hwm21 commited on
Commit
eb4d7b5
·
verified ·
1 Parent(s): da72132

Training in progress, step 12500, checkpoint

Browse files
checkpoint-12500/config.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/detr-resnet-50-dc5",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "relu",
5
+ "architectures": [
6
+ "DetrForObjectDetection"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "auxiliary_loss": false,
10
+ "backbone": "resnet50",
11
+ "backbone_config": null,
12
+ "backbone_kwargs": {
13
+ "in_chans": 3,
14
+ "out_indices": [
15
+ 1,
16
+ 2,
17
+ 3,
18
+ 4
19
+ ],
20
+ "output_stride": 16
21
+ },
22
+ "bbox_cost": 5,
23
+ "bbox_loss_coefficient": 5,
24
+ "class_cost": 1,
25
+ "classifier_dropout": 0.0,
26
+ "d_model": 256,
27
+ "decoder_attention_heads": 8,
28
+ "decoder_ffn_dim": 2048,
29
+ "decoder_layerdrop": 0.0,
30
+ "decoder_layers": 6,
31
+ "dice_loss_coefficient": 1,
32
+ "dilation": true,
33
+ "dropout": 0.1,
34
+ "encoder_attention_heads": 8,
35
+ "encoder_ffn_dim": 2048,
36
+ "encoder_layerdrop": 0.0,
37
+ "encoder_layers": 6,
38
+ "eos_coefficient": 0.1,
39
+ "giou_cost": 2,
40
+ "giou_loss_coefficient": 2,
41
+ "id2label": {
42
+ "0": "segment"
43
+ },
44
+ "init_std": 0.02,
45
+ "init_xavier_std": 1.0,
46
+ "is_encoder_decoder": true,
47
+ "label2id": {
48
+ "segment": 0
49
+ },
50
+ "mask_loss_coefficient": 1,
51
+ "max_position_embeddings": 1024,
52
+ "model_type": "detr",
53
+ "num_channels": 3,
54
+ "num_hidden_layers": 6,
55
+ "num_queries": 500,
56
+ "position_embedding_type": "sine",
57
+ "scale_embedding": false,
58
+ "torch_dtype": "float32",
59
+ "transformers_version": "4.44.2",
60
+ "use_pretrained_backbone": true,
61
+ "use_timm_backbone": true
62
+ }
checkpoint-12500/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:595b4256bc6b08a4dabf7abff0058bc5e1ee2387ebdbc5e2f883d6e228de4f7b
3
+ size 166904424
checkpoint-12500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca785a2351cde23732f9a0cc2441aae042a0296ac71254e41d44412c6c5fe3cd
3
+ size 331310970
checkpoint-12500/preprocessor_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_convert_annotations": true,
3
+ "do_normalize": true,
4
+ "do_pad": true,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "format": "coco_detection",
8
+ "image_mean": [
9
+ 0.485,
10
+ 0.456,
11
+ 0.406
12
+ ],
13
+ "image_processor_type": "DetrImageProcessor",
14
+ "image_std": [
15
+ 0.229,
16
+ 0.224,
17
+ 0.225
18
+ ],
19
+ "pad_size": null,
20
+ "resample": 2,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "longest_edge": 1333,
24
+ "shortest_edge": 800
25
+ }
26
+ }
checkpoint-12500/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bbe7c038bec3ddc2065570a911ae8723c49c70f04b28d405f7d32ebfedcb489b
3
+ size 14244
checkpoint-12500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7338ce6d7fb2152fa4b402cec8e839c6afb4c2d6f9920db5807d3e11e566c53
3
+ size 1064
checkpoint-12500/trainer_state.json ADDED
@@ -0,0 +1,1908 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 192.30769230769232,
5
+ "eval_steps": 100,
6
+ "global_step": 12500,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 1.5384615384615383,
13
+ "grad_norm": 47.11796188354492,
14
+ "learning_rate": 9.990900000000001e-06,
15
+ "loss": 3.6644,
16
+ "step": 100
17
+ },
18
+ {
19
+ "epoch": 1.5384615384615383,
20
+ "eval_loss": 2.4919605255126953,
21
+ "eval_runtime": 12.5517,
22
+ "eval_samples_per_second": 10.437,
23
+ "eval_steps_per_second": 1.354,
24
+ "step": 100
25
+ },
26
+ {
27
+ "epoch": 3.076923076923077,
28
+ "grad_norm": 31.28130531311035,
29
+ "learning_rate": 9.980900000000001e-06,
30
+ "loss": 2.2347,
31
+ "step": 200
32
+ },
33
+ {
34
+ "epoch": 3.076923076923077,
35
+ "eval_loss": 2.156316041946411,
36
+ "eval_runtime": 11.1792,
37
+ "eval_samples_per_second": 11.718,
38
+ "eval_steps_per_second": 1.521,
39
+ "step": 200
40
+ },
41
+ {
42
+ "epoch": 4.615384615384615,
43
+ "grad_norm": 39.616390228271484,
44
+ "learning_rate": 9.970900000000001e-06,
45
+ "loss": 2.0254,
46
+ "step": 300
47
+ },
48
+ {
49
+ "epoch": 4.615384615384615,
50
+ "eval_loss": 2.024153709411621,
51
+ "eval_runtime": 11.1934,
52
+ "eval_samples_per_second": 11.703,
53
+ "eval_steps_per_second": 1.519,
54
+ "step": 300
55
+ },
56
+ {
57
+ "epoch": 6.153846153846154,
58
+ "grad_norm": 28.285825729370117,
59
+ "learning_rate": 9.960900000000001e-06,
60
+ "loss": 1.9361,
61
+ "step": 400
62
+ },
63
+ {
64
+ "epoch": 6.153846153846154,
65
+ "eval_loss": 1.9094743728637695,
66
+ "eval_runtime": 11.3855,
67
+ "eval_samples_per_second": 11.506,
68
+ "eval_steps_per_second": 1.493,
69
+ "step": 400
70
+ },
71
+ {
72
+ "epoch": 7.6923076923076925,
73
+ "grad_norm": 34.14302062988281,
74
+ "learning_rate": 9.950900000000002e-06,
75
+ "loss": 1.8531,
76
+ "step": 500
77
+ },
78
+ {
79
+ "epoch": 7.6923076923076925,
80
+ "eval_loss": 1.8729331493377686,
81
+ "eval_runtime": 11.2935,
82
+ "eval_samples_per_second": 11.6,
83
+ "eval_steps_per_second": 1.505,
84
+ "step": 500
85
+ },
86
+ {
87
+ "epoch": 9.23076923076923,
88
+ "grad_norm": 39.09531784057617,
89
+ "learning_rate": 9.940900000000002e-06,
90
+ "loss": 1.7669,
91
+ "step": 600
92
+ },
93
+ {
94
+ "epoch": 9.23076923076923,
95
+ "eval_loss": 1.831756830215454,
96
+ "eval_runtime": 11.0535,
97
+ "eval_samples_per_second": 11.851,
98
+ "eval_steps_per_second": 1.538,
99
+ "step": 600
100
+ },
101
+ {
102
+ "epoch": 10.76923076923077,
103
+ "grad_norm": 93.24444580078125,
104
+ "learning_rate": 9.930900000000002e-06,
105
+ "loss": 1.7518,
106
+ "step": 700
107
+ },
108
+ {
109
+ "epoch": 10.76923076923077,
110
+ "eval_loss": 1.7832175493240356,
111
+ "eval_runtime": 11.1684,
112
+ "eval_samples_per_second": 11.729,
113
+ "eval_steps_per_second": 1.522,
114
+ "step": 700
115
+ },
116
+ {
117
+ "epoch": 12.307692307692308,
118
+ "grad_norm": 32.21013641357422,
119
+ "learning_rate": 9.920900000000002e-06,
120
+ "loss": 1.7149,
121
+ "step": 800
122
+ },
123
+ {
124
+ "epoch": 12.307692307692308,
125
+ "eval_loss": 1.7581098079681396,
126
+ "eval_runtime": 11.101,
127
+ "eval_samples_per_second": 11.801,
128
+ "eval_steps_per_second": 1.531,
129
+ "step": 800
130
+ },
131
+ {
132
+ "epoch": 13.846153846153847,
133
+ "grad_norm": 59.90657043457031,
134
+ "learning_rate": 9.9109e-06,
135
+ "loss": 1.6734,
136
+ "step": 900
137
+ },
138
+ {
139
+ "epoch": 13.846153846153847,
140
+ "eval_loss": 1.7163844108581543,
141
+ "eval_runtime": 11.1167,
142
+ "eval_samples_per_second": 11.784,
143
+ "eval_steps_per_second": 1.529,
144
+ "step": 900
145
+ },
146
+ {
147
+ "epoch": 15.384615384615385,
148
+ "grad_norm": 20.61592674255371,
149
+ "learning_rate": 9.9009e-06,
