246116823452,
231
+ "grad_norm": 1.5127216577529907,
232
+ "learning_rate": 8.746355685131195e-06,
233
+ "loss": 1.1597,
234
+ "step": 300
235
+ },
236
+ {
237
+ "epoch": 0.27127543207175675,
238
+ "grad_norm": 1.6015137434005737,
239
+ "learning_rate": 9.03790087463557e-06,
240
+ "loss": 1.1398,
241
+ "step": 310
242
+ },
243
+ {
244
+ "epoch": 0.28002625246116825,
245
+ "grad_norm": 1.3944435119628906,
246
+ "learning_rate": 9.329446064139942e-06,
247
+ "loss": 1.1174,
248
+ "step": 320
249
+ },
250
+ {
251
+ "epoch": 0.28877707285057974,
252
+ "grad_norm": 1.2441927194595337,
253
+ "learning_rate": 9.620991253644316e-06,
254
+ "loss": 1.1416,
255
+ "step": 330
256
+ },
257
+ {
258
+ "epoch": 0.29752789323999124,
259
+ "grad_norm": 1.4111624956130981,
260
+ "learning_rate": 9.912536443148689e-06,
261
+ "loss": 1.1352,
262
+ "step": 340
263
+ },
264
+ {
265
+ "epoch": 0.30627871362940273,
266
+ "grad_norm": 1.3007830381393433,
267
+ "learning_rate": 9.999872800049233e-06,
268
+ "loss": 1.1208,
269
+ "step": 350
270
+ },
271
+ {
272
+ "epoch": 0.3150295340188143,
273
+ "grad_norm": 1.3299254179000854,
274
+ "learning_rate": 9.999249795462235e-06,
275
+ "loss": 1.142,
276
+ "step": 360
277
+ },
278
+ {
279
+ "epoch": 0.3237803544082258,
280
+ "grad_norm": 1.2524114847183228,
281
+ "learning_rate": 9.998107687592326e-06,
282
+ "loss": 1.1147,
283
+ "step": 370
284
+ },
285
+ {
286
+ "epoch": 0.3325311747976373,
287
+ "grad_norm": 1.4711254835128784,
288
+ "learning_rate": 9.996446595031691e-06,
289
+ "loss": 1.1182,
290
+ "step": 380
291
+ },
292
+ {
293
+ "epoch": 0.34128199518704877,
294
+ "grad_norm": 1.3600913286209106,
295
+ "learning_rate": 9.994266690261928e-06,
296
+ "loss": 1.1565,
297
+ "step": 390
298
+ },
299
+ {
300
+ "epoch": 0.35003281557646027,
301
+ "grad_norm": 1.3752756118774414,
302
+ "learning_rate": 9.991568199636149e-06,
303
+ "loss": 1.0803,
304
+ "step": 400
305
+ },
306
+ {
307
+ "epoch": 0.3587836359658718,
308
+ "grad_norm": 1.2283236980438232,
309
+ "learning_rate": 9.988351403355455e-06,
310
+ "loss": 1.1371,
311
+ "step": 410
312
+ },
313
+ {
314
+ "epoch": 0.3675344563552833,
315
+ "grad_norm": 1.266762614250183,
316
+ "learning_rate": 9.98461663543987e-06,
317
+ "loss": 1.1027,
318
+ "step": 420
319
+ },
320
+ {
321
+ "epoch": 0.3762852767446948,
322
+ "grad_norm": 1.2771412134170532,
323
+ "learning_rate": 9.980364283693638e-06,
324
+ "loss": 1.0891,
325
+ "step": 430
326
+ },
327
+ {
328
+ "epoch": 0.3850360971341063,
329
+ "grad_norm": 1.2226617336273193,
330
+ "learning_rate": 9.97559478966496e-06,
331
+ "loss": 1.1105,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.39378691752351785,
336
+ "grad_norm": 1.2738267183303833,
337
+ "learning_rate": 9.970308648600142e-06,
338
+ "loss": 1.114,
339
+ "step": 450
340
+ },
341
+ {
342
+ "epoch": 0.40253773791292935,
343
+ "grad_norm": 1.3344817161560059,
344
+ "learning_rate": 9.964506409392178e-06,
345
+ "loss": 1.1178,
346
+ "step": 460
347
+ },
348
+ {
349
+ "epoch": 0.41128855830234085,
350
+ "grad_norm": 1.3076369762420654,
351
+ "learning_rate": 9.958188674523748e-06,
352
+ "loss": 1.1205,
353
+ "step": 470
354
+ },
355
+ {
356
+ "epoch": 0.42003937869175234,
357
+ "grad_norm": 1.251315712928772,
358
+ "learning_rate": 9.951356100004666e-06,
359
+ "loss": 1.0795,
360
+ "step": 480
361
+ },
362
+ {
363
+ "epoch": 0.42879019908116384,
364
+ "grad_norm": 1.1784945726394653,
365
+ "learning_rate": 9.94400939530375e-06,
366
+ "loss": 1.1175,
367
+ "step": 490
368
+ },
369
+ {
370
+ "epoch": 0.4375410194705754,
371
+ "grad_norm": 1.219256043434143,
372
+ "learning_rate": 9.93614932327517e-06,
373
+ "loss": 1.0933,
374
+ "step": 500
375
+ },
376
+ {
377
+ "epoch": 0.4462918398599869,
378
+ "grad_norm": 1.398763656616211,
379
+ "learning_rate": 9.927776700079217e-06,
380
+ "loss": 1.1267,
381
+ "step": 510
382
+ },
383
+ {
384
+ "epoch": 0.4550426602493984,
385
+ "grad_norm": 1.25184166431427,
386
+ "learning_rate": 9.918892395097572e-06,
387
+ "loss": 1.1191,
388
+ "step": 520
389
+ },
390
+ {
391
+ "epoch": 0.4637934806388099,
392
+ "grad_norm": 1.3422456979751587,
393
+ "learning_rate": 9.90949733084303e-06,
394
+ "loss": 1.1003,
395
+ "step": 530
396
+ },
397
+ {
398
+ "epoch": 0.47254430102822137,
399
+ "grad_norm": 1.3312543630599976,
400
+ "learning_rate": 9.8995924828637e-06,
401
+ "loss": 1.1004,
402
+ "step": 540
403
+ },
404
+ {
405
+ "epoch": 0.4812951214176329,
406
+ "grad_norm": 1.2708804607391357,
407
+ "learning_rate": 9.889178879641717e-06,
408
+ "loss": 1.1094,
409
+ "step": 550
410
+ },
411
+ {
412
+ "epoch": 0.4900459418070444,
413
+ "grad_norm": 1.3565181493759155,
414
+ "learning_rate": 9.878257602486445e-06,
415
+ "loss": 1.1252,
416
+ "step": 560
417
+ },
418
+ {
419
+ "epoch": 0.4987967621964559,
420
+ "grad_norm": 1.232021450996399,
421
+ "learning_rate": 9.866829785422201e-06,
422
+ "loss": 1.0976,
423
+ "step": 570
424
+ },
425
+ {
426
+ "epoch": 0.5075475825858674,
427
+ "grad_norm": 1.371229887008667,
428
+ "learning_rate": 9.854896615070495e-06,
429
+ "loss": 1.0941,
430
+ "step": 580
431
+ },
432
+ {
433
+ "epoch": 0.516298402975279,
434
+ "grad_norm": 1.35117769241333,
435
+ "learning_rate": 9.842459330526826e-06,
436
+ "loss": 1.1127,
437
+ "step": 590
438
+ },
439
+ {
440
+ "epoch": 0.5250492233646904,
441
+ "grad_norm": 1.234413504600525,
442
+ "learning_rate": 9.829519223232006e-06,
443
+ "loss": 1.09,
444
+ "step": 600
445
+ },
446
+ {
447
+ "epoch": 0.533800043754102,
448
+ "grad_norm": 1.2576864957809448,
449
+ "learning_rate": 9.816077636838079e-06,
450
+ "loss": 1.1094,
451
+ "step": 610
452
+ },
453
+ {
454
+ "epoch": 0.5425508641435135,
455
+ "grad_norm": 1.340950608253479,
456
+ "learning_rate": 9.802135967068783e-06,
457
+ "loss": 1.0889,
458
+ "step": 620
459
+ },
460
+ {
461
+ "epoch": 0.5513016845329249,
462
+ "grad_norm": 1.350095272064209,
463
+ "learning_rate": 9.787695661574632e-06,
464
+ "loss": 1.0661,
465
+ "step": 630
466
+ },
467
+ {
468
+ "epoch": 0.5600525049223365,
469
+ "grad_norm": 1.2629624605178833,
470
+ "learning_rate": 9.772758219782601e-06,
471
+ "loss": 1.0823,
472
+ "step": 640
473
+ },
474
+ {
475
+ "epoch": 0.5688033253117479,
476
+ "grad_norm": 1.234574556350708,
477
+ "learning_rate": 9.757325192740425e-06,
478
+ "loss": 1.0937,
479
+ "step": 650
480
+ },
481
+ {
482
+ "epoch": 0.5775541457011595,
483
+ "grad_norm": 1.1775031089782715,
484
+ "learning_rate": 9.741398182955548e-06,
485
+ "loss": 1.1066,
486
+ "step": 660
487
+ },
488
+ {
489
+ "epoch": 0.586304966090571,
490
+ "grad_norm": 1.2653073072433472,
491
+ "learning_rate": 9.724978844228717e-06,
492
+ "loss": 1.0964,
493
+ "step": 670
494
+ },
495
+ {
496
+ "epoch": 0.5950557864799825,
497
+ "grad_norm": 1.2665278911590576,
498
+ "learning_rate": 9.70806888148227e-06,
499
+ "loss": 1.1007,
500
+ "step": 680
501
+ },
502
+ {
503
+ "epoch": 0.603806606869394,
504
+ "grad_norm": 1.2909613847732544,
505
+ "learning_rate": 9.690670050583095e-06,
506
+ "loss": 1.0973,
507
+ "step": 690
508
+ },
509
+ {
510
+ "epoch": 0.6125574272588055,
511
+ "grad_norm": 1.2396572828292847,
