Upload trainer_state.json
Browse files- trainer_state.json +216 -0
trainer_state.json
ADDED
@@ -0,0 +1,216 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 82.213,
|
3 |
+
"best_model_checkpoint": "/uoa/home/s02sd1/Desktop/Project data/pytorch_project/pegasus_26.04/checkpoint-5000",
|
4 |
+
"epoch": 95.23809523809524,
|
5 |
+
"global_step": 8000,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 5.95,
|
12 |
+
"learning_rate": 2.8214285714285713e-06,
|
13 |
+
"loss": 0.1312,
|
14 |
+
"step": 500
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 11.9,
|
18 |
+
"learning_rate": 2.642857142857143e-06,
|
19 |
+
"loss": 0.046,
|
20 |
+
"step": 1000
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 11.9,
|
24 |
+
"eval_gen_len": 189.9754,
|
25 |
+
"eval_loss": 0.0635865107178688,
|
26 |
+
"eval_rouge1": 81.7351,
|
27 |
+
"eval_rouge2": 74.9502,
|
28 |
+
"eval_rougeL": 65.0568,
|
29 |
+
"eval_rougeLsum": 65.0619,
|
30 |
+
"eval_runtime": 16215.7902,
|
31 |
+
"eval_samples_per_second": 0.013,
|
32 |
+
"eval_steps_per_second": 0.001,
|
33 |
+
"step": 1000
|
34 |
+
},
|
35 |
+
{
|
36 |
+
"epoch": 17.86,
|
37 |
+
"learning_rate": 2.4642857142857143e-06,
|
38 |
+
"loss": 0.0227,
|
39 |
+
"step": 1500
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 23.81,
|
43 |
+
"learning_rate": 2.2857142857142856e-06,
|
44 |
+
"loss": 0.0126,
|
45 |
+
"step": 2000
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 23.81,
|
49 |
+
"eval_gen_len": 208.1773,
|
50 |
+
"eval_loss": 0.061119239777326584,
|
51 |
+
"eval_rouge1": 82.8965,
|
52 |
+
"eval_rouge2": 76.972,
|
53 |
+
"eval_rougeL": 70.043,
|
54 |
+
"eval_rougeLsum": 70.3743,
|
55 |
+
"eval_runtime": 2572.7511,
|
56 |
+
"eval_samples_per_second": 0.079,
|
57 |
+
"eval_steps_per_second": 0.004,
|
58 |
+
"step": 2000
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 29.76,
|
62 |
+
"learning_rate": 2.1071428571428572e-06,
|
63 |
+
"loss": 0.0072,
|
64 |
+
"step": 2500
|
65 |
+
},
|
66 |
+
{
|
67 |
+
"epoch": 35.71,
|
68 |
+
"learning_rate": 1.928571428571429e-06,
|
69 |
+
"loss": 0.0048,
|
70 |
+
"step": 3000
|
71 |
+
},
|
72 |
+
{
|
73 |
+
"epoch": 35.71,
|
74 |
+
"eval_gen_len": 206.1478,
|
75 |
+
"eval_loss": 0.07027211785316467,
|
76 |
+
"eval_rouge1": 84.3357,
|
77 |
+
"eval_rouge2": 79.2414,
|
78 |
+
"eval_rougeL": 72.9218,
|
79 |
+
"eval_rougeLsum": 73.2212,
|
80 |
+
"eval_runtime": 2652.223,
|
81 |
+
"eval_samples_per_second": 0.077,
|
82 |
+
"eval_steps_per_second": 0.004,
|
83 |
+
"step": 3000
|
84 |
+
},
|
85 |
+
{
|
86 |
+
"epoch": 41.67,
|
87 |
+
"learning_rate": 1.7500000000000002e-06,
|
88 |
+
"loss": 0.0033,
|
89 |
+
"step": 3500
|
90 |
+
},
|
91 |
+
{
|
92 |
+
"epoch": 47.62,
|
93 |
+
"learning_rate": 1.5714285714285714e-06,
|
94 |
+
"loss": 0.0025,
|
95 |
+
"step": 4000
|
96 |
+
},
|
97 |
+
{
|
98 |
+
"epoch": 47.62,
|
99 |
+
"eval_gen_len": 210.4483,
|
100 |
+
"eval_loss": 0.06871643662452698,
|
101 |
+
"eval_rouge1": 83.8269,
|
102 |
+
"eval_rouge2": 78.3266,
|
103 |
+
"eval_rougeL": 72.7844,
|
104 |
+
"eval_rougeLsum": 73.1385,
|
105 |
+
"eval_runtime": 2371.8896,
|
106 |
+
"eval_samples_per_second": 0.086,
|
107 |
+