150
+ "loss": 1.6612,
151
+ "step": 1000
152
+ },
153
+ {
154
+ "epoch": 15.384615384615385,
155
+ "eval_loss": 1.6949567794799805,
156
+ "eval_runtime": 11.0663,
157
+ "eval_samples_per_second": 11.838,
158
+ "eval_steps_per_second": 1.536,
159
+ "step": 1000
160
+ },
161
+ {
162
+ "epoch": 16.923076923076923,
163
+ "grad_norm": 17.60099220275879,
164
+ "learning_rate": 9.8909e-06,
165
+ "loss": 1.6199,
166
+ "step": 1100
167
+ },
168
+ {
169
+ "epoch": 16.923076923076923,
170
+ "eval_loss": 1.6769332885742188,
171
+ "eval_runtime": 11.0531,
172
+ "eval_samples_per_second": 11.852,
173
+ "eval_steps_per_second": 1.538,
174
+ "step": 1100
175
+ },
176
+ {
177
+ "epoch": 18.46153846153846,
178
+ "grad_norm": 20.802692413330078,
179
+ "learning_rate": 9.8809e-06,
180
+ "loss": 1.6008,
181
+ "step": 1200
182
+ },
183
+ {
184
+ "epoch": 18.46153846153846,
185
+ "eval_loss": 1.6524990797042847,
186
+ "eval_runtime": 11.0831,
187
+ "eval_samples_per_second": 11.82,
188
+ "eval_steps_per_second": 1.534,
189
+ "step": 1200
190
+ },
191
+ {
192
+ "epoch": 20.0,
193
+ "grad_norm": 21.809823989868164,
194
+ "learning_rate": 9.8709e-06,
195
+ "loss": 1.5812,
196
+ "step": 1300
197
+ },
198
+ {
199
+ "epoch": 20.0,
200
+ "eval_loss": 1.6428295373916626,
201
+ "eval_runtime": 11.1093,
202
+ "eval_samples_per_second": 11.792,
203
+ "eval_steps_per_second": 1.53,
204
+ "step": 1300
205
+ },
206
+ {
207
+ "epoch": 21.53846153846154,
208
+ "grad_norm": 46.8908576965332,
209
+ "learning_rate": 9.8609e-06,
210
+ "loss": 1.5419,
211
+ "step": 1400
212
+ },
213
+ {
214
+ "epoch": 21.53846153846154,
215
+ "eval_loss": 1.6006404161453247,
216
+ "eval_runtime": 11.2393,
217
+ "eval_samples_per_second": 11.655,
218
+ "eval_steps_per_second": 1.513,
219
+ "step": 1400
220
+ },
221
+ {
222
+ "epoch": 23.076923076923077,
223
+ "grad_norm": 27.15238380432129,
224
+ "learning_rate": 9.8509e-06,
225
+ "loss": 1.5374,
226
+ "step": 1500
227
+ },
228
+ {
229
+ "epoch": 23.076923076923077,
230
+ "eval_loss": 1.5862094163894653,
231
+ "eval_runtime": 11.1815,
232
+ "eval_samples_per_second": 11.716,
233
+ "eval_steps_per_second": 1.52,
234
+ "step": 1500
235
+ },
236
+ {
237
+ "epoch": 24.615384615384617,
238
+ "grad_norm": 40.26778030395508,
239
+ "learning_rate": 9.840900000000001e-06,
240
+ "loss": 1.4923,
241
+ "step": 1600
242
+ },
243
+ {
244
+ "epoch": 24.615384615384617,
245
+ "eval_loss": 1.576373815536499,
246
+ "eval_runtime": 11.1215,
247
+ "eval_samples_per_second": 11.779,
248
+ "eval_steps_per_second": 1.529,
249
+ "step": 1600
250
+ },
251
+ {
252
+ "epoch": 26.153846153846153,
253
+ "grad_norm": 35.266971588134766,
254
+ "learning_rate": 9.830900000000001e-06,
255
+ "loss": 1.4989,
256
+ "step": 1700
257
+ },
258
+ {
259
+ "epoch": 26.153846153846153,
260
+ "eval_loss": 1.5671430826187134,
261
+ "eval_runtime": 11.1873,
262
+ "eval_samples_per_second": 11.71,
263
+ "eval_steps_per_second": 1.52,
264
+ "step": 1700
265
+ },
266
+ {
267
+ "epoch": 27.692307692307693,
268
+ "grad_norm": 26.813480377197266,
269
+ "learning_rate": 9.820900000000001e-06,
270
+ "loss": 1.4711,
271
+ "step": 1800
272
+ },
273
+ {
274
+ "epoch": 27.692307692307693,
275
+ "eval_loss": 1.522908329963684,
276
+ "eval_runtime": 11.2106,
277
+ "eval_samples_per_second": 11.685,
278
+ "eval_steps_per_second": 1.516,
279
+ "step": 1800
280
+ },
281
+ {
282
+ "epoch": 29.23076923076923,
283
+ "grad_norm": 24.576723098754883,
284
+ "learning_rate": 9.810900000000001e-06,
285
+ "loss": 1.4421,
286
+ "step": 1900
287
+ },
288
+ {
289
+ "epoch": 29.23076923076923,
290
+ "eval_loss": 1.5039104223251343,
291
+ "eval_runtime": 11.257,
292
+ "eval_samples_per_second": 11.637,
293
+ "eval_steps_per_second": 1.51,
294
+ "step": 1900
295
+ },
296
+ {
297
+ "epoch": 30.76923076923077,
298
+ "grad_norm": 28.480438232421875,
299
+ "learning_rate": 9.800900000000001e-06,
300
+ "loss": 1.4347,
301
+ "step": 2000
302
+ },
303
+ {
304
+ "epoch": 30.76923076923077,
305
+ "eval_loss": 1.5123459100723267,
306
+ "eval_runtime": 11.187,
307
+ "eval_samples_per_second": 11.71,
308
+ "eval_steps_per_second": 1.52,
309
+ "step": 2000
310
+ },
311
+ {
312
+ "epoch": 32.30769230769231,
313
+ "grad_norm": 56.582088470458984,
314
+ "learning_rate": 9.790900000000001e-06,
315
+ "loss": 1.4212,
316
+ "step": 2100
317
+ },
318
+ {
319
+ "epoch": 32.30769230769231,
320
+ "eval_loss": 1.481844425201416,
321
+ "eval_runtime": 11.2075,
322
+ "eval_samples_per_second": 11.689,
323
+ "eval_steps_per_second": 1.517,
324
+ "step": 2100
325
+ },
326
+ {
327
+ "epoch": 33.84615384615385,
328
+ "grad_norm": 38.5254020690918,
329
+ "learning_rate": 9.780900000000002e-06,
330
+ "loss": 1.3908,
331
+ "step": 2200
332
+ },
333
+ {
334
+ "epoch": 33.84615384615385,
335
+ "eval_loss": 1.4529048204421997,
336
+ "eval_runtime": 11.197,
337
+ "eval_samples_per_second": 11.7,
338
+ "eval_steps_per_second": 1.518,
339
+ "step": 2200
340
+ },
341
+ {
342
+ "epoch": 35.38461538461539,
343
+ "grad_norm": 44.74857711791992,
344
+ "learning_rate": 9.770900000000002e-06,
345
+ "loss": 1.3734,
346
+ "step": 2300
347
+ },
348
+ {
349
+ "epoch": 35.38461538461539,
350
+ "eval_loss": 1.4617427587509155,
351
+ "eval_runtime": 11.1235,
352
+ "eval_samples_per_second": 11.777,
353
+ "eval_steps_per_second": 1.528,
354
+ "step": 2300
355
+ },
356
+ {
357
+ "epoch": 36.92307692307692,
358
+ "grad_norm": 28.926782608032227,
359
+ "learning_rate": 9.760900000000002e-06,
360
+ "loss": 1.365,
361
+ "step": 2400
362
+ },
363
+ {
364
+ "epoch": 36.92307692307692,
365
+ "eval_loss": 1.4297789335250854,
366
+ "eval_runtime": 11.3007,
367
+ "eval_samples_per_second": 11.592,
368
+ "eval_steps_per_second": 1.504,
369
+ "step": 2400
370
+ },
371
+ {
372
+ "epoch": 38.46153846153846,
373
+ "grad_norm": 27.750213623046875,
374
+ "learning_rate": 9.7509e-06,
375
+ "loss": 1.3306,
376
+ "step": 2500
377
+ },
378
+ {
379
+ "epoch": 38.46153846153846,
380
+ "eval_loss": 1.4345914125442505,
381
+ "eval_runtime": 11.2795,
382
+ "eval_samples_per_second": 11.614,
383
+ "eval_steps_per_second": 1.507,
384
+ "step": 2500
385
+ },
386
+ {
387
+ "epoch": 40.0,
388
+ "grad_norm": 35.288352966308594,
389
+ "learning_rate": 9.7409e-06,
390
+ "loss": 1.3677,
391
+ "step": 2600
392
+ },
393
+ {
394
+ "epoch": 40.0,
395
+ "eval_loss": 1.447089433670044,
396
+ "eval_runtime": 11.1366,
397
+ "eval_samples_per_second": 11.763,
398
+ "eval_steps_per_second": 1.527,
399
+ "step": 2600
400
+ },
401
+ {
402
+ "epoch": 41.53846153846154,
403
+ "grad_norm": 65.49736022949219,
404
+ "learning_rate": 9.7309e-06,
405
+ "loss": 1.3453,
406
+ "step": 2700
407
+ },
408
+ {