512
+ "learning_rate": 9.672784158160307e-06,
513
+ "loss": 1.0423,
514
+ "step": 700
515
+ },
516
+ {
517
+ "epoch": 0.621308247648217,
518
+ "grad_norm": 1.2569255828857422,
519
+ "learning_rate": 9.654413061417655e-06,
520
+ "loss": 1.0764,
521
+ "step": 710
522
+ },
523
+ {
524
+ "epoch": 0.6300590680376286,
525
+ "grad_norm": 1.3747756481170654,
526
+ "learning_rate": 9.635558667940683e-06,
527
+ "loss": 1.0563,
528
+ "step": 720
529
+ },
530
+ {
531
+ "epoch": 0.63880988842704,
532
+ "grad_norm": 1.2752923965454102,
533
+ "learning_rate": 9.616222935498647e-06,
534
+ "loss": 1.1084,
535
+ "step": 730
536
+ },
537
+ {
538
+ "epoch": 0.6475607088164516,
539
+ "grad_norm": 1.218320369720459,
540
+ "learning_rate": 9.596407871841229e-06,
541
+ "loss": 1.0674,
542
+ "step": 740
543
+ },
544
+ {
545
+ "epoch": 0.656311529205863,
546
+ "grad_norm": 1.2466013431549072,
547
+ "learning_rate": 9.57611553449006e-06,
548
+ "loss": 1.1048,
549
+ "step": 750
550
+ },
551
+ {
552
+ "epoch": 0.6650623495952745,
553
+ "grad_norm": 1.2956342697143555,
554
+ "learning_rate": 9.555348030525076e-06,
555
+ "loss": 1.0723,
556
+ "step": 760
557
+ },
558
+ {
559
+ "epoch": 0.6738131699846861,
560
+ "grad_norm": 1.29911208152771,
561
+ "learning_rate": 9.534107516365725e-06,
562
+ "loss": 1.0743,
563
+ "step": 770
564
+ },
565
+ {
566
+ "epoch": 0.6825639903740975,
567
+ "grad_norm": 1.2063359022140503,
568
+ "learning_rate": 9.512396197547053e-06,
569
+ "loss": 1.0717,
570
+ "step": 780
571
+ },
572
+ {
573
+ "epoch": 0.6913148107635091,
574
+ "grad_norm": 1.184462547302246,
575
+ "learning_rate": 9.490216328490693e-06,
576
+ "loss": 1.0895,
577
+ "step": 790
578
+ },
579
+ {
580
+ "epoch": 0.7000656311529205,
581
+ "grad_norm": 1.186874270439148,
582
+ "learning_rate": 9.467570212270767e-06,
583
+ "loss": 1.1476,
584
+ "step": 800
585
+ },
586
+ {
587
+ "epoch": 0.7088164515423321,
588
+ "grad_norm": 1.2901315689086914,
589
+ "learning_rate": 9.444460200374747e-06,
590
+ "loss": 1.0538,
591
+ "step": 810
592
+ },
593
+ {
594
+ "epoch": 0.7175672719317436,
595
+ "grad_norm": 1.2175313234329224,
596
+ "learning_rate": 9.42088869245929e-06,
597
+ "loss": 1.079,
598
+ "step": 820
599
+ },
600
+ {
601
+ "epoch": 0.7263180923211551,
602
+ "grad_norm": 1.1771786212921143,
603
+ "learning_rate": 9.396858136101066e-06,
604
+ "loss": 1.0756,
605
+ "step": 830
606
+ },
607
+ {
608
+ "epoch": 0.7350689127105666,
609
+ "grad_norm": 1.2814205884933472,
610
+ "learning_rate": 9.372371026542595e-06,
611
+ "loss": 1.0697,
612
+ "step": 840
613
+ },
614
+ {
615
+ "epoch": 0.7438197330999782,
616
+ "grad_norm": 1.2603764533996582,
617
+ "learning_rate": 9.347429906433175e-06,
618
+ "loss": 1.1079,
619
+ "step": 850
620
+ },
621
+ {
622
+ "epoch": 0.7525705534893896,
623
+ "grad_norm": 1.2228299379348755,
624
+ "learning_rate": 9.322037365564844e-06,
625
+ "loss": 1.0746,
626
+ "step": 860
627
+ },
628
+ {
629
+ "epoch": 0.7613213738788012,
630
+ "grad_norm": 1.1895763874053955,
631
+ "learning_rate": 9.296196040603476e-06,
632
+ "loss": 1.0831,
633
+ "step": 870
634
+ },
635
+ {
636
+ "epoch": 0.7700721942682126,
637
+ "grad_norm": 1.2288453578948975,
638
+ "learning_rate": 9.269908614814995e-06,
639
+ "loss": 1.0649,
640
+ "step": 880
641
+ },
642
+ {
643
+ "epoch": 0.7788230146576242,
644
+ "grad_norm": 1.1580345630645752,
645
+ "learning_rate": 9.243177817786757e-06,
646
+ "loss": 1.1053,
647
+ "step": 890
648
+ },
649
+ {
650
+ "epoch": 0.7875738350470357,
651
+ "grad_norm": 1.1613861322402954,
652
+ "learning_rate": 9.216006425144118e-06,
653
+ "loss": 1.0854,
654
+ "step": 900
655
+ },
656
+ {
657
+ "epoch": 0.7963246554364471,
658
+ "grad_norm": 1.0809310674667358,
659
+ "learning_rate": 9.188397258262221e-06,
660
+ "loss": 1.059,
661
+ "step": 910
662
+ },
663
+ {
664
+ "epoch": 0.8050754758258587,
665
+ "grad_norm": 1.2021125555038452,
666
+ "learning_rate": 9.160353183973045e-06,
667
+ "loss": 1.064,
668
+ "step": 920
669
+ },
670
+ {
671
+ "epoch": 0.8138262962152701,
672
+ "grad_norm": 1.286584496498108,
673
+ "learning_rate": 9.131877114267711e-06,
674
+ "loss": 1.0516,
675
+ "step": 930
676
+ },
677
+ {
678
+ "epoch": 0.8225771166046817,
679
+ "grad_norm": 1.177673578262329,
680
+ "learning_rate": 9.102972005994123e-06,
681
+ "loss": 1.0358,
682
+ "step": 940
683
+ },
684
+ {
685
+ "epoch": 0.8313279369940932,
686
+ "grad_norm": 1.2117091417312622,
687
+ "learning_rate": 9.073640860549931e-06,
688
+ "loss": 1.0701,
689
+ "step": 950
690
+ },
691
+ {
692
+ "epoch": 0.8400787573835047,
693
+ "grad_norm": 1.2018595933914185,
694
+ "learning_rate": 9.043886723570885e-06,
695
+ "loss": 1.0833,
696
+ "step": 960
697
+ },
698
+ {
699
+ "epoch": 0.8488295777729162,
700
+ "grad_norm": 1.1337984800338745,
701
+ "learning_rate": 9.013712684614582e-06,
702
+ "loss": 1.0785,
703
+ "step": 970
704
+ },
705
+ {
706
+ "epoch": 0.8575803981623277,
707
+ "grad_norm": 1.2217832803726196,
708
+ "learning_rate": 8.983121876839658e-06,
709
+ "loss": 1.0623,
710
+ "step": 980
711
+ },
712
+ {
713
+ "epoch": 0.8663312185517392,
714
+ "grad_norm": 1.2249287366867065,
715
+ "learning_rate": 8.95211747668046e-06,
716
+ "loss": 1.0285,
717
+ "step": 990
718
+ },
719
+ {
720
+ "epoch": 0.8750820389411508,
721
+ "grad_norm": 1.1673548221588135,
722
+ "learning_rate": 8.920702703517205e-06,
723
+ "loss": 1.0354,
724
+ "step": 1000
725
+ },
726
+ {
727
+ "epoch": 0.8838328593305622,
728
+ "grad_norm": 1.1298819780349731,
729
+ "learning_rate": 8.888880819341703e-06,
730
+ "loss": 1.0538,
731
+ "step": 1010
732
+ },
733
+ {
734
+ "epoch": 0.8925836797199738,
735
+ "grad_norm": 1.1600337028503418,
736
+ "learning_rate": 8.856655128418638e-06,
737
+ "loss": 1.0527,
738
+ "step": 1020
739
+ },
740
+ {
741
+ "epoch": 0.9013345001093852,
742
+ "grad_norm": 1.2292169332504272,
743
+ "learning_rate": 8.824028976942465e-06,
744
+ "loss": 1.0613,
745
+ "step": 1030
746
+ },
747
+ {
748
+ "epoch": 0.9100853204987968,
749
+ "grad_norm": 1.1878681182861328,
750
+ "learning_rate": 8.791005752689958e-06,
751
+ "loss": 1.0489,
752
+ "step": 1040
753
+ },
754
+ {
755
+ "epoch": 0.9188361408882083,
756
+ "grad_norm": 1.2165477275848389,
757
+ "learning_rate": 8.757588884668435e-06,
758
+ "loss": 1.0675,
759
+ "step": 1050
760
+ },
761
+ {
762
+ "epoch": 0.9275869612776197,
763
+ "grad_norm": 1.1682305335998535,
764
+ "learning_rate": 8.723781842759696e-06,
765
+ "loss": 1.084,
766
+ "step": 1060
767
+ },
768
+ {
769
+ "epoch": 0.9363377816670313,
770
+ "grad_norm": 1.3387882709503174,
771
+ "learning_rate": 8.689588137359731e-06,
772
+ "loss": 1.0564,
773
+ "step": 1070
774
+ },
775
+ {
776
+ "epoch": 0.9450886020564427,
777
+ "grad_norm": 1.2144585847854614,
778
+ "learning_rate": 8.655011319014217e-06,
779
+ "loss": 1.08,
780
+ "step": 1080
781
+ },
782
+ {
783
+ "epoch": 0.9538394224458543,
784
+ "grad_norm": 1.2637611627578735,
785
+ "learning_rate": 8.620054978049828e-06,
786
+ "loss": 1.0716,
787
+ "step": 1090
788
+ },
789
+ {
790
+ "epoch": 0.9625902428352658,
791
+ "grad_norm": 1.230635404586792,
792
+ "learning_rate": 8.584722744201448e-06,
