aleandrananu commited on
Commit
5a4ba70
1 Parent(s): 7941a07

Delete trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +0 -196
trainer_state.json DELETED
@@ -1,196 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 0.9765625,
5
- "global_step": 12000,
6
- "is_hyper_param_search": false,
7
- "is_local_process_zero": true,
8
- "is_world_process_zero": true,
9
- "log_history": [
10
- {
11
- "epoch": 0.04,
12
- "learning_rate": 0.0005,
13
- "loss": 2.0116,
14
- "step": 500
15
- },
16
- {
17
- "epoch": 0.08,
18
- "learning_rate": 0.0005,
19
- "loss": 1.467,
20
- "step": 1000
21
- },
22
- {
23
- "epoch": 0.12,
24
- "learning_rate": 0.0005,
25
- "loss": 1.2612,
26
- "step": 1500
27
- },
28
- {
29
- "epoch": 0.16,
30
- "learning_rate": 0.0005,
31
- "loss": 1.1703,
32
- "step": 2000
33
- },
34
- {
35
- "epoch": 0.2,
36
- "learning_rate": 0.0005,
37
- "loss": 1.0993,
38
- "step": 2500
39
- },
40
- {
41
- "epoch": 0.2,
42
- "eval_bleu": 14.0784,
43
- "eval_gen_len": 14.7514,
44
- "eval_loss": 1.683070421218872,
45
- "eval_runtime": 937.2645,
46
- "eval_samples_per_second": 30.017,
47
- "step": 2500
48
- },
49
- {
50
- "epoch": 0.24,
51
- "learning_rate": 0.0005,
52
- "loss": 1.0415,
53
- "step": 3000
54
- },
55
- {
56
- "epoch": 0.28,
57
- "learning_rate": 0.0005,
58
- "loss": 0.9833,
59
- "step": 3500
60
- },
61
- {
62
- "epoch": 0.33,
63
- "learning_rate": 0.0005,
64
- "loss": 0.9562,
65
- "step": 4000
66
- },
67
- {
68
- "epoch": 0.37,
69
- "learning_rate": 0.0005,
70
- "loss": 0.9222,
71
- "step": 4500
72
- },
73
- {
74
- "epoch": 0.41,
75
- "learning_rate": 0.0005,
76
- "loss": 0.8916,
77
- "step": 5000
78
- },
79
- {
80
- "epoch": 0.41,
81
- "eval_bleu": 13.693,
82
- "eval_gen_len": 14.7668,
83
- "eval_loss": 1.786014199256897,
84
- "eval_runtime": 948.4426,
85
- "eval_samples_per_second": 29.663,
86
- "step": 5000
87
- },
88
- {
89
- "epoch": 0.45,
90
- "learning_rate": 0.0005,
91
- "loss": 0.857,
92
- "step": 5500
93
- },
94
- {
95
- "epoch": 0.49,
96
- "learning_rate": 0.0005,
97
- "loss": 0.8205,
98
- "step": 6000
99
- },
100
- {
101
- "epoch": 0.53,
102
- "learning_rate": 0.0005,
103
- "loss": 0.827,
104
- "step": 6500
105
- },
106
- {
107
- "epoch": 0.57,
108
- "learning_rate": 0.0005,
109
- "loss": 0.791,
110
- "step": 7000
111
- },
112
- {
113
- "epoch": 0.61,
114
- "learning_rate": 0.0005,
115
- "loss": 0.7599,
116
- "step": 7500
117
- },
118
- {
119
- "epoch": 0.61,
120
- "eval_bleu": 13.4015,
121
- "eval_gen_len": 14.7892,
122
- "eval_loss": 1.7776522636413574,
123
- "eval_runtime": 936.6248,
124
- "eval_samples_per_second": 30.038,
125
- "step": 7500
126
- },
127
- {
128
- "epoch": 0.65,
129
- "learning_rate": 0.0005,
130
- "loss": 0.7712,
131
- "step": 8000
132
- },
133
- {
134
- "epoch": 0.69,
135
- "learning_rate": 0.0005,
136
- "loss": 0.7531,
137
- "step": 8500
138
- },
139
- {
140
- "epoch": 0.73,
141
- "learning_rate": 0.0005,
142
- "loss": 0.7272,
143
- "step": 9000
144
- },
145
- {
146
- "epoch": 0.77,
147
- "learning_rate": 0.0005,
148
- "loss": 0.7285,
149
- "step": 9500
150
- },
151
- {
152
- "epoch": 0.81,
153
- "learning_rate": 0.0005,
154
- "loss": 0.7226,
155
- "step": 10000
156
- },
157
- {
158
- "epoch": 0.81,
159
- "eval_bleu": 13.1149,
160
- "eval_gen_len": 15.0103,
161
- "eval_loss": 1.895828366279602,
162
- "eval_runtime": 937.7408,
163
- "eval_samples_per_second": 30.002,
164
- "step": 10000
165
- },
166
- {
167
- "epoch": 0.85,
168
- "learning_rate": 0.0005,
169
- "loss": 0.7017,
170
- "step": 10500
171
- },
172
- {
173
- "epoch": 0.9,
174
- "learning_rate": 0.0005,
175
- "loss": 0.6853,
176
- "step": 11000
177
- },
178
- {
179
- "epoch": 0.94,
180
- "learning_rate": 0.0005,
181
- "loss": 0.6854,
182
- "step": 11500
183
- },
184
- {
185
- "epoch": 0.98,
186
- "learning_rate": 0.0005,
187
- "loss": 0.6591,
188
- "step": 12000
189
- }
190
- ],
191
- "max_steps": 12288,
192
- "num_train_epochs": 1,
193
- "total_flos": 1134507311529984.0,
194
- "trial_name": null,
195
- "trial_params": null
196
- }