409
+ "epoch": 41.53846153846154,
410
+ "eval_loss": 1.405300259590149,
411
+ "eval_runtime": 11.1549,
412
+ "eval_samples_per_second": 11.744,
413
+ "eval_steps_per_second": 1.524,
414
+ "step": 2700
415
+ },
416
+ {
417
+ "epoch": 43.07692307692308,
418
+ "grad_norm": 24.333518981933594,
419
+ "learning_rate": 9.7209e-06,
420
+ "loss": 1.3206,
421
+ "step": 2800
422
+ },
423
+ {
424
+ "epoch": 43.07692307692308,
425
+ "eval_loss": 1.4218717813491821,
426
+ "eval_runtime": 11.3113,
427
+ "eval_samples_per_second": 11.581,
428
+ "eval_steps_per_second": 1.503,
429
+ "step": 2800
430
+ },
431
+ {
432
+ "epoch": 44.61538461538461,
433
+ "grad_norm": 45.50777816772461,
434
+ "learning_rate": 9.7109e-06,
435
+ "loss": 1.3363,
436
+ "step": 2900
437
+ },
438
+ {
439
+ "epoch": 44.61538461538461,
440
+ "eval_loss": 1.4220006465911865,
441
+ "eval_runtime": 11.156,
442
+ "eval_samples_per_second": 11.743,
443
+ "eval_steps_per_second": 1.524,
444
+ "step": 2900
445
+ },
446
+ {
447
+ "epoch": 46.15384615384615,
448
+ "grad_norm": 25.898344039916992,
449
+ "learning_rate": 9.7009e-06,
450
+ "loss": 1.2995,
451
+ "step": 3000
452
+ },
453
+ {
454
+ "epoch": 46.15384615384615,
455
+ "eval_loss": 1.3942431211471558,
456
+ "eval_runtime": 11.2282,
457
+ "eval_samples_per_second": 11.667,
458
+ "eval_steps_per_second": 1.514,
459
+ "step": 3000
460
+ },
461
+ {
462
+ "epoch": 47.69230769230769,
463
+ "grad_norm": 56.0889892578125,
464
+ "learning_rate": 9.6909e-06,
465
+ "loss": 1.2994,
466
+ "step": 3100
467
+ },
468
+ {
469
+ "epoch": 47.69230769230769,
470
+ "eval_loss": 1.3970586061477661,
471
+ "eval_runtime": 11.1682,
472
+ "eval_samples_per_second": 11.73,
473
+ "eval_steps_per_second": 1.522,
474
+ "step": 3100
475
+ },
476
+ {
477
+ "epoch": 49.23076923076923,
478
+ "grad_norm": 58.10311508178711,
479
+ "learning_rate": 9.6809e-06,
480
+ "loss": 1.2761,
481
+ "step": 3200
482
+ },
483
+ {
484
+ "epoch": 49.23076923076923,
485
+ "eval_loss": 1.390371561050415,
486
+ "eval_runtime": 11.2406,
487
+ "eval_samples_per_second": 11.654,
488
+ "eval_steps_per_second": 1.512,
489
+ "step": 3200
490
+ },
491
+ {
492
+ "epoch": 50.76923076923077,
493
+ "grad_norm": 17.050870895385742,
494
+ "learning_rate": 9.670900000000001e-06,
495
+ "loss": 1.2712,
496
+ "step": 3300
497
+ },
498
+ {
499
+ "epoch": 50.76923076923077,
500
+ "eval_loss": 1.3936753273010254,
501
+ "eval_runtime": 11.2364,
502
+ "eval_samples_per_second": 11.659,
503
+ "eval_steps_per_second": 1.513,
504
+ "step": 3300
505
+ },
506
+ {
507
+ "epoch": 52.30769230769231,
508
+ "grad_norm": 232.21804809570312,
509
+ "learning_rate": 9.660900000000001e-06,
510
+ "loss": 1.262,
511
+ "step": 3400
512
+ },
513
+ {
514
+ "epoch": 52.30769230769231,
515
+ "eval_loss": 1.4037091732025146,
516
+ "eval_runtime": 11.2231,
517
+ "eval_samples_per_second": 11.672,
518
+ "eval_steps_per_second": 1.515,
519
+ "step": 3400
520
+ },
521
+ {
522
+ "epoch": 53.84615384615385,
523
+ "grad_norm": 35.11832046508789,
524
+ "learning_rate": 9.650900000000001e-06,
525
+ "loss": 1.2788,
526
+ "step": 3500
527
+ },
528
+ {
529
+ "epoch": 53.84615384615385,
530
+ "eval_loss": 1.3545280694961548,
531
+ "eval_runtime": 11.1609,
532
+ "eval_samples_per_second": 11.737,
533
+ "eval_steps_per_second": 1.523,
534
+ "step": 3500
535
+ },
536
+ {
537
+ "epoch": 55.38461538461539,
538
+ "grad_norm": 27.077022552490234,
539
+ "learning_rate": 9.640900000000001e-06,
540
+ "loss": 1.2711,
541
+ "step": 3600
542
+ },
543
+ {
544
+ "epoch": 55.38461538461539,
545
+ "eval_loss": 1.3528518676757812,
546
+ "eval_runtime": 11.2852,
547
+ "eval_samples_per_second": 11.608,
548
+ "eval_steps_per_second": 1.506,
549
+ "step": 3600
550
+ },
551
+ {
552
+ "epoch": 56.92307692307692,
553
+ "grad_norm": 46.97712326049805,
554
+ "learning_rate": 9.630900000000001e-06,
555
+ "loss": 1.2492,
556
+ "step": 3700
557
+ },
558
+ {
559
+ "epoch": 56.92307692307692,
560
+ "eval_loss": 1.3534098863601685,
561
+ "eval_runtime": 11.2368,
562
+ "eval_samples_per_second": 11.658,
563
+ "eval_steps_per_second": 1.513,
564
+ "step": 3700
565
+ },
566
+ {
567
+ "epoch": 58.46153846153846,
568
+ "grad_norm": 42.06857681274414,
569
+ "learning_rate": 9.620900000000001e-06,
570
+ "loss": 1.2506,
571
+ "step": 3800
572
+ },
573
+ {
574
+ "epoch": 58.46153846153846,
575
+ "eval_loss": 1.3613977432250977,
576
+ "eval_runtime": 11.2206,
577
+ "eval_samples_per_second": 11.675,
578
+ "eval_steps_per_second": 1.515,
579
+ "step": 3800
580
+ },
581
+ {
582
+ "epoch": 60.0,
583
+ "grad_norm": 19.298952102661133,
584
+ "learning_rate": 9.610900000000001e-06,
585
+ "loss": 1.2201,
586
+ "step": 3900
587
+ },
588
+ {
589
+ "epoch": 60.0,
590
+ "eval_loss": 1.3586474657058716,
591
+ "eval_runtime": 11.2045,
592
+ "eval_samples_per_second": 11.692,
593
+ "eval_steps_per_second": 1.517,
594
+ "step": 3900
595
+ },
596
+ {
597
+ "epoch": 61.53846153846154,
598
+ "grad_norm": 30.0198974609375,
599
+ "learning_rate": 9.600900000000002e-06,
600
+ "loss": 1.2086,
601
+ "step": 4000
602
+ },
603
+ {
604
+ "epoch": 61.53846153846154,
605
+ "eval_loss": 1.3304755687713623,
606
+ "eval_runtime": 11.193,
607
+ "eval_samples_per_second": 11.704,
608
+ "eval_steps_per_second": 1.519,
609
+ "step": 4000
610
+ },
611
+ {
612
+ "epoch": 63.07692307692308,
613
+ "grad_norm": 37.59902572631836,
614
+ "learning_rate": 9.5909e-06,
615
+ "loss": 1.2375,
616
+ "step": 4100
617
+ },
618
+ {
619
+ "epoch": 63.07692307692308,
620
+ "eval_loss": 1.331407904624939,
621
+ "eval_runtime": 10.6714,
622
+ "eval_samples_per_second": 12.276,
623
+ "eval_steps_per_second": 1.593,
624
+ "step": 4100
625
+ },
626
+ {
627
+ "epoch": 64.61538461538461,
628
+ "grad_norm": 36.82079315185547,
629
+ "learning_rate": 9.5809e-06,
630
+ "loss": 1.2148,
631
+ "step": 4200
632
+ },
633
+ {
634
+ "epoch": 64.61538461538461,
635
+ "eval_loss": 1.3441548347473145,
636
+ "eval_runtime": 10.7472,
637
+ "eval_samples_per_second": 12.189,
638
+ "eval_steps_per_second": 1.582,
639
+ "step": 4200
640
+ },
641
+ {
642
+ "epoch": 66.15384615384616,
643
+ "grad_norm": 30.974130630493164,
644
+ "learning_rate": 9.5709e-06,
645
+ "loss": 1.197,
646
+ "step": 4300
647
+ },
648
+ {
649
+ "epoch": 66.15384615384616,
650
+ "eval_loss": 1.34512197971344,
651
+ "eval_runtime": 10.7398,
652
+ "eval_samples_per_second": 12.198,
653
+ "eval_steps_per_second": 1.583,
654
+ "step": 4300
655
+ },
656
+ {
657
+ "epoch": 67.6923076923077,
658
+ "grad_norm": 20.45345115661621,
659
+ "learning_rate": 9.5609e-06,
660
+ "loss": 1.2361,
661
+ "step": 4400
662
+ },
663
+ {
664
+ "epoch": 67.6923076923077,
665
+ "eval_loss": 1.3371080160140991,