793
+ "loss": 1.0429,
794
+ "step": 1100
795
+ },
796
+ {
797
+ "epoch": 0.9713410632246773,
798
+ "grad_norm": 1.2495629787445068,
799
+ "learning_rate": 8.549018286235255e-06,
800
+ "loss": 1.0464,
801
+ "step": 1110
802
+ },
803
+ {
804
+ "epoch": 0.9800918836140888,
805
+ "grad_norm": 1.1702367067337036,
806
+ "learning_rate": 8.512945311567786e-06,
807
+ "loss": 1.0476,
808
+ "step": 1120
809
+ },
810
+ {
811
+ "epoch": 0.9888427040035004,
812
+ "grad_norm": 1.1367738246917725,
813
+ "learning_rate": 8.476507565880959e-06,
814
+ "loss": 1.0392,
815
+ "step": 1130
816
+ },
817
+ {
818
+ "epoch": 0.9975935243929118,
819
+ "grad_norm": 1.1903005838394165,
820
+ "learning_rate": 8.439708832733142e-06,
821
+ "loss": 1.0414,
822
+ "step": 1140
823
+ },
824
+ {
825
+ "epoch": 1.006125574272588,
826
+ "grad_norm": 1.310563325881958,
827
+ "learning_rate": 8.402552933166288e-06,
828
+ "loss": 0.9966,
829
+ "step": 1150
830
+ },
831
+ {
832
+ "epoch": 1.0148763946619996,
833
+ "grad_norm": 1.2823761701583862,
834
+ "learning_rate": 8.36504372530916e-06,
835
+ "loss": 0.9961,
836
+ "step": 1160
837
+ },
838
+ {
839
+ "epoch": 1.023627215051411,
840
+ "grad_norm": 1.340758204460144,
841
+ "learning_rate": 8.327185103976731e-06,
842
+ "loss": 0.9904,
843
+ "step": 1170
844
+ },
845
+ {
846
+ "epoch": 1.0323780354408225,
847
+ "grad_norm": 1.2751260995864868,
848
+ "learning_rate": 8.288981000265748e-06,
849
+ "loss": 1.004,
850
+ "step": 1180
851
+ },
852
+ {
853
+ "epoch": 1.041128855830234,
854
+ "grad_norm": 1.1516119241714478,
855
+ "learning_rate": 8.25043538114655e-06,
856
+ "loss": 0.9757,
857
+ "step": 1190
858
+ },
859
+ {
860
+ "epoch": 1.0498796762196456,
861
+ "grad_norm": 1.206026315689087,
862
+ "learning_rate": 8.211552249051149e-06,
863
+ "loss": 0.9799,
864
+ "step": 1200
865
+ },
866
+ {
867
+ "epoch": 1.0586304966090572,
868
+ "grad_norm": 1.2714760303497314,
869
+ "learning_rate": 8.17233564145764e-06,
870
+ "loss": 0.9731,
871
+ "step": 1210
872
+ },
873
+ {
874
+ "epoch": 1.0673813169984685,
875
+ "grad_norm": 1.2047169208526611,
876
+ "learning_rate": 8.132789630470952e-06,
877
+ "loss": 0.971,
878
+ "step": 1220
879
+ },
880
+ {
881
+ "epoch": 1.07613213738788,
882
+ "grad_norm": 1.282832384109497,
883
+ "learning_rate": 8.092918322400028e-06,
884
+ "loss": 0.9695,
885
+ "step": 1230
886
+ },
887
+ {
888
+ "epoch": 1.0848829577772916,
889
+ "grad_norm": 1.2463730573654175,
890
+ "learning_rate": 8.05272585733143e-06,
891
+ "loss": 0.9727,
892
+ "step": 1240
893
+ },
894
+ {
895
+ "epoch": 1.0936337781667032,
896
+ "grad_norm": 1.230220079421997,
897
+ "learning_rate": 8.012216408699456e-06,
898
+ "loss": 0.9909,
899
+ "step": 1250
900
+ },
901
+ {
902
+ "epoch": 1.1023845985561147,
903
+ "grad_norm": 1.3110296726226807,
904
+ "learning_rate": 7.97139418285279e-06,
905
+ "loss": 0.9887,
906
+ "step": 1260
907
+ },
908
+ {
909
+ "epoch": 1.1111354189455263,
910
+ "grad_norm": 1.234381914138794,
911
+ "learning_rate": 7.930263418617714e-06,
912
+ "loss": 1.0014,
913
+ "step": 1270
914
+ },
915
+ {
916
+ "epoch": 1.1198862393349376,
917
+ "grad_norm": 1.2342872619628906,
918
+ "learning_rate": 7.888828386857985e-06,
919
+ "loss": 0.9795,
920
+ "step": 1280
921
+ },
922
+ {
923
+ "epoch": 1.1286370597243491,
924
+ "grad_norm": 1.279505729675293,
925
+ "learning_rate": 7.847093390031351e-06,
926
+ "loss": 0.9647,
927
+ "step": 1290
928
+ },
929
+ {
930
+ "epoch": 1.1373878801137607,
931
+ "grad_norm": 1.3248056173324585,
932
+ "learning_rate": 7.8050627617428e-06,
933
+ "loss": 0.9881,
934
+ "step": 1300
935
+ },
936
+ {
937
+ "epoch": 1.1461387005031722,
938
+ "grad_norm": 1.234870433807373,
939
+ "learning_rate": 7.76274086629458e-06,
940
+ "loss": 0.9728,
941
+ "step": 1310
942
+ },
943
+ {
944
+ "epoch": 1.1548895208925836,
945
+ "grad_norm": 1.2661969661712646,
946
+ "learning_rate": 7.72013209823303e-06,
947
+ "loss": 0.9716,
948
+ "step": 1320
949
+ },
950
+ {
951
+ "epoch": 1.1636403412819951,
952
+ "grad_norm": 1.180408239364624,
953
+ "learning_rate": 7.677240881892259e-06,
954
+ "loss": 0.9641,
955
+ "step": 1330
956
+ },
957
+ {
958
+ "epoch": 1.1723911616714067,
959
+ "grad_norm": 1.2939667701721191,
960
+ "learning_rate": 7.634071670934734e-06,
961
+ "loss": 0.9832,
962
+ "step": 1340
963
+ },
964
+ {
965
+ "epoch": 1.1811419820608182,
966
+ "grad_norm": 1.1289607286453247,
967
+ "learning_rate": 7.5906289478888475e-06,
968
+ "loss": 0.9733,
969
+ "step": 1350
970
+ },
971
+ {
972
+ "epoch": 1.1898928024502298,
973
+ "grad_norm": 1.2281969785690308,
974
+ "learning_rate": 7.546917223683453e-06,
975
+ "loss": 0.9645,
976
+ "step": 1360
977
+ },
978
+ {
979
+ "epoch": 1.1986436228396413,
980
+ "grad_norm": 1.226364254951477,
981
+ "learning_rate": 7.502941037179474e-06,
982
+ "loss": 0.9668,
983
+ "step": 1370
984
+ },
985
+ {
986
+ "epoch": 1.2073944432290526,
987
+ "grad_norm": 1.2580183744430542,
988
+ "learning_rate": 7.4587049546986e-06,
989
+ "loss": 1.0167,
990
+ "step": 1380
991
+ },
992
+ {
993
+ "epoch": 1.2161452636184642,
994
+ "grad_norm": 1.261915683746338,
995
+ "learning_rate": 7.414213569549145e-06,
996
+ "loss": 0.9896,
997
+ "step": 1390
998
+ },
999
+ {
1000
+ "epoch": 1.2248960840078758,
1001
+ "grad_norm": 1.2158194780349731,
1002
+ "learning_rate": 7.369471501549087e-06,
1003
+ "loss": 0.9893,
1004
+ "step": 1400
1005
+ },
1006
+ {
1007
+ "epoch": 1.2336469043972873,
1008
+ "grad_norm": 1.2699130773544312,
1009
+ "learning_rate": 7.324483396546371e-06,
1010
+ "loss": 1.0171,
1011
+ "step": 1410
1012
+ },
1013
+ {
1014
+ "epoch": 1.2423977247866986,
1015
+ "grad_norm": 1.1923282146453857,
1016
+ "learning_rate": 7.279253925936498e-06,
1017
+ "loss": 0.9802,
1018
+ "step": 1420
1019
+ },
1020
+ {
1021
+ "epoch": 1.2511485451761102,
1022
+ "grad_norm": 1.1498876810073853,
1023
+ "learning_rate": 7.233787786177464e-06,
1024
+ "loss": 0.9683,
1025
+ "step": 1430
1026
+ },
1027
+ {
1028
+ "epoch": 1.2598993655655217,
1029
+ "grad_norm": 1.2153609991073608,
1030
+ "learning_rate": 7.188089698302097e-06,
1031
+ "loss": 0.9916,
1032
+ "step": 1440
1033
+ },
1034
+ {
1035
+ "epoch": 1.2686501859549333,
1036
+ "grad_norm": 1.1934350728988647,
1037
+ "learning_rate": 7.142164407427849e-06,
1038
+ "loss": 0.9788,
1039
+ "step": 1450
1040
+ },
1041
+ {
1042
+ "epoch": 1.2774010063443448,
1043
+ "grad_norm": 1.2987148761749268,
1044
+ "learning_rate": 7.096016682264074e-06,
1045
+ "loss": 0.9865,
1046
+ "step": 1460
1047
+ },
1048
+ {
1049
+ "epoch": 1.2861518267337564,
1050
+ "grad_norm": 1.2581541538238525,
1051
+ "learning_rate": 7.049651314616862e-06,
1052
+ "loss": 0.9494,
1053
+ "step": 1470
1054
+ },
1055
+ {
1056
+ "epoch": 1.2949026471231677,
1057
+ "grad_norm": 1.257711410522461,
1058
+ "learning_rate": 7.003073118891487e-06,
1059
+ "loss": 0.9534,
1060
+ "step": 1480
1061
+ },
1062
+ {
1063
+ "epoch": 1.3036534675125793,
1064
+ "grad_norm": 1.3140227794647217,
1065
+ "learning_rate": 6.956286931592481e-06,
1066
+ "loss": 0.9888,
1067
+ "step": 1490
1068
+ },
1069
+ {
1070
+ "epoch": 1.3124042879019908,
1071
+ "grad_norm": 1.2232003211975098,