"eval_steps_per_second": 0.005,
|
108 |
+
"step": 4000
|
109 |
+
},
|
110 |
+
{
|
111 |
+
"epoch": 53.57,
|
112 |
+
"learning_rate": 1.392857142857143e-06,
|
113 |
+
"loss": 0.0018,
|
114 |
+
"step": 4500
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 59.52,
|
118 |
+
"learning_rate": 1.2142857142857144e-06,
|
119 |
+
"loss": 0.0016,
|
120 |
+
"step": 5000
|
121 |
+
},
|
122 |
+
{
|
123 |
+
"epoch": 59.52,
|
124 |
+
"eval_gen_len": 199.7389,
|
125 |
+
"eval_loss": 0.07927948236465454,
|
126 |
+
"eval_rouge1": 86.8825,
|
127 |
+
"eval_rouge2": 82.213,
|
128 |
+
"eval_rougeL": 77.2112,
|
129 |
+
"eval_rougeLsum": 77.4106,
|
130 |
+
"eval_runtime": 2398.724,
|
131 |
+
"eval_samples_per_second": 0.085,
|
132 |
+
"eval_steps_per_second": 0.005,
|
133 |
+
"step": 5000
|
134 |
+
},
|
135 |
+
{
|
136 |
+
"epoch": 65.48,
|
137 |
+
"learning_rate": 1.0357142857142857e-06,
|
138 |
+
"loss": 0.0012,
|
139 |
+
"step": 5500
|
140 |
+
},
|
141 |
+
{
|
142 |
+
"epoch": 71.43,
|
143 |
+
"learning_rate": 8.571428571428571e-07,
|
144 |
+
"loss": 0.001,
|
145 |
+
"step": 6000
|
146 |
+
},
|
147 |
+
{
|
148 |
+
"epoch": 71.43,
|
149 |
+
"eval_gen_len": 202.665,
|
150 |
+
"eval_loss": 0.08206681907176971,
|
151 |
+
"eval_rouge1": 86.0452,
|
152 |
+
"eval_rouge2": 81.3296,
|
153 |
+
"eval_rougeL": 75.9829,
|
154 |
+
"eval_rougeLsum": 76.3571,
|
155 |
+
"eval_runtime": 2447.0707,
|
156 |
+
"eval_samples_per_second": 0.083,
|
157 |
+
"eval_steps_per_second": 0.004,
|
158 |
+
"step": 6000
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 77.38,
|
162 |
+
"learning_rate": 6.785714285714286e-07,
|
163 |
+
"loss": 0.0009,
|
164 |
+
"step": 6500
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 83.33,
|
168 |
+
"learning_rate": 5e-07,
|
169 |
+
"loss": 0.0008,
|
170 |
+
"step": 7000
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 83.33,
|
174 |
+
"eval_gen_len": 205.0788,
|
175 |
+
"eval_loss": 0.08218137919902802,
|
176 |
+
"eval_rouge1": 85.9623,
|
177 |
+
"eval_rouge2": 81.1543,
|
178 |
+
"eval_rougeL": 76.8811,
|
179 |
+
"eval_rougeLsum": 77.0526,
|
180 |
+
"eval_runtime": 2415.0035,
|
181 |
+
"eval_samples_per_second": 0.084,
|
182 |
+
"eval_steps_per_second": 0.005,
|
183 |
+
"step": 7000
|
184 |
+
},
|
185 |
+
{
|
186 |
+
"epoch": 89.29,
|
187 |
+
"learning_rate": 3.214285714285714e-07,
|
188 |
+
"loss": 0.0007,
|
189 |
+
"step": 7500
|
190 |
+
},
|
191 |
+
{
|
192 |
+
"epoch": 95.24,
|
193 |
+
"learning_rate": 1.4285714285714285e-07,
|
194 |
+
"loss": 0.0006,
|
195 |
+
"step": 8000
|
196 |
+
},
|
197 |
+
{
|
198 |
+
"epoch": 95.24,
|
199 |
+
"eval_gen_len": 206.7488,
|
200 |
+
"eval_loss": 0.07930376380681992,
|
201 |
+
"eval_rouge1": 85.5313,
|
202 |
+
"eval_rouge2": 80.6524,
|
203 |
+
"eval_rougeL": 75.6534,
|
204 |
+
"eval_rougeLsum": 75.8372,
|
205 |
+
"eval_runtime": 2606.6221,
|
206 |
+
"eval_samples_per_second": 0.078,
|
207 |
+
"eval_steps_per_second": 0.004,
|
208 |
+
"step": 8000
|
209 |
+
}
|
210 |
+
],
|
211 |
+
"max_steps": 8400,
|
212 |
+
"num_train_epochs": 100,
|
213 |
+
"total_flos": 1.1606068796537242e+17,
|
214 |
+
"trial_name": null,
|
215 |
+
"trial_params": null
|
216 |
+
}
|