666
+ "eval_runtime": 10.654,
667
+ "eval_samples_per_second": 12.296,
668
+ "eval_steps_per_second": 1.596,
669
+ "step": 4400
670
+ },
671
+ {
672
+ "epoch": 69.23076923076923,
673
+ "grad_norm": 19.758630752563477,
674
+ "learning_rate": 9.5509e-06,
675
+ "loss": 1.2001,
676
+ "step": 4500
677
+ },
678
+ {
679
+ "epoch": 69.23076923076923,
680
+ "eval_loss": 1.3270760774612427,
681
+ "eval_runtime": 10.6928,
682
+ "eval_samples_per_second": 12.251,
683
+ "eval_steps_per_second": 1.59,
684
+ "step": 4500
685
+ },
686
+ {
687
+ "epoch": 70.76923076923077,
688
+ "grad_norm": 45.2899055480957,
689
+ "learning_rate": 9.5409e-06,
690
+ "loss": 1.192,
691
+ "step": 4600
692
+ },
693
+ {
694
+ "epoch": 70.76923076923077,
695
+ "eval_loss": 1.3184590339660645,
696
+ "eval_runtime": 10.7154,
697
+ "eval_samples_per_second": 12.225,
698
+ "eval_steps_per_second": 1.586,
699
+ "step": 4600
700
+ },
701
+ {
702
+ "epoch": 72.3076923076923,
703
+ "grad_norm": 45.60734939575195,
704
+ "learning_rate": 9.5309e-06,
705
+ "loss": 1.2081,
706
+ "step": 4700
707
+ },
708
+ {
709
+ "epoch": 72.3076923076923,
710
+ "eval_loss": 1.3107666969299316,
711
+ "eval_runtime": 10.7029,
712
+ "eval_samples_per_second": 12.24,
713
+ "eval_steps_per_second": 1.588,
714
+ "step": 4700
715
+ },
716
+ {
717
+ "epoch": 73.84615384615384,
718
+ "grad_norm": 26.859603881835938,
719
+ "learning_rate": 9.5209e-06,
720
+ "loss": 1.1729,
721
+ "step": 4800
722
+ },
723
+ {
724
+ "epoch": 73.84615384615384,
725
+ "eval_loss": 1.310062289237976,
726
+ "eval_runtime": 10.7544,
727
+ "eval_samples_per_second": 12.181,
728
+ "eval_steps_per_second": 1.581,
729
+ "step": 4800
730
+ },
731
+ {
732
+ "epoch": 75.38461538461539,
733
+ "grad_norm": 19.90914535522461,
734
+ "learning_rate": 9.5109e-06,
735
+ "loss": 1.1899,
736
+ "step": 4900
737
+ },
738
+ {
739
+ "epoch": 75.38461538461539,
740
+ "eval_loss": 1.3038017749786377,
741
+ "eval_runtime": 10.6532,
742
+ "eval_samples_per_second": 12.297,
743
+ "eval_steps_per_second": 1.596,
744
+ "step": 4900
745
+ },
746
+ {
747
+ "epoch": 76.92307692307692,
748
+ "grad_norm": 22.629934310913086,
749
+ "learning_rate": 9.5009e-06,
750
+ "loss": 1.1875,
751
+ "step": 5000
752
+ },
753
+ {
754
+ "epoch": 76.92307692307692,
755
+ "eval_loss": 1.2785409688949585,
756
+ "eval_runtime": 10.6388,
757
+ "eval_samples_per_second": 12.313,
758
+ "eval_steps_per_second": 1.598,
759
+ "step": 5000
760
+ },
761
+ {
762
+ "epoch": 78.46153846153847,
763
+ "grad_norm": 45.462059020996094,
764
+ "learning_rate": 9.490900000000001e-06,
765
+ "loss": 1.1717,
766
+ "step": 5100
767
+ },
768
+ {
769
+ "epoch": 78.46153846153847,
770
+ "eval_loss": 1.278078317642212,
771
+ "eval_runtime": 10.9666,
772
+ "eval_samples_per_second": 11.945,
773
+ "eval_steps_per_second": 1.55,
774
+ "step": 5100
775
+ },
776
+ {
777
+ "epoch": 80.0,
778
+ "grad_norm": 34.85255432128906,
779
+ "learning_rate": 9.480900000000001e-06,
780
+ "loss": 1.1657,
781
+ "step": 5200
782
+ },
783
+ {
784
+ "epoch": 80.0,
785
+ "eval_loss": 1.2711185216903687,
786
+ "eval_runtime": 11.0211,
787
+ "eval_samples_per_second": 11.886,
788
+ "eval_steps_per_second": 1.543,
789
+ "step": 5200
790
+ },
791
+ {
792
+ "epoch": 81.53846153846153,
793
+ "grad_norm": 19.078449249267578,
794
+ "learning_rate": 9.470900000000001e-06,
795
+ "loss": 1.1814,
796
+ "step": 5300
797
+ },
798
+ {
799
+ "epoch": 81.53846153846153,
800
+ "eval_loss": 1.2781996726989746,
801
+ "eval_runtime": 11.0663,
802
+ "eval_samples_per_second": 11.838,
803
+ "eval_steps_per_second": 1.536,
804
+ "step": 5300
805
+ },
806
+ {
807
+ "epoch": 83.07692307692308,
808
+ "grad_norm": 31.05898094177246,
809
+ "learning_rate": 9.460900000000001e-06,
810
+ "loss": 1.1452,
811
+ "step": 5400
812
+ },
813
+ {
814
+ "epoch": 83.07692307692308,
815
+ "eval_loss": 1.2848775386810303,
816
+ "eval_runtime": 11.01,
817
+ "eval_samples_per_second": 11.898,
818
+ "eval_steps_per_second": 1.544,
819
+ "step": 5400
820
+ },
821
+ {
822
+ "epoch": 84.61538461538461,
823
+ "grad_norm": 28.712461471557617,
824
+ "learning_rate": 9.450900000000001e-06,
825
+ "loss": 1.1465,
826
+ "step": 5500
827
+ },
828
+ {
829
+ "epoch": 84.61538461538461,
830
+ "eval_loss": 1.2928494215011597,
831
+ "eval_runtime": 10.9253,
832
+ "eval_samples_per_second": 11.991,
833
+ "eval_steps_per_second": 1.556,
834
+ "step": 5500
835
+ },
836
+ {
837
+ "epoch": 86.15384615384616,
838
+ "grad_norm": 19.871828079223633,
839
+ "learning_rate": 9.440900000000001e-06,
840
+ "loss": 1.1736,
841
+ "step": 5600
842
+ },
843
+ {
844
+ "epoch": 86.15384615384616,
845
+ "eval_loss": 1.2648124694824219,
846
+ "eval_runtime": 11.0314,
847
+ "eval_samples_per_second": 11.875,
848
+ "eval_steps_per_second": 1.541,
849
+ "step": 5600
850
+ },
851
+ {
852
+ "epoch": 87.6923076923077,
853
+ "grad_norm": 22.47665023803711,
854
+ "learning_rate": 9.4309e-06,
855
+ "loss": 1.1184,
856
+ "step": 5700
857
+ },
858
+ {
859
+ "epoch": 87.6923076923077,
860
+ "eval_loss": 1.2936598062515259,
861
+ "eval_runtime": 10.9322,
862
+ "eval_samples_per_second": 11.983,
863
+ "eval_steps_per_second": 1.555,
864
+ "step": 5700
865
+ },
866
+ {
867
+ "epoch": 89.23076923076923,
868
+ "grad_norm": 38.79877471923828,
869
+ "learning_rate": 9.4209e-06,
870
+ "loss": 1.1616,
871
+ "step": 5800
872
+ },
873
+ {
874
+ "epoch": 89.23076923076923,
875
+ "eval_loss": 1.2650004625320435,
876
+ "eval_runtime": 10.9434,
877
+ "eval_samples_per_second": 11.971,
878
+ "eval_steps_per_second": 1.553,
879
+ "step": 5800
880
+ },
881
+ {
882
+ "epoch": 90.76923076923077,
883
+ "grad_norm": 40.851097106933594,
884
+ "learning_rate": 9.4109e-06,
885
+ "loss": 1.1469,
886
+ "step": 5900
887
+ },
888
+ {
889
+ "epoch": 90.76923076923077,
890
+ "eval_loss": 1.252148151397705,
891
+ "eval_runtime": 10.9867,
892
+ "eval_samples_per_second": 11.923,
893
+ "eval_steps_per_second": 1.547,
894
+ "step": 5900
895
+ },
896
+ {
897
+ "epoch": 92.3076923076923,
898
+ "grad_norm": 21.35544204711914,
899
+ "learning_rate": 9.4009e-06,
900
+ "loss": 1.1489,
901
+ "step": 6000
902
+ },
903
+ {
904
+ "epoch": 92.3076923076923,
905
+ "eval_loss": 1.259344220161438,
906
+ "eval_runtime": 10.9649,
907
+ "eval_samples_per_second": 11.947,
908
+ "eval_steps_per_second": 1.55,
909
+ "step": 6000
910
+ },
911
+ {
912
+ "epoch": 93.84615384615384,
913
+ "grad_norm": 33.265132904052734,
914
+ "learning_rate": 9.3909e-06,
915
+ "loss": 1.1315,
916
+ "step": 6100
917
+ },
918
+ {
919
+ "epoch": 93.84615384615384,
920
+ "eval_loss": 1.252693772315979,
921
+ "eval_runtime": 10.9852,
922
+ "eval_samples_per_second": 11.925,
923