1072
+ "learning_rate": 6.909297610821444e-06,
1073
+ "loss": 0.9632,
1074
+ "step": 1500
1075
+ },
1076
+ {
1077
+ "epoch": 1.3211551082914024,
1078
+ "grad_norm": 1.190757393836975,
1079
+ "learning_rate": 6.86211003577259e-06,
1080
+ "loss": 0.9598,
1081
+ "step": 1510
1082
+ },
1083
+ {
1084
+ "epoch": 1.3299059286808137,
1085
+ "grad_norm": 1.180031657218933,
1086
+ "learning_rate": 6.81472910622611e-06,
1087
+ "loss": 0.9804,
1088
+ "step": 1520
1089
+ },
1090
+ {
1091
+ "epoch": 1.3386567490702252,
1092
+ "grad_norm": 1.2404285669326782,
1093
+ "learning_rate": 6.7671597420394e-06,
1094
+ "loss": 0.9915,
1095
+ "step": 1530
1096
+ },
1097
+ {
1098
+ "epoch": 1.3474075694596368,
1099
+ "grad_norm": 1.2322285175323486,
1100
+ "learning_rate": 6.719406882636196e-06,
1101
+ "loss": 0.9908,
1102
+ "step": 1540
1103
+ },
1104
+ {
1105
+ "epoch": 1.3561583898490484,
1106
+ "grad_norm": 1.1908921003341675,
1107
+ "learning_rate": 6.671475486493691e-06,
1108
+ "loss": 0.9716,
1109
+ "step": 1550
1110
+ },
1111
+ {
1112
+ "epoch": 1.36490921023846,
1113
+ "grad_norm": 1.2217786312103271,
1114
+ "learning_rate": 6.623370530627661e-06,
1115
+ "loss": 0.9602,
1116
+ "step": 1560
1117
+ },
1118
+ {
1119
+ "epoch": 1.3736600306278715,
1120
+ "grad_norm": 1.2622575759887695,
1121
+ "learning_rate": 6.575097010075665e-06,
1122
+ "loss": 0.9509,
1123
+ "step": 1570
1124
+ },
1125
+ {
1126
+ "epoch": 1.382410851017283,
1127
+ "grad_norm": 1.2419332265853882,
1128
+ "learning_rate": 6.526659937378392e-06,
1129
+ "loss": 0.9821,
1130
+ "step": 1580
1131
+ },
1132
+ {
1133
+ "epoch": 1.3911616714066943,
1134
+ "grad_norm": 1.3119240999221802,
1135
+ "learning_rate": 6.478064342059165e-06,
1136
+ "loss": 0.9823,
1137
+ "step": 1590
1138
+ },
1139
+ {
1140
+ "epoch": 1.3999124917961059,
1141
+ "grad_norm": 1.3174080848693848,
1142
+ "learning_rate": 6.429315270101702e-06,
1143
+ "loss": 0.9844,
1144
+ "step": 1600
1145
+ },
1146
+ {
1147
+ "epoch": 1.4086633121855174,
1148
+ "grad_norm": 1.2387707233428955,
1149
+ "learning_rate": 6.380417783426154e-06,
1150
+ "loss": 0.9545,
1151
+ "step": 1610
1152
+ },
1153
+ {
1154
+ "epoch": 1.4174141325749288,
1155
+ "grad_norm": 1.1914652585983276,
1156
+ "learning_rate": 6.331376959363501e-06,
1157
+ "loss": 0.9792,
1158
+ "step": 1620
1159
+ },
1160
+ {
1161
+ "epoch": 1.4261649529643403,
1162
+ "grad_norm": 1.2240160703659058,
1163
+ "learning_rate": 6.2821978901283386e-06,
1164
+ "loss": 0.9573,
1165
+ "step": 1630
1166
+ },
1167
+ {
1168
+ "epoch": 1.4349157733537519,
1169
+ "grad_norm": 1.2401772737503052,
1170
+ "learning_rate": 6.232885682290119e-06,
1171
+ "loss": 0.9669,
1172
+ "step": 1640
1173
+ },
1174
+ {
1175
+ "epoch": 1.4436665937431634,
1176
+ "grad_norm": 1.2027502059936523,
1177
+ "learning_rate": 6.183445456242905e-06,
1178
+ "loss": 0.9815,
1179
+ "step": 1650
1180
+ },
1181
+ {
1182
+ "epoch": 1.452417414132575,
1183
+ "grad_norm": 1.2344744205474854,
1184
+ "learning_rate": 6.13388234567369e-06,
1185
+ "loss": 0.9361,
1186
+ "step": 1660
1187
+ },
1188
+ {
1189
+ "epoch": 1.4611682345219865,
1190
+ "grad_norm": 1.2103825807571411,
1191
+ "learning_rate": 6.084201497029333e-06,
1192
+ "loss": 0.9744,
1193
+ "step": 1670
1194
+ },
1195
+ {
1196
+ "epoch": 1.469919054911398,
1197
+ "grad_norm": 1.3063782453536987,
1198
+ "learning_rate": 6.034408068982171e-06,
1199
+ "loss": 0.9681,
1200
+ "step": 1680
1201
+ },
1202
+ {
1203
+ "epoch": 1.4786698753008094,
1204
+ "grad_norm": 1.2130053043365479,
1205
+ "learning_rate": 5.984507231894364e-06,
1206
+ "loss": 0.9586,
1207
+ "step": 1690
1208
+ },
1209
+ {
1210
+ "epoch": 1.487420695690221,
1211
+ "grad_norm": 1.2866191864013672,
1212
+ "learning_rate": 5.934504167281022e-06,
1213
+ "loss": 0.9681,
1214
+ "step": 1700
1215
+ },
1216
+ {
1217
+ "epoch": 1.4961715160796325,
1218
+ "grad_norm": 1.3027279376983643,
1219
+ "learning_rate": 5.884404067272175e-06,
1220
+ "loss": 0.975,
1221
+ "step": 1710
1222
+ },
1223
+ {
1224
+ "epoch": 1.5049223364690438,
1225
+ "grad_norm": 1.2601852416992188,
1226
+ "learning_rate": 5.834212134073645e-06,
1227
+ "loss": 0.9388,
1228
+ "step": 1720
1229
+ },
1230
+ {
1231
+ "epoch": 1.5136731568584554,
1232
+ "grad_norm": 1.2413978576660156,
1233
+ "learning_rate": 5.783933579426867e-06,
1234
+ "loss": 0.9723,
1235
+ "step": 1730
1236
+ },
1237
+ {
1238
+ "epoch": 1.522423977247867,
1239
+ "grad_norm": 1.2116490602493286,
1240
+ "learning_rate": 5.733573624067717e-06,
1241
+ "loss": 0.973,
1242
+ "step": 1740
1243
+ },
1244
+ {
1245
+ "epoch": 1.5311747976372785,
1246
+ "grad_norm": 1.2060950994491577,
1247
+ "learning_rate": 5.683137497184424e-06,
1248
+ "loss": 0.9375,
1249
+ "step": 1750
1250
+ },
1251
+ {
1252
+ "epoch": 1.53992561802669,
1253
+ "grad_norm": 1.2472314834594727,
1254
+ "learning_rate": 5.63263043587457e-06,
1255
+ "loss": 1.0112,
1256
+ "step": 1760
1257
+ },
1258
+ {
1259
+ "epoch": 1.5486764384161016,
1260
+ "grad_norm": 1.2653073072433472,
1261
+ "learning_rate": 5.582057684601308e-06,
1262
+ "loss": 0.9778,
1263
+ "step": 1770
1264
+ },
1265
+ {
1266
+ "epoch": 1.5574272588055131,
1267
+ "grad_norm": 1.2270764112472534,
1268
+ "learning_rate": 5.531424494648789e-06,
1269
+ "loss": 0.9807,
1270
+ "step": 1780
1271
+ },
1272
+ {
1273
+ "epoch": 1.5661780791949247,
1274
+ "grad_norm": 1.2816975116729736,
1275
+ "learning_rate": 5.480736123576886e-06,
1276
+ "loss": 0.9847,
1277
+ "step": 1790
1278
+ },
1279
+ {
1280
+ "epoch": 1.574928899584336,
1281
+ "grad_norm": 1.2429735660552979,
1282
+ "learning_rate": 5.429997834675276e-06,
1283
+ "loss": 0.9886,
1284
+ "step": 1800
1285
+ },
1286
+ {
1287
+ "epoch": 1.5836797199737476,
1288
+ "grad_norm": 1.3056929111480713,
1289
+ "learning_rate": 5.379214896416908e-06,
1290
+ "loss": 0.9514,
1291
+ "step": 1810
1292
+ },
1293
+ {
1294
+ "epoch": 1.592430540363159,
1295
+ "grad_norm": 1.155527949333191,
1296
+ "learning_rate": 5.328392581910962e-06,
1297
+ "loss": 0.9593,
1298
+ "step": 1820
1299
+ },
1300
+ {
1301
+ "epoch": 1.6011813607525704,
1302
+ "grad_norm": 1.2867212295532227,
1303
+ "learning_rate": 5.277536168355293e-06,
1304
+ "loss": 0.9722,
1305
+ "step": 1830
1306
+ },
1307
+ {
1308
+ "epoch": 1.609932181141982,
1309
+ "grad_norm": 1.3213430643081665,
1310
+ "learning_rate": 5.226650936488474e-06,
1311
+ "loss": 0.9588,
1312
+ "step": 1840
1313
+ },
1314
+ {
1315
+ "epoch": 1.6186830015313936,
1316
+ "grad_norm": 1.2772150039672852,
1317
+ "learning_rate": 5.1757421700414645e-06,
1318
+ "loss": 0.9729,
1319
+ "step": 1850
1320
+ },
1321
+ {
1322
+ "epoch": 1.627433821920805,
1323
+ "grad_norm": 1.2589421272277832,
1324
+ "learning_rate": 5.124815155188966e-06,
1325
+ "loss": 1.0171,
1326
+ "step": 1860
1327
+ },
1328
+ {
1329
+ "epoch": 1.6361846423102167,
1330
+ "grad_norm": 1.2326509952545166,
1331
+ "learning_rate": 5.073875180000523e-06,
1332
+ "loss": 0.9622,
1333
+ "step": 1870
1334
+ },
1335
+ {
1336
+ "epoch": 1.6449354626996282,
1337
+ "grad_norm": 1.1593009233474731,
1338
+ "learning_rate": 5.022927533891434e-06,
1339
+ "loss": 0.9584,
1340
+ "step": 1880
1341
+ },