+ "eval_steps_per_second": 1.548,
924
+ "step": 6100
925
+ },
926
+ {
927
+ "epoch": 95.38461538461539,
928
+ "grad_norm": 23.43667221069336,
929
+ "learning_rate": 9.381e-06,
930
+ "loss": 1.119,
931
+ "step": 6200
932
+ },
933
+ {
934
+ "epoch": 95.38461538461539,
935
+ "eval_loss": 1.254772424697876,
936
+ "eval_runtime": 11.0937,
937
+ "eval_samples_per_second": 11.808,
938
+ "eval_steps_per_second": 1.532,
939
+ "step": 6200
940
+ },
941
+ {
942
+ "epoch": 96.92307692307692,
943
+ "grad_norm": 51.93602752685547,
944
+ "learning_rate": 9.371e-06,
945
+ "loss": 1.1333,
946
+ "step": 6300
947
+ },
948
+ {
949
+ "epoch": 96.92307692307692,
950
+ "eval_loss": 1.249408483505249,
951
+ "eval_runtime": 11.0388,
952
+ "eval_samples_per_second": 11.867,
953
+ "eval_steps_per_second": 1.54,
954
+ "step": 6300
955
+ },
956
+ {
957
+ "epoch": 98.46153846153847,
958
+ "grad_norm": 23.473421096801758,
959
+ "learning_rate": 9.361e-06,
960
+ "loss": 1.1164,
961
+ "step": 6400
962
+ },
963
+ {
964
+ "epoch": 98.46153846153847,
965
+ "eval_loss": 1.2438522577285767,
966
+ "eval_runtime": 11.1581,
967
+ "eval_samples_per_second": 11.74,
968
+ "eval_steps_per_second": 1.524,
969
+ "step": 6400
970
+ },
971
+ {
972
+ "epoch": 100.0,
973
+ "grad_norm": 24.228403091430664,
974
+ "learning_rate": 9.351e-06,
975
+ "loss": 1.1333,
976
+ "step": 6500
977
+ },
978
+ {
979
+ "epoch": 100.0,
980
+ "eval_loss": 1.248632788658142,
981
+ "eval_runtime": 10.9675,
982
+ "eval_samples_per_second": 11.944,
983
+ "eval_steps_per_second": 1.55,
984
+ "step": 6500
985
+ },
986
+ {
987
+ "epoch": 101.53846153846153,
988
+ "grad_norm": 18.29631996154785,
989
+ "learning_rate": 9.341000000000001e-06,
990
+ "loss": 1.1082,
991
+ "step": 6600
992
+ },
993
+ {
994
+ "epoch": 101.53846153846153,
995
+ "eval_loss": 1.2509865760803223,
996
+ "eval_runtime": 11.0693,
997
+ "eval_samples_per_second": 11.834,
998
+ "eval_steps_per_second": 1.536,
999
+ "step": 6600
1000
+ },
1001
+ {
1002
+ "epoch": 103.07692307692308,
1003
+ "grad_norm": 42.855491638183594,
1004
+ "learning_rate": 9.331000000000001e-06,
1005
+ "loss": 1.1178,
1006
+ "step": 6700
1007
+ },
1008
+ {
1009
+ "epoch": 103.07692307692308,
1010
+ "eval_loss": 1.2890292406082153,
1011
+ "eval_runtime": 11.0366,
1012
+ "eval_samples_per_second": 11.87,
1013
+ "eval_steps_per_second": 1.54,
1014
+ "step": 6700
1015
+ },
1016
+ {
1017
+ "epoch": 104.61538461538461,
1018
+ "grad_norm": 46.675655364990234,
1019
+ "learning_rate": 9.321000000000001e-06,
1020
+ "loss": 1.1106,
1021
+ "step": 6800
1022
+ },
1023
+ {
1024
+ "epoch": 104.61538461538461,
1025
+ "eval_loss": 1.2719863653182983,
1026
+ "eval_runtime": 11.1266,
1027
+ "eval_samples_per_second": 11.774,
1028
+ "eval_steps_per_second": 1.528,
1029
+ "step": 6800
1030
+ },
1031
+ {
1032
+ "epoch": 106.15384615384616,
1033
+ "grad_norm": 26.414846420288086,
1034
+ "learning_rate": 9.311000000000001e-06,
1035
+ "loss": 1.1216,
1036
+ "step": 6900
1037
+ },
1038
+ {
1039
+ "epoch": 106.15384615384616,
1040
+ "eval_loss": 1.2423394918441772,
1041
+ "eval_runtime": 11.0197,
1042
+ "eval_samples_per_second": 11.888,
1043
+ "eval_steps_per_second": 1.543,
1044
+ "step": 6900
1045
+ },
1046
+ {
1047
+ "epoch": 107.6923076923077,
1048
+ "grad_norm": 31.4022274017334,
1049
+ "learning_rate": 9.301000000000001e-06,
1050
+ "loss": 1.1052,
1051
+ "step": 7000
1052
+ },
1053
+ {
1054
+ "epoch": 107.6923076923077,
1055
+ "eval_loss": 1.2372961044311523,
1056
+ "eval_runtime": 11.1127,
1057
+ "eval_samples_per_second": 11.788,
1058
+ "eval_steps_per_second": 1.53,
1059
+ "step": 7000
1060
+ },
1061
+ {
1062
+ "epoch": 109.23076923076923,
1063
+ "grad_norm": 23.16703987121582,
1064
+ "learning_rate": 9.291000000000001e-06,
1065
+ "loss": 1.0911,
1066
+ "step": 7100
1067
+ },
1068
+ {
1069
+ "epoch": 109.23076923076923,
1070
+ "eval_loss": 1.2309863567352295,
1071
+ "eval_runtime": 10.9572,
1072
+ "eval_samples_per_second": 11.956,
1073
+ "eval_steps_per_second": 1.551,
1074
+ "step": 7100
1075
+ },
1076
+ {
1077
+ "epoch": 110.76923076923077,
1078
+ "grad_norm": 21.648773193359375,
1079
+ "learning_rate": 9.281000000000001e-06,
1080
+ "loss": 1.0956,
1081
+ "step": 7200
1082
+ },
1083
+ {
1084
+ "epoch": 110.76923076923077,
1085
+ "eval_loss": 1.2261079549789429,
1086
+ "eval_runtime": 10.9351,
1087
+ "eval_samples_per_second": 11.98,
1088
+ "eval_steps_per_second": 1.555,
1089
+ "step": 7200
1090
+ },
1091
+ {
1092
+ "epoch": 112.3076923076923,
1093
+ "grad_norm": 24.5791072845459,
1094
+ "learning_rate": 9.271000000000002e-06,
1095
+ "loss": 1.0751,
1096
+ "step": 7300
1097
+ },
1098
+ {
1099
+ "epoch": 112.3076923076923,
1100
+ "eval_loss": 1.2161471843719482,
1101
+ "eval_runtime": 11.1052,
1102
+ "eval_samples_per_second": 11.796,
1103
+ "eval_steps_per_second": 1.531,
1104
+ "step": 7300
1105
+ },
1106
+ {
1107
+ "epoch": 113.84615384615384,
1108
+ "grad_norm": 35.867801666259766,
1109
+ "learning_rate": 9.261000000000002e-06,
1110
+ "loss": 1.086,
1111
+ "step": 7400
1112
+ },
1113
+ {
1114
+ "epoch": 113.84615384615384,
1115
+ "eval_loss": 1.2092362642288208,
1116
+ "eval_runtime": 11.2048,
1117
+ "eval_samples_per_second": 11.691,
1118
+ "eval_steps_per_second": 1.517,
1119
+ "step": 7400
1120
+ },
1121
+ {
1122
+ "epoch": 115.38461538461539,
1123
+ "grad_norm": 67.91041564941406,
1124
+ "learning_rate": 9.251000000000002e-06,
1125
+ "loss": 1.092,
1126
+ "step": 7500
1127
+ },
1128
+ {
1129
+ "epoch": 115.38461538461539,
1130
+ "eval_loss": 1.241829514503479,
1131
+ "eval_runtime": 10.9937,
1132
+ "eval_samples_per_second": 11.916,
1133
+ "eval_steps_per_second": 1.546,
1134
+ "step": 7500
1135
+ },
1136
+ {
1137
+ "epoch": 116.92307692307692,
1138
+ "grad_norm": 128.73751831054688,
1139
+ "learning_rate": 9.241000000000002e-06,
1140
+ "loss": 1.0764,
1141
+ "step": 7600
1142
+ },
1143
+ {
1144
+ "epoch": 116.92307692307692,
1145
+ "eval_loss": 1.2462713718414307,
1146
+ "eval_runtime": 10.9625,
1147
+ "eval_samples_per_second": 11.95,
1148
+ "eval_steps_per_second": 1.551,
1149
+ "step": 7600
1150
+ },
1151
+ {
1152
+ "epoch": 118.46153846153847,
1153
+ "grad_norm": 86.5144271850586,
1154
+ "learning_rate": 9.231000000000002e-06,
1155
+ "loss": 1.0643,
1156
+ "step": 7700
1157
+ },
1158
+ {
1159
+ "epoch": 118.46153846153847,
1160
+ "eval_loss": 1.2187525033950806,
1161
+ "eval_runtime": 10.9492,
1162
+ "eval_samples_per_second": 11.964,
1163
+ "eval_steps_per_second": 1.553,
1164
+ "step": 7700
1165
+ },
1166
+ {
1167
+ "epoch": 120.0,
1168
+ "grad_norm": 19.4710750579834,
1169
+ "learning_rate": 9.221e-06,
1170
+ "loss": 1.0966,
1171
+ "step": 7800
1172
+ },
1173
+ {
1174
+ "epoch": 120.0,
1175
+ "eval_loss": 1.2282384634017944,
1176
+ "eval_runtime": 10.9078,
1177
+ "eval_samples_per_second": 12.01,
1178
+ "eval_steps_per_second": 1.559,
1179
+ "step": 7800
1180
+ },
1181
+ {
1182
+ "epoch": 121.53846153846153,
1183
+ "grad_norm": 37.673744201660156,
1184
+ "learning_rate": 9.211e-06,
1185
+ "loss": 1.0632,
1186
+ "step": 7900
1187
+ },
1188
+ {
1189
+ "epoch": 121.53846153846153,
1190
+ "eval_loss": 1.2206230163574219,
1191
+ "eval_runtime": 11.0018,
1192
+ "eval_samples_per_second": 11.907,
1193
+ "eval_steps_per_second": 1.545,
1194
+ "step": 7900
1195
+ },
1196
+ {
1197
+ "epoch": 123.07692307692308,
1198
+ "grad_norm": 25.10326385498047,
1199
+ "learning_rate": 9.201e-06,
1200
+ "loss": 1.0873,
1201
+ "step": 8000
1202
+ },
1203
+ {
1204
+ "epoch": 123.07692307692308,
1205
+ "eval_loss": 1.2137339115142822,
1206
+ "eval_runtime": 10.9232,
1207
+ "eval_samples_per_second": 11.993,
1208
+ "eval_steps_per_second": 1.556,
1209
+ "step": 8000
1210
+ },
1211
+ {
1212
+ "epoch": 124.61538461538461,
1213
+ "grad_norm": 32.02176284790039,
1214
+ "learning_rate": 9.191e-06,
1215
+ "loss": 1.0568,
1216
+ "step": 8100
1217
+ },
1218
+ {
1219
+ "epoch": 124.61538461538461,
1220
+ "eval_loss": 1.2065187692642212,
1221
+ "eval_runtime": 10.9614,
1222
+ "eval_samples_per_second": 11.951,
1223
+ "eval_steps_per_second": 1.551,
1224
+ "step": 8100
1225
+ },
1226
+ {
1227
+ "epoch": 126.15384615384616,
1228
+ "grad_norm": 19.97406005859375,
1229
+ "learning_rate": 9.181e-06,
1230
+ "loss": 1.065,
1231
+ "step": 8200
1232
+ },
1233
+ {
1234
+ "epoch": 126.15384615384616,
1235
+ "eval_loss": 1.2094841003417969,
1236
+ "eval_runtime": 11.0274,
1237
+ "eval_samples_per_second": 11.879,
1238
+ "eval_steps_per_second": 1.542,
1239
+ "step": 8200
1240
+ },
1241
+ {
1242
+ "epoch": 127.6923076923077,
1243
+ "grad_norm": 31.624399185180664,
1244
+ "learning_rate": 9.171e-06,
1245
+ "loss": 1.0805,
1246
+ "step": 8300
1247
+ },
1248
+ {
1249
+ "epoch": 127.6923076923077,
1250
+ "eval_loss": 1.2149733304977417,
1251
+ "eval_runtime": 11.2014,
1252
+ "eval_samples_per_second": 11.695,
1253
+ "eval_steps_per_second": 1.518,
1254
+ "step": 8300
1255
+ },
1256
+ {
1257
+ "epoch": 129.23076923076923,
1258
+ "grad_norm": 29.24848747253418,
1259
+ "learning_rate": 9.161000000000001e-06,
1260
+ "loss": 1.0463,
1261
+ "step": 8400
1262
+ },
1263
+ {
1264
+ "epoch": 129.23076923076923,
1265
+ "eval_loss": 1.2077099084854126,
1266
+ "eval_runtime": 10.9432,
1267
+ "eval_samples_per_second": 11.971,
1268
+ "eval_steps_per_second": 1.553,
1269
+ "step": 8400
1270
+ },
1271
+ {
1272
+ "epoch": 130.76923076923077,
1273
+ "grad_norm": 27.14787483215332,
1274
+ "learning_rate": 9.151000000000001e-06,
1275
+ "loss": 1.0607,
1276
+ "step": 8500
1277
+ },
1278
+ {
1279
+ "epoch": 130.76923076923077,
1280
+ "eval_loss": 1.2046644687652588,
1281
+ "eval_runtime": 11.001,
1282
+ "eval_samples_per_second": 11.908,
1283
+ "eval_steps_per_second": 1.545,
1284
+ "step": 8500
1285
+ },
1286
+ {
1287
+ "epoch": 132.30769230769232,
1288
+ "grad_norm": 32.416194915771484,
1289
+ "learning_rate": 9.141000000000001e-06,
1290
+ "loss": 1.0365,
1291
+ "step": 8600
1292
+ },
1293
+ {
1294
+ "epoch": 132.30769230769232,
1295
+ "eval_loss": 1.195080041885376,
1296
+ "eval_runtime": 11.0205,
1297
+ "eval_samples_per_second": 11.887,
1298
+ "eval_steps_per_second": 1.543,
1299
+ "step": 8600
1300
+ },
1301
+ {
1302
+ "epoch": 133.84615384615384,
1303
+ "grad_norm": 58.863582611083984,
1304
+ "learning_rate": 9.131000000000001e-06,
1305
+ "loss": 1.0564,
1306
+ "step": 8700
1307
+ },
1308
+ {
1309
+ "epoch": 133.84615384615384,
1310
+ "eval_loss": 1.1966980695724487,
1311
+ "eval_runtime": 10.9043,
1312
+ "eval_samples_per_second": 12.014,
1313
+ "eval_steps_per_second": 1.559,
1314
+ "step": 8700
1315
+ },
1316
+ {
1317
+ "epoch": 135.3846153846154,
1318
+ "grad_norm": 26.08232307434082,
1319
+ "learning_rate": 9.121000000000001e-06,
1320
+ "loss": 1.0507,
1321
+ "step": 8800
1322
+ },
1323
+ {
1324
+ "epoch": 135.3846153846154,
1325
+ "eval_loss": 1.201367735862732,
1326
+ "eval_runtime": 10.9714,
1327
+ "eval_samples_per_second": 11.94,
1328
+ "eval_steps_per_second": 1.549,
1329
+ "step": 8800
1330
+ },
1331
+ {
1332
+ "epoch": 136.92307692307693,
1333
+ "grad_norm": 59.368019104003906,
1334
+ "learning_rate": 9.111000000000001e-06,
1335
+ "loss": 1.0508,
1336
+ "step": 8900
1337
+ },
1338
+ {
1339
+ "epoch": 136.92307692307693,
1340
+ "eval_loss": 1.208795428276062,
1341
+ "eval_runtime": 11.0529,
1342
+ "eval_samples_per_second": 11.852,
1343
+ "eval_steps_per_second": 1.538,
1344
+ "step": 8900
1345
+ },
1346
+ {
1347
+ "epoch": 138.46153846153845,
1348
+ "grad_norm": 43.07460021972656,
1349
+ "learning_rate": 9.101000000000001e-06,
1350
+ "loss": 1.0359,
1351
+ "step": 9000
1352
+ },
1353
+ {
1354
+ "epoch": 138.46153846153845,
1355
+ "eval_loss": 1.1782392263412476,
1356
+ "eval_runtime": 11.1519,
1357
+ "eval_samples_per_second": 11.747,
1358
+ "eval_steps_per_second": 1.524,
1359
+ "step": 9000
1360
+ },
1361
+ {
1362
+ "epoch": 140.0,
1363
+ "grad_norm": 22.39567756652832,
1364
+ "learning_rate": 9.091000000000002e-06,
1365
+ "loss": 1.0584,
1366
+ "step": 9100
1367
+ },
1368
+ {
1369
+ "epoch": 140.0,
1370
+ "eval_loss": 1.1873483657836914,
1371
+ "eval_runtime": 10.8732,
1372
+ "eval_samples_per_second": 12.048,
1373
+ "eval_steps_per_second": 1.563,
1374
+ "step": 9100
1375
+ },
1376
+ {
1377
+ "epoch": 141.53846153846155,
1378
+ "grad_norm": 22.813621520996094,
1379
+ "learning_rate": 9.081000000000002e-06,
1380
+ "loss": 1.0354,
1381
+ "step": 9200
1382
+ },
1383
+ {
1384
+ "epoch": 141.53846153846155,
1385
+ "eval_loss": 1.1731117963790894,
1386
+ "eval_runtime": 11.0847,
1387
+ "eval_samples_per_second": 11.818,
1388
+ "eval_steps_per_second": 1.534,
1389
+ "step": 9200
1390
+ },
1391
+ {
1392
+ "epoch": 143.07692307692307,
1393
+ "grad_norm": 31.44073486328125,
1394
+ "learning_rate": 9.071000000000002e-06,
1395
+ "loss": 1.0457,
1396
+ "step": 9300
1397
+ },
1398
+ {
1399
+ "epoch": 143.07692307692307,
1400
+ "eval_loss": 1.1962807178497314,
1401
+ "eval_runtime": 11.0668,
1402
+ "eval_samples_per_second": 11.837,
1403
+ "eval_steps_per_second": 1.536,
1404
+ "step": 9300
1405
+ },
1406
+ {
1407
+ "epoch": 144.6153846153846,
1408
+ "grad_norm": 18.18711280822754,
1409
+ "learning_rate": 9.061e-06,
1410
+ "loss": 1.0481,
1411
+ "step": 9400
1412
+ },
1413
+ {
1414
+ "epoch": 144.6153846153846,
1415
+ "eval_loss": 1.185339093208313,
1416
+ "eval_runtime": 10.9916,
1417
+ "eval_samples_per_second": 11.918,
1418
+ "eval_steps_per_second": 1.547,