1342
+ {
1343
+ "epoch": 1.6536862830890398,
1344
+ "grad_norm": 1.303597331047058,
1345
+ "learning_rate": 4.97197750707352e-06,
1346
+ "loss": 0.9994,
1347
+ "step": 1890
1348
+ },
1349
+ {
1350
+ "epoch": 1.662437103478451,
1351
+ "grad_norm": 1.2183244228363037,
1352
+ "learning_rate": 4.9210303900058025e-06,
1353
+ "loss": 0.9582,
1354
+ "step": 1900
1355
+ },
1356
+ {
1357
+ "epoch": 1.6711879238678626,
1358
+ "grad_norm": 1.2271572351455688,
1359
+ "learning_rate": 4.870091472845166e-06,
1360
+ "loss": 0.9775,
1361
+ "step": 1910
1362
+ },
1363
+ {
1364
+ "epoch": 1.679938744257274,
1365
+ "grad_norm": 1.222212791442871,
1366
+ "learning_rate": 4.819166044897047e-06,
1367
+ "loss": 0.9578,
1368
+ "step": 1920
1369
+ },
1370
+ {
1371
+ "epoch": 1.6886895646466855,
1372
+ "grad_norm": 1.2827479839324951,
1373
+ "learning_rate": 4.7682593940662156e-06,
1374
+ "loss": 0.982,
1375
+ "step": 1930
1376
+ },
1377
+ {
1378
+ "epoch": 1.697440385036097,
1379
+ "grad_norm": 1.2533340454101562,
1380
+ "learning_rate": 4.717376806307696e-06,
1381
+ "loss": 0.9735,
1382
+ "step": 1940
1383
+ },
1384
+ {
1385
+ "epoch": 1.7061912054255086,
1386
+ "grad_norm": 1.267432451248169,
1387
+ "learning_rate": 4.666523565077889e-06,
1388
+ "loss": 0.9598,
1389
+ "step": 1950
1390
+ },
1391
+ {
1392
+ "epoch": 1.7149420258149202,
1393
+ "grad_norm": 1.2450517416000366,
1394
+ "learning_rate": 4.615704950785965e-06,
1395
+ "loss": 0.9751,
1396
+ "step": 1960
1397
+ },
1398
+ {
1399
+ "epoch": 1.7236928462043317,
1400
+ "grad_norm": 1.2049565315246582,
1401
+ "learning_rate": 4.5649262402455705e-06,
1402
+ "loss": 0.9737,
1403
+ "step": 1970
1404
+ },
1405
+ {
1406
+ "epoch": 1.7324436665937433,
1407
+ "grad_norm": 1.2524805068969727,
1408
+ "learning_rate": 4.514192706126885e-06,
1409
+ "loss": 0.9662,
1410
+ "step": 1980
1411
+ },
1412
+ {
1413
+ "epoch": 1.7411944869831548,
1414
+ "grad_norm": 1.2566184997558594,
1415
+ "learning_rate": 4.463509616409144e-06,
1416
+ "loss": 0.9618,
1417
+ "step": 1990
1418
+ },
1419
+ {
1420
+ "epoch": 1.7499453073725662,
1421
+ "grad_norm": 1.176069736480713,
1422
+ "learning_rate": 4.4128822338336295e-06,
1423
+ "loss": 0.9521,
1424
+ "step": 2000
1425
+ },
1426
+ {
1427
+ "epoch": 1.7586961277619777,
1428
+ "grad_norm": 1.1588139533996582,
1429
+ "learning_rate": 4.362315815357197e-06,
1430
+ "loss": 0.9488,
1431
+ "step": 2010
1432
+ },
1433
+ {
1434
+ "epoch": 1.767446948151389,
1435
+ "grad_norm": 1.1984119415283203,
1436
+ "learning_rate": 4.311815611606423e-06,
1437
+ "loss": 0.9702,
1438
+ "step": 2020
1439
+ },
1440
+ {
1441
+ "epoch": 1.7761977685408006,
1442
+ "grad_norm": 1.1387180089950562,
1443
+ "learning_rate": 4.261386866332397e-06,
1444
+ "loss": 0.9528,
1445
+ "step": 2030
1446
+ },
1447
+ {
1448
+ "epoch": 1.7849485889302121,
1449
+ "grad_norm": 1.1617263555526733,
1450
+ "learning_rate": 4.211034815866228e-06,
1451
+ "loss": 0.951,
1452
+ "step": 2040
1453
+ },
1454
+ {
1455
+ "epoch": 1.7936994093196237,
1456
+ "grad_norm": 1.1776400804519653,
1457
+ "learning_rate": 4.160764688575321e-06,
1458
+ "loss": 0.9634,
1459
+ "step": 2050
1460
+ },
1461
+ {
1462
+ "epoch": 1.8024502297090352,
1463
+ "grad_norm": 1.25057852268219,
1464
+ "learning_rate": 4.110581704320495e-06,
1465
+ "loss": 0.9633,
1466
+ "step": 2060
1467
+ },
1468
+ {
1469
+ "epoch": 1.8112010500984468,
1470
+ "grad_norm": 1.2121225595474243,
1471
+ "learning_rate": 4.0604910739139575e-06,
1472
+ "loss": 0.9749,
1473
+ "step": 2070
1474
+ },
1475
+ {
1476
+ "epoch": 1.8199518704878583,
1477
+ "grad_norm": 1.3024873733520508,
1478
+ "learning_rate": 4.0104979985782394e-06,
1479
+ "loss": 0.9744,
1480
+ "step": 2080
1481
+ },
1482
+ {
1483
+ "epoch": 1.8287026908772699,
1484
+ "grad_norm": 1.1998348236083984,
1485
+ "learning_rate": 3.960607669406129e-06,
1486
+ "loss": 0.9622,
1487
+ "step": 2090
1488
+ },
1489
+ {
1490
+ "epoch": 1.8374535112666812,
1491
+ "grad_norm": 1.2691677808761597,
1492
+ "learning_rate": 3.9108252668216294e-06,
1493
+ "loss": 0.9702,
1494
+ "step": 2100
1495
+ },
1496
+ {
1497
+ "epoch": 1.8462043316560928,
1498
+ "grad_norm": 1.2797294855117798,
1499
+ "learning_rate": 3.861155960042057e-06,
1500
+ "loss": 0.9294,
1501
+ "step": 2110
1502
+ },
1503
+ {
1504
+ "epoch": 1.8549551520455043,
1505
+ "grad_norm": 1.2609057426452637,
1506
+ "learning_rate": 3.811604906541287e-06,
1507
+ "loss": 0.9624,
1508
+ "step": 2120
1509
+ },
1510
+ {
1511
+ "epoch": 1.8637059724349156,
1512
+ "grad_norm": 1.2273874282836914,
1513
+ "learning_rate": 3.762177251514218e-06,
1514
+ "loss": 0.9563,
1515
+ "step": 2130
1516
+ },
1517
+ {
1518
+ "epoch": 1.8724567928243272,
1519
+ "grad_norm": 1.225128173828125,
1520
+ "learning_rate": 3.7128781273425156e-06,
1521
+ "loss": 0.9587,
1522
+ "step": 2140
1523
+ },
1524
+ {
1525
+ "epoch": 1.8812076132137387,
1526
+ "grad_norm": 1.2284705638885498,
1527
+ "learning_rate": 3.6637126530616816e-06,
1528
+ "loss": 0.9401,
1529
+ "step": 2150
1530
+ },
1531
+ {
1532
+ "epoch": 1.8899584336031503,
1533
+ "grad_norm": 1.2802456617355347,
1534
+ "learning_rate": 3.6146859338295248e-06,
1535
+ "loss": 0.9386,
1536
+ "step": 2160
1537
+ },
1538
+ {
1539
+ "epoch": 1.8987092539925619,
1540
+ "grad_norm": 1.22531259059906,
1541
+ "learning_rate": 3.5658030603960437e-06,
1542
+ "loss": 0.9652,
1543
+ "step": 2170
1544
+ },
1545
+ {
1546
+ "epoch": 1.9074600743819734,
1547
+ "grad_norm": 1.2205225229263306,
1548
+ "learning_rate": 3.5170691085748325e-06,
1549
+ "loss": 0.9418,
1550
+ "step": 2180
1551
+ },
1552
+ {
1553
+ "epoch": 1.916210894771385,
1554
+ "grad_norm": 1.1588237285614014,
1555
+ "learning_rate": 3.4684891387160292e-06,
1556
+ "loss": 0.9624,
1557
+ "step": 2190
1558
+ },
1559
+ {
1560
+ "epoch": 1.9249617151607963,
1561
+ "grad_norm": 1.2422109842300415,
1562
+ "learning_rate": 3.42006819518086e-06,
1563
+ "loss": 0.9714,
1564
+ "step": 2200
1565
+ },
1566
+ {
1567
+ "epoch": 1.9337125355502078,
1568
+ "grad_norm": 1.2960257530212402,
1569
+ "learning_rate": 3.371811305817858e-06,
1570
+ "loss": 0.9521,
1571
+ "step": 2210
1572
+ },
1573
+ {
1574
+ "epoch": 1.9424633559396194,
1575
+ "grad_norm": 1.1022367477416992,
1576
+ "learning_rate": 3.3237234814407904e-06,
1577
+ "loss": 0.9775,
1578
+ "step": 2220
1579
+ },
1580
+ {
1581
+ "epoch": 1.9512141763290307,
1582
+ "grad_norm": 1.160735011100769,
1583
+ "learning_rate": 3.275809715308349e-06,
1584
+ "loss": 0.9497,
1585
+ "step": 2230
1586
+ },
1587
+ {
1588
+ "epoch": 1.9599649967184423,
1589
+ "grad_norm": 1.2306522130966187,
1590
+ "learning_rate": 3.2280749826056736e-06,
1591
+ "loss": 0.9544,
1592
+ "step": 2240
1593
+ },
1594
+ {
1595
+ "epoch": 1.9687158171078538,
1596
+ "grad_norm": 1.1739164590835571,
1597
+ "learning_rate": 3.180524239927749e-06,
1598
+ "loss": 0.9526,
1599
+ "step": 2250
1600
+ },
1601
+ {
1602
+ "epoch": 1.9774666374972654,
1603
+ "grad_norm": 1.2096645832061768,
1604
+ "learning_rate": 3.1331624247647276e-06,
1605
+ "loss": 0.9503,
1606
+ "step": 2260
1607
+ },
1608
+ {
1609
+ "epoch": 1.986217457886677,
1610
+ "grad_norm": 1.2347972393035889,
1611
+ "learning_rate": 3.085994454989232e-06,
1612
+ "loss": 0.9473,
1613
+ "step": 2270
1614
+ },
1615
+ {
1616
+ "epoch": 1.9949682782760885,
1617
+ "grad_norm": 1.28371000289917,
1618
+ "learning_rate": 3.039025228345725e-06,
1619
+ "loss": 0.9787,
1620
+ "step": 2280
1621
+ },
1622
+ {
1623
+ "epoch": 2.0035003281557646,
1624
+ "grad_norm": 1.1137796640396118,
1625
+ "learning_rate": 2.992259621941916e-06,
1626
+ "loss": 0.9234,
1627
+ "step": 2290
1628
+ },
1629
+ {
1630
+ "epoch": 2.012251148545176,
1631
+ "grad_norm": 1.3428007364273071,
1632
+ "learning_rate": 2.9457024917423594e-06,
1633
+ "loss": 0.8978,
1634
+ "step": 2300
1635
+ },
1636
+ {
1637
+ "epoch": 2.0210019689345877,
1638
+ "grad_norm": 1.2651664018630981,
1639
+ "learning_rate": 2.8993586720642307e-06,
1640
+ "loss": 0.8992,
1641
+ "step": 2310
1642
+ },
1643
+ {
1644
+ "epoch": 2.0297527893239993,
1645
+ "grad_norm": 1.1616652011871338,
1646
+ "learning_rate": 2.85323297507534e-06,
1647
+ "loss": 0.8695,
1648
+ "step": 2320
1649
+ },
1650
+ {
1651
+ "epoch": 2.038503609713411,
1652
+ "grad_norm": 1.2452731132507324,
1653
+ "learning_rate": 2.8073301902944518e-06,
1654
+ "loss": 0.901,
1655
+ "step": 2330
1656
+ },
1657
+ {
1658
+ "epoch": 2.047254430102822,
1659
+ "grad_norm": 1.2381782531738281,
1660
+ "learning_rate": 2.761655084093976e-06,
1661
+ "loss": 0.9096,
1662
+ "step": 2340
1663
+ },
1664
+ {
1665
+ "epoch": 2.0560052504922335,
1666
+ "grad_norm": 1.265434980392456,
1667
+ "learning_rate": 2.7162123992050316e-06,
1668
+ "loss": 0.8984,
1669
+ "step": 2350
1670
+ },
1671
+ {
1672
+ "epoch": 2.064756070881645,
1673
+ "grad_norm": 1.2683861255645752,
1674
+ "learning_rate": 2.671006854224982e-06,
1675
+ "loss": 0.8864,
1676
+ "step": 2360
1677
+ },
1678
+ {
1679
+ "epoch": 2.0735068912710566,
1680
+ "grad_norm": 1.2859617471694946,
1681
+ "learning_rate": 2.626043143127477e-06,
1682
+ "loss": 0.8728,
1683
+ "step": 2370
1684
+ },
1685
+ {
1686
+ "epoch": 2.082257711660468,
1687
+ "grad_norm": 1.3000761270523071,
1688
+ "learning_rate": 2.5813259347750445e-06,
1689
+ "loss": 0.932,
1690
+ "step": 2380
1691
+ },
1692
+ {
1693
+ "epoch": 2.0910085320498797,
1694
+ "grad_norm": 1.23272705078125,
1695
+ "learning_rate": 2.5368598724342974e-06,
1696
+ "loss": 0.9114,
1697
+ "step": 2390
1698
+ },
1699
+ {
1700
+ "epoch": 2.0997593524392912,
1701
+ "grad_norm": 1.243643879890442,
1702
+ "learning_rate": 2.4926495732937877e-06,
1703
+ "loss": 0.8987,
1704
+ "step": 2400
1705
+ },
1706
+ {
1707
+ "epoch": 2.108510172828703,
1708
+ "grad_norm": 1.2681580781936646,
1709
+ "learning_rate": 2.4486996279845864e-06,
1710
+ "loss": 0.9277,
1711
+ "step": 2410
1712
+ },
1713
+ {
1714
+ "epoch": 2.1172609932181143,
1715
+ "grad_norm": 1.2045451402664185,
1716
+ "learning_rate": 2.4050146001035983e-06,
1717
+ "loss": 0.8988,
1718
+ "step": 2420
1719
+ },
1720
+ {
1721
+ "epoch": 2.126011813607526,
1722
+ "grad_norm": 1.2446624040603638,
1723
+ "learning_rate": 2.361599025739695e-06,
1724
+ "loss": 0.889,
1725
+ "step": 2430
1726
+ },
1727
+ {
1728
+ "epoch": 2.134762633996937,
1729
+ "grad_norm": 1.298621654510498,
1730
+ "learning_rate": 2.318457413002718e-06,
1731
+ "loss": 0.898,
1732
+ "step": 2440
1733
+ },
1734
+ {
1735
+ "epoch": 2.1435134543863485,
1736
+ "grad_norm": 1.2545632123947144,
1737
+ "learning_rate": 2.2755942415553593e-06,
1738
+ "loss": 0.8675,
1739
+ "step": 2450
1740
+ },
1741
+ {
1742
+ "epoch": 2.15226427477576,
1743
+ "grad_norm": 1.1765785217285156,
1744
+ "learning_rate": 2.233013962148016e-06,
1745
+ "loss": 0.853,
1746
+ "step": 2460
1747
+ },
1748
+ {
1749
+ "epoch": 2.1610150951651717,
1750
+ "grad_norm": 1.3674156665802002,
1751
+ "learning_rate": 2.1907209961566418e-06,
1752
+ "loss": 0.9134,
1753
+ "step": 2470
1754
+ },
1755
+ {
1756
+ "epoch": 2.169765915554583,
1757
+ "grad_norm": 1.337667465209961,
1758
+ "learning_rate": 2.1487197351236443e-06,
1759
+ "loss": 0.8766,
1760
+ "step": 2480
1761
+ },
1762
+ {
1763
+ "epoch": 2.1785167359439948,
1764
+ "grad_norm": 1.240268349647522,
1765
+ "learning_rate": 2.107014540301886e-06,
1766
+ "loss": 0.8921,
1767
+ "step": 2490
1768
+ },
1769
+ {
1770
+ "epoch": 2.1872675563334063,
1771
+ "grad_norm": 1.2723026275634766,
1772
+ "learning_rate": 2.0656097422018346e-06,
1773
+ "loss": 0.8948,
1774
+ "step": 2500
1775
+ },
1776
+ {
1777
+ "epoch": 2.196018376722818,
1778
+ "grad_norm": 1.244240164756775,
1779
+ "learning_rate": 2.0245096401418853e-06,
1780
+ "loss": 0.902,
1781
+ "step": 2510
1782
+ },
1783
+ {
1784
+ "epoch": 2.2047691971122294,
1785
+ "grad_norm": 1.2332072257995605,
1786
+ "learning_rate": 1.9837185018019465e-06,
1787
+ "loss": 0.9042,
1788
+ "step": 2520
1789
+ },
1790
+ {
1791
+ "epoch": 2.213520017501641,
1792
+ "grad_norm": 1.3075368404388428,
1793
+ "learning_rate": 1.9432405627802963e-06,
1794
+ "loss": 0.8743,
1795
+ "step": 2530
1796
+ },
1797
+ {
1798
+ "epoch": 2.2222708378910525,
1799
+ "grad_norm": 1.2511454820632935,
1800
+ "learning_rate": 1.9030800261537742e-06,
1801
+ "loss": 0.8919,
1802
+ "step": 2540
1803
+ },
1804
+ {
1805
+ "epoch": 2.2310216582804636,
1806
+ "grad_norm": 1.2848820686340332,
1807
+ "learning_rate": 1.8632410620413505e-06,
1808
+ "loss": 0.8948,
1809
+ "step": 2550
1810
+ },
1811
+ {
1812
+ "epoch": 2.239772478669875,
1813
+ "grad_norm": 1.1853325366973877,
1814
+ "learning_rate": 1.823727807171115e-06,
1815
+ "loss": 0.9262,
1816
+ "step": 2560
1817
+ },
1818
+ {
1819
+ "epoch": 2.2485232990592867,
1820
+ "grad_norm": 1.2825355529785156,
1821
+ "learning_rate": 1.7845443644507355e-06,
1822
+ "loss": 0.8744,
1823
+ "step": 2570
1824
+ },
1825
+ {
1826
+ "epoch": 2.2572741194486983,
1827
+ "grad_norm": 1.2997822761535645,
1828
+ "learning_rate": 1.7456948025414288e-06,
1829
+ "loss": 0.9153,
1830
+ "step": 2580
1831
+ },
1832
+ {
1833
+ "epoch": 2.26602493983811,
1834
+ "grad_norm": 1.2967973947525024,
1835
+ "learning_rate": 1.7071831554354822e-06,
1836
+ "loss": 0.8702,
1837
+ "step": 2590
1838
+ },
1839
+ {
1840
+ "epoch": 2.2747757602275214,
1841
+ "grad_norm": 1.2506667375564575,
1842
+ "learning_rate": 1.6690134220373866e-06,
1843
+ "loss": 0.9055,
1844
+ "step": 2600
1845
+ },
1846
+ {
1847
+ "epoch": 2.283526580616933,
1848
+ "grad_norm": 1.2581911087036133,
1849
+ "learning_rate": 1.631189565748596e-06,
1850
+ "loss": 0.8953,
1851
+ "step": 2610
1852
+ },
1853
+ {
1854
+ "epoch": 2.2922774010063445,
1855
+ "grad_norm": 1.2846101522445679,
1856
+ "learning_rate": 1.5937155140559885e-06,
1857
+ "loss": 0.8857,
1858
+ "step": 2620
1859
+ },
1860
+ {
1861
+ "epoch": 2.301028221395756,
1862
+ "grad_norm": 1.317673683166504,
1863
+ "learning_rate": 1.556595158124049e-06,
1864
+ "loss": 0.8936,
1865
+ "step": 2630
1866
+ },
1867
+ {
1868
+ "epoch": 2.309779041785167,
1869
+ "grad_norm": 1.3196806907653809,
1870
+ "learning_rate": 1.5198323523908266e-06,
1871
+ "loss": 0.8889,
1872
+ "step": 2640
1873
+ },
1874
+ {
1875
+ "epoch": 2.3185298621745787,
1876
+ "grad_norm": 1.2810077667236328,
1877
+ "learning_rate": 1.483430914167701e-06,
1878
+ "loss": 0.9023,
1879
+ "step": 2650
1880