1419
+ "step": 9400
1420
+ },
1421
+ {
1422
+ "epoch": 146.15384615384616,
1423
+ "grad_norm": 38.05665969848633,
1424
+ "learning_rate": 9.051e-06,
1425
+ "loss": 1.0391,
1426
+ "step": 9500
1427
+ },
1428
+ {
1429
+ "epoch": 146.15384615384616,
1430
+ "eval_loss": 1.1856777667999268,
1431
+ "eval_runtime": 11.0391,
1432
+ "eval_samples_per_second": 11.867,
1433
+ "eval_steps_per_second": 1.54,
1434
+ "step": 9500
1435
+ },
1436
+ {
1437
+ "epoch": 147.69230769230768,
1438
+ "grad_norm": 19.963260650634766,
1439
+ "learning_rate": 9.041e-06,
1440
+ "loss": 1.0322,
1441
+ "step": 9600
1442
+ },
1443
+ {
1444
+ "epoch": 147.69230769230768,
1445
+ "eval_loss": 1.1843148469924927,
1446
+ "eval_runtime": 11.0382,
1447
+ "eval_samples_per_second": 11.868,
1448
+ "eval_steps_per_second": 1.54,
1449
+ "step": 9600
1450
+ },
1451
+ {
1452
+ "epoch": 149.23076923076923,
1453
+ "grad_norm": 28.58108901977539,
1454
+ "learning_rate": 9.031e-06,
1455
+ "loss": 1.0369,
1456
+ "step": 9700
1457
+ },
1458
+ {
1459
+ "epoch": 149.23076923076923,
1460
+ "eval_loss": 1.1637182235717773,
1461
+ "eval_runtime": 11.2766,
1462
+ "eval_samples_per_second": 11.617,
1463
+ "eval_steps_per_second": 1.508,
1464
+ "step": 9700
1465
+ },
1466
+ {
1467
+ "epoch": 150.76923076923077,
1468
+ "grad_norm": 64.3956527709961,
1469
+ "learning_rate": 9.021e-06,
1470
+ "loss": 1.0519,
1471
+ "step": 9800
1472
+ },
1473
+ {
1474
+ "epoch": 150.76923076923077,
1475
+ "eval_loss": 1.1796802282333374,
1476
+ "eval_runtime": 11.3418,
1477
+ "eval_samples_per_second": 11.55,
1478
+ "eval_steps_per_second": 1.499,
1479
+ "step": 9800
1480
+ },
1481
+ {
1482
+ "epoch": 152.30769230769232,
1483
+ "grad_norm": 18.857580184936523,
1484
+ "learning_rate": 9.011e-06,
1485
+ "loss": 1.0272,
1486
+ "step": 9900
1487
+ },
1488
+ {
1489
+ "epoch": 152.30769230769232,
1490
+ "eval_loss": 1.1634279489517212,
1491
+ "eval_runtime": 11.6749,
1492
+ "eval_samples_per_second": 11.221,
1493
+ "eval_steps_per_second": 1.456,
1494
+ "step": 9900
1495
+ },
1496
+ {
1497
+ "epoch": 153.84615384615384,
1498
+ "grad_norm": 42.541683197021484,
1499
+ "learning_rate": 9.001e-06,
1500
+ "loss": 1.0287,
1501
+ "step": 10000
1502
+ },
1503
+ {
1504
+ "epoch": 153.84615384615384,
1505
+ "eval_loss": 1.1698088645935059,
1506
+ "eval_runtime": 11.068,
1507
+ "eval_samples_per_second": 11.836,
1508
+ "eval_steps_per_second": 1.536,
1509
+ "step": 10000
1510
+ },
1511
+ {
1512
+ "epoch": 155.3846153846154,
1513
+ "grad_norm": 30.52286720275879,
1514
+ "learning_rate": 8.991e-06,
1515
+ "loss": 1.0237,
1516
+ "step": 10100
1517
+ },
1518
+ {
1519
+ "epoch": 155.3846153846154,
1520
+ "eval_loss": 1.178311824798584,
1521
+ "eval_runtime": 11.1107,
1522
+ "eval_samples_per_second": 11.79,
1523
+ "eval_steps_per_second": 1.53,
1524
+ "step": 10100
1525
+ },
1526
+ {
1527
+ "epoch": 156.92307692307693,
1528
+ "grad_norm": 32.60612487792969,
1529
+ "learning_rate": 8.981000000000001e-06,
1530
+ "loss": 1.0362,
1531
+ "step": 10200
1532
+ },
1533
+ {
1534
+ "epoch": 156.92307692307693,
1535
+ "eval_loss": 1.157893180847168,
1536
+ "eval_runtime": 11.0484,
1537
+ "eval_samples_per_second": 11.857,
1538
+ "eval_steps_per_second": 1.539,
1539
+ "step": 10200
1540
+ },
1541
+ {
1542
+ "epoch": 158.46153846153845,
1543
+ "grad_norm": 26.15647315979004,
1544
+ "learning_rate": 8.971000000000001e-06,
1545
+ "loss": 0.998,
1546
+ "step": 10300
1547
+ },
1548
+ {
1549
+ "epoch": 158.46153846153845,
1550
+ "eval_loss": 1.1724605560302734,
1551
+ "eval_runtime": 11.2323,
1552
+ "eval_samples_per_second": 11.663,
1553
+ "eval_steps_per_second": 1.513,
1554
+ "step": 10300
1555
+ },
1556
+ {
1557
+ "epoch": 160.0,
1558
+ "grad_norm": 45.76993942260742,
1559
+ "learning_rate": 8.961000000000001e-06,
1560
+ "loss": 1.0432,
1561
+ "step": 10400
1562
+ },
1563
+ {
1564
+ "epoch": 160.0,
1565
+ "eval_loss": 1.1688075065612793,
1566
+ "eval_runtime": 11.0984,
1567
+ "eval_samples_per_second": 11.803,
1568
+ "eval_steps_per_second": 1.532,
1569
+ "step": 10400
1570
+ },
1571
+ {
1572
+ "epoch": 161.53846153846155,
1573
+ "grad_norm": 40.384578704833984,
1574
+ "learning_rate": 8.951000000000001e-06,
1575
+ "loss": 1.0421,
1576
+ "step": 10500
1577
+ },
1578
+ {
1579
+ "epoch": 161.53846153846155,
1580
+ "eval_loss": 1.1621845960617065,
1581
+ "eval_runtime": 11.1624,
1582
+ "eval_samples_per_second": 11.736,
1583
+ "eval_steps_per_second": 1.523,
1584
+ "step": 10500
1585
+ },
1586
+ {
1587
+ "epoch": 163.07692307692307,
1588
+ "grad_norm": 22.886058807373047,
1589
+ "learning_rate": 8.941000000000001e-06,
1590
+ "loss": 0.9912,
1591
+ "step": 10600
1592
+ },
1593
+ {
1594
+ "epoch": 163.07692307692307,
1595
+ "eval_loss": 1.1695055961608887,
1596
+ "eval_runtime": 10.9842,
1597
+ "eval_samples_per_second": 11.926,
1598
+ "eval_steps_per_second": 1.548,
1599
+ "step": 10600
1600
+ },
1601
+ {
1602
+ "epoch": 164.6153846153846,
1603
+ "grad_norm": 26.940736770629883,
1604
+ "learning_rate": 8.931000000000001e-06,
1605
+ "loss": 1.011,
1606
+ "step": 10700
1607
+ },
1608
+ {
1609
+ "epoch": 164.6153846153846,
1610
+ "eval_loss": 1.1458157300949097,
1611
+ "eval_runtime": 11.025,
1612
+ "eval_samples_per_second": 11.882,
1613
+ "eval_steps_per_second": 1.542,
1614
+ "step": 10700
1615
+ },
1616
+ {
1617
+ "epoch": 166.15384615384616,
1618
+ "grad_norm": 26.5013484954834,
1619
+ "learning_rate": 8.921000000000001e-06,
1620
+ "loss": 0.9876,
1621
+ "step": 10800
1622
+ },
1623
+ {
1624
+ "epoch": 166.15384615384616,
1625
+ "eval_loss": 1.1569631099700928,
1626
+ "eval_runtime": 11.101,
1627
+ "eval_samples_per_second": 11.801,
1628
+ "eval_steps_per_second": 1.531,
1629
+ "step": 10800
1630
+ },
1631
+ {
1632
+ "epoch": 167.69230769230768,
1633
+ "grad_norm": 29.859987258911133,
1634
+ "learning_rate": 8.911000000000002e-06,
1635
+ "loss": 1.0374,
1636
+ "step": 10900
1637
+ },
1638
+ {
1639
+ "epoch": 167.69230769230768,
1640
+ "eval_loss": 1.149316668510437,
1641
+ "eval_runtime": 11.1242,
1642
+ "eval_samples_per_second": 11.776,
1643
+ "eval_steps_per_second": 1.528,
1644
+ "step": 10900
1645
+ },
1646
+ {
1647
+ "epoch": 169.23076923076923,
1648
+ "grad_norm": 27.777738571166992,
1649
+ "learning_rate": 8.901e-06,
1650
+ "loss": 0.985,
1651
+ "step": 11000
1652
+ },
1653
+ {
1654
+ "epoch": 169.23076923076923,
1655
+ "eval_loss": 1.1608328819274902,
1656
+ "eval_runtime": 11.2215,
1657
+ "eval_samples_per_second": 11.674,
1658
+ "eval_steps_per_second": 1.515,
1659
+ "step": 11000
1660
+ },
1661
+ {
1662
+ "epoch": 170.76923076923077,
1663
+ "grad_norm": 39.21344757080078,
1664
+ "learning_rate": 8.891e-06,
1665