+ },
1881
+ {
1882
+ "epoch": 2.3272806825639902,
1883
+ "grad_norm": 1.3437639474868774,
1884
+ "learning_rate": 1.447394623243018e-06,
1885
+ "loss": 0.9372,
1886
+ "step": 2660
1887
+ },
1888
+ {
1889
+ "epoch": 2.336031502953402,
1890
+ "grad_norm": 1.2470753192901611,
1891
+ "learning_rate": 1.41172722148959e-06,
1892
+ "loss": 0.8692,
1893
+ "step": 2670
1894
+ },
1895
+ {
1896
+ "epoch": 2.3447823233428133,
1897
+ "grad_norm": 1.226980209350586,
1898
+ "learning_rate": 1.3764324124761735e-06,
1899
+ "loss": 0.9238,
1900
+ "step": 2680
1901
+ },
1902
+ {
1903
+ "epoch": 2.353533143732225,
1904
+ "grad_norm": 1.2623775005340576,
1905
+ "learning_rate": 1.3415138610828987e-06,
1906
+ "loss": 0.8965,
1907
+ "step": 2690
1908
+ },
1909
+ {
1910
+ "epoch": 2.3622839641216364,
1911
+ "grad_norm": 1.1816771030426025,
1912
+ "learning_rate": 1.3069751931207164e-06,
1913
+ "loss": 0.863,
1914
+ "step": 2700
1915
+ },
1916
+ {
1917
+ "epoch": 2.371034784511048,
1918
+ "grad_norm": 1.0856176614761353,
1919
+ "learning_rate": 1.2728199949549143e-06,
1920
+ "loss": 0.8694,
1921
+ "step": 2710
1922
+ },
1923
+ {
1924
+ "epoch": 2.3797856049004595,
1925
+ "grad_norm": 1.1386377811431885,
1926
+ "learning_rate": 1.2390518131327173e-06,
1927
+ "loss": 0.8742,
1928
+ "step": 2720
1929
+ },
1930
+ {
1931
+ "epoch": 2.388536425289871,
1932
+ "grad_norm": 1.1820670366287231,
1933
+ "learning_rate": 1.205674154015033e-06,
1934
+ "loss": 0.8797,
1935
+ "step": 2730
1936
+ },
1937
+ {
1938
+ "epoch": 2.3972872456792826,
1939
+ "grad_norm": 1.3150583505630493,
1940
+ "learning_rate": 1.1726904834123582e-06,
1941
+ "loss": 0.9003,
1942
+ "step": 2740
1943
+ },
1944
+ {
1945
+ "epoch": 2.4060380660686937,
1946
+ "grad_norm": 1.30619478225708,
1947
+ "learning_rate": 1.1401042262249056e-06,
1948
+ "loss": 0.9086,
1949
+ "step": 2750
1950
+ },
1951
+ {
1952
+ "epoch": 2.4147888864581053,
1953
+ "grad_norm": 1.3051726818084717,
1954
+ "learning_rate": 1.10791876608698e-06,
1955
+ "loss": 0.8602,
1956
+ "step": 2760
1957
+ },
1958
+ {
1959
+ "epoch": 2.423539706847517,
1960
+ "grad_norm": 1.225061297416687,
1961
+ "learning_rate": 1.0761374450156203e-06,
1962
+ "loss": 0.8669,
1963
+ "step": 2770
1964
+ },
1965
+ {
1966
+ "epoch": 2.4322905272369284,
1967
+ "grad_norm": 1.1500948667526245,
1968
+ "learning_rate": 1.04476356306359e-06,
1969
+ "loss": 0.8733,
1970
+ "step": 2780
1971
+ },
1972
+ {
1973
+ "epoch": 2.44104134762634,
1974
+ "grad_norm": 1.2501094341278076,
1975
+ "learning_rate": 1.0138003779767102e-06,
1976
+ "loss": 0.9025,
1977
+ "step": 2790
1978
+ },
1979
+ {
1980
+ "epoch": 2.4497921680157515,
1981
+ "grad_norm": 1.1673866510391235,
1982
+ "learning_rate": 9.832511048555825e-07,
1983
+ "loss": 0.918,
1984
+ "step": 2800
1985
+ },
1986
+ {
1987
+ "epoch": 2.458542988405163,
1988
+ "grad_norm": 1.2898247241973877,
1989
+ "learning_rate": 9.53118915821748e-07,
1990
+ "loss": 0.8854,
1991
+ "step": 2810
1992
+ },
1993
+ {
1994
+ "epoch": 2.4672938087945746,
1995
+ "grad_norm": 1.201026201248169,
1996
+ "learning_rate": 9.234069396883066e-07,
1997
+ "loss": 0.9144,
1998
+ "step": 2820
1999
+ },
2000
+ {
2001
+ "epoch": 2.476044629183986,
2002
+ "grad_norm": 1.1994189023971558,
2003
+ "learning_rate": 8.941182616350319e-07,
2004
+ "loss": 0.889,
2005
+ "step": 2830
2006
+ },
2007
+ {
2008
+ "epoch": 2.4847954495733973,
2009
+ "grad_norm": 1.2872554063796997,
2010
+ "learning_rate": 8.652559228880169e-07,
2011
+ "loss": 0.8989,
2012
+ "step": 2840
2013
+ },
2014
+ {
2015
+ "epoch": 2.493546269962809,
2016
+ "grad_norm": 1.286000370979309,
2017
+ "learning_rate": 8.368229204038897e-07,
2018
+ "loss": 0.9244,
2019
+ "step": 2850
2020
+ },
2021
+ {
2022
+ "epoch": 2.5022970903522204,
2023
+ "grad_norm": 1.227409839630127,
2024
+ "learning_rate": 8.088222065586132e-07,
2025
+ "loss": 0.9,
2026
+ "step": 2860
2027
+ },
2028
+ {
2029
+ "epoch": 2.511047910741632,
2030
+ "grad_norm": 1.2374303340911865,
2031
+ "learning_rate": 7.812566888409228e-07,
2032
+ "loss": 0.9094,
2033
+ "step": 2870
2034
+ },
2035
+ {
2036
+ "epoch": 2.5197987311310435,
2037
+ "grad_norm": 1.3504208326339722,
2038
+ "learning_rate": 7.541292295504327e-07,
2039
+ "loss": 0.8945,
2040
+ "step": 2880
2041
+ },
2042
+ {
2043
+ "epoch": 2.528549551520455,
2044
+ "grad_norm": 1.203357219696045,
2045
+ "learning_rate": 7.27442645500413e-07,
2046
+ "loss": 0.8938,
2047
+ "step": 2890
2048
+ },
2049
+ {
2050
+ "epoch": 2.5373003719098666,
2051
+ "grad_norm": 1.2361124753952026,
2052
+ "learning_rate": 7.011997077253113e-07,
2053
+ "loss": 0.8804,
2054
+ "step": 2900
2055
+ },
2056
+ {
2057
+ "epoch": 2.546051192299278,
2058
+ "grad_norm": 1.2695845365524292,
2059
+ "learning_rate": 6.75403141193014e-07,
2060
+ "loss": 0.901,
2061
+ "step": 2910
2062
+ },
2063
+ {
2064
+ "epoch": 2.5548020126886897,
2065
+ "grad_norm": 1.3083370923995972,
2066
+ "learning_rate": 6.500556245218986e-07,
2067
+ "loss": 0.903,
2068
+ "step": 2920
2069
+ },
2070
+ {
2071
+ "epoch": 2.5635528330781012,
2072
+ "grad_norm": 1.253382921218872,
2073
+ "learning_rate": 6.25159789702695e-07,
2074
+ "loss": 0.8944,
2075
+ "step": 2930
2076
+ },
2077
+ {
2078
+ "epoch": 2.5723036534675128,
2079
+ "grad_norm": 1.2046483755111694,
2080
+ "learning_rate": 6.007182218251883e-07,
2081
+ "loss": 0.8654,
2082
+ "step": 2940
2083
+ },
2084
+ {
2085
+ "epoch": 2.5810544738569243,
2086
+ "grad_norm": 1.286098837852478,
2087
+ "learning_rate": 5.767334588097979e-07,
2088
+ "loss": 0.8889,
2089
+ "step": 2950
2090
+ },
2091
+ {
2092
+ "epoch": 2.5898052942463354,
2093
+ "grad_norm": 1.212591290473938,
2094
+ "learning_rate": 5.532079911440419e-07,
2095
+ "loss": 0.8724,
2096
+ "step": 2960
2097
+ },
2098
+ {
2099
+ "epoch": 2.598556114635747,
2100
+ "grad_norm": 1.2204617261886597,
2101
+ "learning_rate": 5.301442616239371e-07,
2102
+ "loss": 0.8696,
2103
+ "step": 2970
2104
+ },
2105
+ {
2106
+ "epoch": 2.6073069350251585,
2107
+ "grad_norm": 1.1826462745666504,
2108
+ "learning_rate": 5.075446651003535e-07,
2109
+ "loss": 0.8542,
2110
+ "step": 2980
2111
+ },
2112
+ {
2113
+ "epoch": 2.61605775541457,
2114
+ "grad_norm": 1.270414113998413,
2115
+ "learning_rate": 4.854115482303345e-07,
2116
+ "loss": 0.8681,
2117
+ "step": 2990
2118
+ },
2119
+ {
2120
+ "epoch": 2.6248085758039816,
2121
+ "grad_norm": 1.2174789905548096,
2122
+ "learning_rate": 4.6374720923343275e-07,
2123
+ "loss": 0.9086,
2124
+ "step": 3000
2125
+ },
2126
+ {
2127
+ "epoch": 2.633559396193393,
2128
+ "grad_norm": 1.2752397060394287,
2129
+ "learning_rate": 4.4255389765307077e-07,
2130
+ "loss": 0.8978,
2131
+ "step": 3010
2132
+ },
2133
+ {
2134
+ "epoch": 2.6423102165828047,
2135
+ "grad_norm": 1.2312636375427246,
2136
+ "learning_rate": 4.2183381412295666e-07,
2137
+ "loss": 0.8834,
2138
+ "step": 3020
2139
+ },
2140
+ {
2141
+ "epoch": 2.6510610369722163,
2142
+ "grad_norm": 1.1354252099990845,
2143
+ "learning_rate": 4.015891101385794e-07,
2144
+ "loss": 0.8863,
2145
+ "step": 3030
2146
+ },
2147
+ {
2148
+ "epoch": 2.6598118573616274,
2149
+ "grad_norm": 1.2844924926757812,
2150
+ "learning_rate": 3.8182188783380534e-07,