+ "loss": 1.0049,
1666
+ "step": 11100
1667
+ },
1668
+ {
1669
+ "epoch": 170.76923076923077,
1670
+ "eval_loss": 1.1642228364944458,
1671
+ "eval_runtime": 11.1947,
1672
+ "eval_samples_per_second": 11.702,
1673
+ "eval_steps_per_second": 1.519,
1674
+ "step": 11100
1675
+ },
1676
+ {
1677
+ "epoch": 172.30769230769232,
1678
+ "grad_norm": 29.880149841308594,
1679
+ "learning_rate": 8.881e-06,
1680
+ "loss": 0.9843,
1681
+ "step": 11200
1682
+ },
1683
+ {
1684
+ "epoch": 172.30769230769232,
1685
+ "eval_loss": 1.1574000120162964,
1686
+ "eval_runtime": 11.2634,
1687
+ "eval_samples_per_second": 11.631,
1688
+ "eval_steps_per_second": 1.509,
1689
+ "step": 11200
1690
+ },
1691
+ {
1692
+ "epoch": 173.84615384615384,
1693
+ "grad_norm": 63.53031539916992,
1694
+ "learning_rate": 8.871e-06,
1695
+ "loss": 1.0354,
1696
+ "step": 11300
1697
+ },
1698
+ {
1699
+ "epoch": 173.84615384615384,
1700
+ "eval_loss": 1.1575734615325928,
1701
+ "eval_runtime": 11.0265,
1702
+ "eval_samples_per_second": 11.88,
1703
+ "eval_steps_per_second": 1.542,
1704
+ "step": 11300
1705
+ },
1706
+ {
1707
+ "epoch": 175.3846153846154,
1708
+ "grad_norm": 26.937786102294922,
1709
+ "learning_rate": 8.861e-06,
1710
+ "loss": 0.9964,
1711
+ "step": 11400
1712
+ },
1713
+ {
1714
+ "epoch": 175.3846153846154,
1715
+ "eval_loss": 1.1552445888519287,
1716
+ "eval_runtime": 11.0325,
1717
+ "eval_samples_per_second": 11.874,
1718
+ "eval_steps_per_second": 1.541,
1719
+ "step": 11400
1720
+ },
1721
+ {
1722
+ "epoch": 176.92307692307693,
1723
+ "grad_norm": 87.28536987304688,
1724
+ "learning_rate": 8.851e-06,
1725
+ "loss": 0.9932,
1726
+ "step": 11500
1727
+ },
1728
+ {
1729
+ "epoch": 176.92307692307693,
1730
+ "eval_loss": 1.1411677598953247,
1731
+ "eval_runtime": 11.1527,
1732
+ "eval_samples_per_second": 11.746,
1733
+ "eval_steps_per_second": 1.524,
1734
+ "step": 11500
1735
+ },
1736
+ {
1737
+ "epoch": 178.46153846153845,
1738
+ "grad_norm": 25.903568267822266,
1739
+ "learning_rate": 8.841e-06,
1740
+ "loss": 0.9768,
1741
+ "step": 11600
1742
+ },
1743
+ {
1744
+ "epoch": 178.46153846153845,
1745
+ "eval_loss": 1.1635726690292358,
1746
+ "eval_runtime": 11.123,
1747
+ "eval_samples_per_second": 11.777,
1748
+ "eval_steps_per_second": 1.528,
1749
+ "step": 11600
1750
+ },
1751
+ {
1752
+ "epoch": 180.0,
1753
+ "grad_norm": 24.315654754638672,
1754
+ "learning_rate": 8.831e-06,
1755
+ "loss": 0.9984,
1756
+ "step": 11700
1757
+ },
1758
+ {
1759
+ "epoch": 180.0,
1760
+ "eval_loss": 1.1710366010665894,
1761
+ "eval_runtime": 11.1375,
1762
+ "eval_samples_per_second": 11.762,
1763
+ "eval_steps_per_second": 1.526,
1764
+ "step": 11700
1765
+ },
1766
+ {
1767
+ "epoch": 181.53846153846155,
1768
+ "grad_norm": 60.17182540893555,
1769
+ "learning_rate": 8.821e-06,
1770
+ "loss": 0.9703,
1771
+ "step": 11800
1772
+ },
1773
+ {
1774
+ "epoch": 181.53846153846155,
1775
+ "eval_loss": 1.1556936502456665,
1776
+ "eval_runtime": 11.0387,
1777
+ "eval_samples_per_second": 11.867,
1778
+ "eval_steps_per_second": 1.54,
1779
+ "step": 11800
1780
+ },
1781
+ {
1782
+ "epoch": 183.07692307692307,
1783
+ "grad_norm": 32.373477935791016,
1784
+ "learning_rate": 8.811000000000001e-06,
1785
+ "loss": 0.9996,
1786
+ "step": 11900
1787
+ },
1788
+ {
1789
+ "epoch": 183.07692307692307,
1790
+ "eval_loss": 1.146790623664856,
1791
+ "eval_runtime": 11.1805,
1792
+ "eval_samples_per_second": 11.717,
1793
+ "eval_steps_per_second": 1.521,
1794
+ "step": 11900
1795
+ },
1796
+ {
1797
+ "epoch": 184.6153846153846,
1798
+ "grad_norm": 42.578575134277344,
1799
+ "learning_rate": 8.801000000000001e-06,
1800
+ "loss": 0.9795,
1801
+ "step": 12000
1802
+ },
1803
+ {
1804
+ "epoch": 184.6153846153846,
1805
+ "eval_loss": 1.144544005393982,
1806
+ "eval_runtime": 11.0704,
1807
+ "eval_samples_per_second": 11.833,
1808
+ "eval_steps_per_second": 1.536,
1809
+ "step": 12000
1810
+ },
1811
+ {
1812
+ "epoch": 186.15384615384616,
1813
+ "grad_norm": 22.79789161682129,
1814
+ "learning_rate": 8.791000000000001e-06,
1815
+ "loss": 0.9905,
1816
+ "step": 12100
1817
+ },
1818
+ {
1819
+ "epoch": 186.15384615384616,
1820
+ "eval_loss": 1.1581685543060303,
1821
+ "eval_runtime": 11.272,
1822
+ "eval_samples_per_second": 11.622,
1823
+ "eval_steps_per_second": 1.508,
1824
+ "step": 12100
1825
+ },
1826
+ {
1827
+ "epoch": 187.69230769230768,
1828
+ "grad_norm": 52.061012268066406,
1829
+ "learning_rate": 8.781200000000002e-06,
1830
+ "loss": 0.9817,
1831
+ "step": 12200
1832
+ },
1833
+ {
1834
+ "epoch": 187.69230769230768,
1835
+ "eval_loss": 1.159809947013855,
1836
+ "eval_runtime": 11.2021,
1837
+ "eval_samples_per_second": 11.694,
1838
+ "eval_steps_per_second": 1.518,
1839
+ "step": 12200
1840
+ },
1841
+ {
1842
+ "epoch": 189.23076923076923,
1843
+ "grad_norm": 35.95882034301758,
1844
+ "learning_rate": 8.7712e-06,
1845
+ "loss": 1.0071,
1846
+ "step": 12300
1847
+ },
1848
+ {
1849
+ "epoch": 189.23076923076923,
1850
+ "eval_loss": 1.1944890022277832,
1851
+ "eval_runtime": 10.9263,
1852
+ "eval_samples_per_second": 11.989,
1853
+ "eval_steps_per_second": 1.556,
1854
+ "step": 12300
1855
+ },
1856
+ {
1857
+ "epoch": 190.76923076923077,
1858
+ "grad_norm": 213.48587036132812,
1859
+ "learning_rate": 8.7612e-06,
1860
+ "loss": 0.9997,
1861
+ "step": 12400
1862
+ },
1863
+ {
1864
+ "epoch": 190.76923076923077,
1865
+ "eval_loss": 1.191455602645874,
1866
+ "eval_runtime": 10.917,
1867
+ "eval_samples_per_second": 12.0,
1868
+ "eval_steps_per_second": 1.557,
1869
+ "step": 12400
1870
+ },
1871
+ {
1872
+ "epoch": 192.30769230769232,
1873
+ "grad_norm": 19.97510528564453,
1874
+ "learning_rate": 8.7512e-06,
1875
+ "loss": 1.001,
1876
+ "step": 12500
1877
+ },
1878
+ {
1879
+ "epoch": 192.30769230769232,
1880
+ "eval_loss": 1.167776346206665,
1881
+ "eval_runtime": 11.033,
1882
+ "eval_samples_per_second": 11.873,
1883
+ "eval_steps_per_second": 1.541,
1884
+ "step": 12500
1885
+ }
1886
+ ],
1887
+ "logging_steps": 100,
1888
+ "max_steps": 100000,
1889
+ "num_input_tokens_seen": 0,
1890
+ "num_train_epochs": 1539,
1891
+ "save_steps": 500,
1892
+ "stateful_callbacks": {
1893
+ "TrainerControl": {
1894
+ "args": {
1895
+ "should_epoch_stop": false,
1896
+ "should_evaluate": false,
1897
+ "should_log": false,
1898
+ "should_save": true,
1899
+ "should_training_stop": false
1900
+ },
1901
+ "attributes": {}
1902
+ }
1903
+ },
1904
+ "total_flos": 4.795622763245568e+19,
1905
+ "train_batch_size": 8,
1906
+ "trial_name": null,
1907
+ "trial_params": null
1908
+ }
checkpoint-12500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2ad9d755c74ae76083891287d009fdddb96b730c67f38d8947352435e2637ed7
3
+ size 5304