2151
+ "loss": 0.8984,
2152
+ "step": 3040
2153
+ },
2154
+ {
2155
+ "epoch": 2.668562677751039,
2156
+ "grad_norm": 1.2184648513793945,
2157
+ "learning_rate": 3.6253419976260027e-07,
2158
+ "loss": 0.904,
2159
+ "step": 3050
2160
+ },
2161
+ {
2162
+ "epoch": 2.6773134981404505,
2163
+ "grad_norm": 1.2693063020706177,
2164
+ "learning_rate": 3.437280486858996e-07,
2165
+ "loss": 0.8813,
2166
+ "step": 3060
2167
+ },
2168
+ {
2169
+ "epoch": 2.686064318529862,
2170
+ "grad_norm": 1.3009639978408813,
2171
+ "learning_rate": 3.2540538736364977e-07,
2172
+ "loss": 0.8851,
2173
+ "step": 3070
2174
+ },
2175
+ {
2176
+ "epoch": 2.6948151389192736,
2177
+ "grad_norm": 1.2336896657943726,
2178
+ "learning_rate": 3.075681183520407e-07,
2179
+ "loss": 0.8994,
2180
+ "step": 3080
2181
+ },
2182
+ {
2183
+ "epoch": 2.703565959308685,
2184
+ "grad_norm": 1.2099403142929077,
2185
+ "learning_rate": 2.9021809380595066e-07,
2186
+ "loss": 0.899,
2187
+ "step": 3090
2188
+ },
2189
+ {
2190
+ "epoch": 2.7123167796980967,
2191
+ "grad_norm": 1.1838281154632568,
2192
+ "learning_rate": 2.7335711528663047e-07,
2193
+ "loss": 0.8951,
2194
+ "step": 3100
2195
+ },
2196
+ {
2197
+ "epoch": 2.7210676000875083,
2198
+ "grad_norm": 1.2549548149108887,
2199
+ "learning_rate": 2.569869335746267e-07,
2200
+ "loss": 0.8795,
2201
+ "step": 3110
2202
+ },
2203
+ {
2204
+ "epoch": 2.72981842047692,
2205
+ "grad_norm": 1.189772605895996,
2206
+ "learning_rate": 2.411092484879957e-07,
2207
+ "loss": 0.8632,
2208
+ "step": 3120
2209
+ },
2210
+ {
2211
+ "epoch": 2.7385692408663314,
2212
+ "grad_norm": 1.4026386737823486,
2213
+ "learning_rate": 2.2572570870579845e-07,
2214
+ "loss": 0.905,
2215
+ "step": 3130
2216
+ },
2217
+ {
2218
+ "epoch": 2.747320061255743,
2219
+ "grad_norm": 1.2943899631500244,
2220
+ "learning_rate": 2.1083791159690615e-07,
2221
+ "loss": 0.8858,
2222
+ "step": 3140
2223
+ },
2224
+ {
2225
+ "epoch": 2.7560708816451545,
2226
+ "grad_norm": 1.3465709686279297,
2227
+ "learning_rate": 1.9644740305413636e-07,
2228
+ "loss": 0.8963,
2229
+ "step": 3150
2230
+ },
2231
+ {
2232
+ "epoch": 2.764821702034566,
2233
+ "grad_norm": 1.2346800565719604,
2234
+ "learning_rate": 1.8255567733373392e-07,
2235
+ "loss": 0.8928,
2236
+ "step": 3160
2237
+ },
2238
+ {
2239
+ "epoch": 2.773572522423977,
2240
+ "grad_norm": 1.255714774131775,
2241
+ "learning_rate": 1.691641769002139e-07,
2242
+ "loss": 0.8977,
2243
+ "step": 3170
2244
+ },
2245
+ {
2246
+ "epoch": 2.7823233428133887,
2247
+ "grad_norm": 1.3175842761993408,
2248
+ "learning_rate": 1.5627429227657864e-07,
2249
+ "loss": 0.9226,
2250
+ "step": 3180
2251
+ },
2252
+ {
2253
+ "epoch": 2.7910741632028,
2254
+ "grad_norm": 1.333662748336792,
2255
+ "learning_rate": 1.4388736189993257e-07,
2256
+ "loss": 0.8932,
2257
+ "step": 3190
2258
+ },
2259
+ {
2260
+ "epoch": 2.7998249835922118,
2261
+ "grad_norm": 1.2100019454956055,
2262
+ "learning_rate": 1.320046719825069e-07,
2263
+ "loss": 0.8886,
2264
+ "step": 3200
2265
+ },
2266
+ {
2267
+ "epoch": 2.8085758039816233,
2268
+ "grad_norm": 1.2037444114685059,
2269
+ "learning_rate": 1.2062745637809737e-07,
2270
+ "loss": 0.8912,
2271
+ "step": 3210
2272
+ },
2273
+ {
2274
+ "epoch": 2.817326624371035,
2275
+ "grad_norm": 1.2076711654663086,
2276
+ "learning_rate": 1.0975689645395193e-07,
2277
+ "loss": 0.8627,
2278
+ "step": 3220
2279
+ },
2280
+ {
2281
+ "epoch": 2.8260774447604464,
2282
+ "grad_norm": 1.2997610569000244,
2283
+ "learning_rate": 9.939412096809986e-08,
2284
+ "loss": 0.899,
2285
+ "step": 3230
2286
+ },
2287
+ {
2288
+ "epoch": 2.8348282651498575,
2289
+ "grad_norm": 1.171425461769104,
2290
+ "learning_rate": 8.954020595214441e-08,
2291
+ "loss": 0.9032,
2292
+ "step": 3240
2293
+ },
2294
+ {
2295
+ "epoch": 2.843579085539269,
2296
+ "grad_norm": 1.2966265678405762,
2297
+ "learning_rate": 8.019617459953333e-08,
2298
+ "loss": 0.8888,
2299
+ "step": 3250
2300
+ },
2301
+ {
2302
+ "epoch": 2.8523299059286806,
2303
+ "grad_norm": 1.27556574344635,
2304
+ "learning_rate": 7.136299715931438e-08,
2305
+ "loss": 0.8775,
2306
+ "step": 3260
2307
+ },
2308
+ {
2309
+ "epoch": 2.861080726318092,
2310
+ "grad_norm": 1.2437477111816406,
2311
+ "learning_rate": 6.304159083538652e-08,
2312
+ "loss": 0.8696,
2313
+ "step": 3270
2314
+ },
2315
+ {
2316
+ "epoch": 2.8698315467075037,
2317
+ "grad_norm": 1.2502518892288208,
2318
+ "learning_rate": 5.523281969126271e-08,
2319
+ "loss": 0.8915,
2320
+ "step": 3280
2321
+ },
2322
+ {
2323
+ "epoch": 2.8785823670969153,
2324
+ "grad_norm": 1.250112771987915,
2325
+ "learning_rate": 4.7937494560349464e-08,
2326
+ "loss": 0.8792,
2327
+ "step": 3290
2328
+ },
2329
+ {
2330
+ "epoch": 2.887333187486327,
2331
+ "grad_norm": 1.2390645742416382,
2332
+ "learning_rate": 4.115637296174979e-08,
2333
+ "loss": 0.8995,
2334
+ "step": 3300
2335
+ },
2336
+ {
2337
+ "epoch": 2.8960840078757384,
2338
+ "grad_norm": 1.2656710147857666,
2339
+ "learning_rate": 3.4890159021608305e-08,
2340
+ "loss": 0.8741,
2341
+ "step": 3310
2342
+ },
2343
+ {
2344
+ "epoch": 2.90483482826515,
2345
+ "grad_norm": 1.149044156074524,
2346
+ "learning_rate": 2.9139503399996382e-08,
2347
+ "loss": 0.8869,
2348
+ "step": 3320
2349
+ },
2350
+ {
2351
+ "epoch": 2.9135856486545615,
2352
+ "grad_norm": 1.2886285781860352,
2353
+ "learning_rate": 2.3905003223349544e-08,
2354
+ "loss": 0.8875,
2355
+ "step": 3330
2356
+ },
2357
+ {
2358
+ "epoch": 2.922336469043973,
2359
+ "grad_norm": 1.2327951192855835,
2360
+ "learning_rate": 1.9187202022464822e-08,
2361
+ "loss": 0.8899,
2362
+ "step": 3340
2363
+ },
2364
+ {
2365
+ "epoch": 2.9310872894333846,
2366
+ "grad_norm": 1.3543472290039062,
2367
+ "learning_rate": 1.4986589676063147e-08,
2368
+ "loss": 0.8787,
2369
+ "step": 3350
2370
+ },
2371
+ {
2372
+ "epoch": 2.939838109822796,
2373
+ "grad_norm": 1.3077126741409302,
2374
+ "learning_rate": 1.1303602359919452e-08,
2375
+ "loss": 0.9173,
2376
+ "step": 3360
2377
+ },
2378
+ {
2379
+ "epoch": 2.9485889302122072,
2380
+ "grad_norm": 1.254578948020935,
2381
+ "learning_rate": 8.1386225015756e-09,
2382
+ "loss": 0.8375,
2383
+ "step": 3370
2384
+ },
2385
+ {
2386
+ "epoch": 2.957339750601619,
2387
+ "grad_norm": 1.3026719093322754,
2388
+ "learning_rate": 5.491978740627679e-09,
2389
+ "loss": 0.8827,
2390
+ "step": 3380
2391
+ },
2392
+ {
2393
+ "epoch": 2.9660905709910304,
2394
+ "grad_norm": 1.2489607334136963,
2395
+ "learning_rate": 3.3639458946027603e-09,
2396
+ "loss": 0.8835,
2397
+ "step": 3390
2398
+ },
2399
+ {
2400
+ "epoch": 2.974841391380442,
2401
+ "grad_norm": 1.264045000076294,
2402
+ "learning_rate": 1.7547449304222696e-09,
2403
+ "loss": 0.8905,
2404
+ "step": 3400
2405
+ },
2406
+ {
2407
+ "epoch": 2.9835922117698535,
2408
+ "grad_norm": 1.2398723363876343,
2409
+ "learning_rate": 6.645429414570137e-10,
2410
+ "loss": 0.8854,
2411
+ "step": 3410
2412
+ },
2413
+ {
2414
+ "epoch": 2.992343032159265,
2415
+ "grad_norm": 1.2165299654006958,
2416
+ "learning_rate": 9.345313017938839e-11,
2417
+ "loss": 0.8911,
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
+ }