removing none-essential files for running predictions
Browse files- optimizer.pt +0 -3
- rng_state.pth +0 -3
- scheduler.pt +0 -3
- trainer_state.json +0 -226
- training_args.bin +0 -3
optimizer.pt
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:b39ef12875e83f443c75ec35530cc20fe735318dc7da605d6e6d7e502e61b775
|
3 |
-
size 338331642
|
|
|
|
|
|
|
|
rng_state.pth
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:73c2594a1e721a20cff9f81a9c6b35735e5ef856ae026e6f451a155090123133
|
3 |
-
size 14244
|
|
|
|
|
|
|
|
scheduler.pt
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:49a2812a42769889644fb265b287a3cb6966f030d4b6988b60a19dba3930cea8
|
3 |
-
size 1064
|
|
|
|
|
|
|
|
trainer_state.json
DELETED
@@ -1,226 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"best_metric": null,
|
3 |
-
"best_model_checkpoint": null,
|
4 |
-
"epoch": 9.659090909090908,
|
5 |
-
"global_step": 8500,
|
6 |
-
"is_hyper_param_search": false,
|
7 |
-
"is_local_process_zero": true,
|
8 |
-
"is_world_process_zero": true,
|
9 |
-
"log_history": [
|
10 |
-
{
|
11 |
-
"epoch": 0.57,
|
12 |
-
"learning_rate": 1.8863636363636366e-05,
|
13 |
-
"loss": 0.9603,
|
14 |
-
"step": 500
|
15 |
-
},
|
16 |
-
{
|
17 |
-
"epoch": 1.0,
|
18 |
-
"eval_accuracy": 0.917317101584343,
|
19 |
-
"eval_f1": 0.780907047467936,
|
20 |
-
"eval_loss": 0.3360126316547394,
|
21 |
-
"eval_precision": 0.7684100962789487,
|
22 |
-
"eval_recall": 0.7938172043010753,
|
23 |
-
"eval_runtime": 4.4585,
|
24 |
-
"eval_samples_per_second": 211.507,
|
25 |
-
"eval_steps_per_second": 42.391,
|
26 |
-
"step": 880
|
27 |
-
},
|
28 |
-
{
|
29 |
-
"epoch": 1.14,
|
30 |
-
"learning_rate": 1.772727272727273e-05,
|
31 |
-
"loss": 0.4145,
|
32 |
-
"step": 1000
|
33 |
-
},
|
34 |
-
{
|
35 |
-
"epoch": 1.7,
|
36 |
-
"learning_rate": 1.6590909090909094e-05,
|
37 |
-
"loss": 0.2846,
|
38 |
-
"step": 1500
|
39 |
-
},
|
40 |
-
{
|
41 |
-
"epoch": 2.0,
|
42 |
-
"eval_accuracy": 0.940004659832246,
|
43 |
-
"eval_f1": 0.8470136913817051,
|
44 |
-
"eval_loss": 0.23408983647823334,
|
45 |
-
"eval_precision": 0.8338325302773799,
|
46 |
-
"eval_recall": 0.8606182795698925,
|
47 |
-
"eval_runtime": 5.0803,
|
48 |
-
"eval_samples_per_second": 185.62,
|
49 |
-
"eval_steps_per_second": 37.203,
|
50 |
-
"step": 1760
|
51 |
-
},
|
52 |
-
{
|
53 |
-
"epoch": 2.27,
|
54 |
-
"learning_rate": 1.5454545454545454e-05,
|
55 |
-
"loss": 0.2114,
|
56 |
-
"step": 2000
|
57 |
-
},
|
58 |
-
{
|
59 |
-
"epoch": 2.84,
|
60 |
-
"learning_rate": 1.431818181818182e-05,
|
61 |
-
"loss": 0.1634,
|
62 |
-
"step": 2500
|
63 |
-
},
|
64 |
-
{
|
65 |
-
"epoch": 3.0,
|
66 |
-
"eval_accuracy": 0.9476351351351351,
|
67 |
-
"eval_f1": 0.8742690058479533,
|
68 |
-
"eval_loss": 0.20282697677612305,
|
69 |
-
"eval_precision": 0.8646161934805467,
|
70 |
-
"eval_recall": 0.8841397849462366,
|
71 |
-
"eval_runtime": 3.3011,
|
72 |
-
"eval_samples_per_second": 285.658,
|
73 |
-
"eval_steps_per_second": 57.253,
|
74 |
-
"step": 2640
|
75 |
-
},
|
76 |
-
{
|
77 |
-
"epoch": 3.41,
|
78 |
-
"learning_rate": 1.3181818181818183e-05,
|
79 |
-
"loss": 0.1243,
|
80 |
-
"step": 3000
|
81 |
-
},
|
82 |
-
{
|
83 |
-
"epoch": 3.98,
|
84 |
-
"learning_rate": 1.2045454545454547e-05,
|
85 |
-
"loss": 0.1158,
|
86 |
-
"step": 3500
|
87 |
-
},
|
88 |
-
{
|
89 |
-
"epoch": 4.0,
|
90 |
-
"eval_accuracy": 0.9500232991612302,
|
91 |
-
"eval_f1": 0.8811803278688525,
|
92 |
-
"eval_loss": 0.1980859786272049,
|
93 |
-
"eval_precision": 0.8603072983354674,
|
94 |
-
"eval_recall": 0.9030913978494624,
|
95 |
-
"eval_runtime": 4.9698,
|
96 |
-
"eval_samples_per_second": 189.745,
|
97 |
-
"eval_steps_per_second": 38.03,
|
98 |
-
"step": 3520
|
99 |
-
},
|
100 |
-
{
|
101 |
-
"epoch": 4.55,
|
102 |
-
"learning_rate": 1.0909090909090909e-05,
|
103 |
-
"loss": 0.0808,
|
104 |
-
"step": 4000
|
105 |
-
},
|
106 |
-
{
|
107 |
-
"epoch": 5.0,
|
108 |
-
"eval_accuracy": 0.9508970177073626,
|
109 |
-
"eval_f1": 0.8861842105263158,
|
110 |
-
"eval_loss": 0.19827169179916382,
|
111 |
-
"eval_precision": 0.8679123711340206,
|
112 |
-
"eval_recall": 0.905241935483871,
|
113 |
-
"eval_runtime": 4.2259,
|
114 |
-
"eval_samples_per_second": 223.15,
|
115 |
-
"eval_steps_per_second": 44.725,
|
116 |
-
"step": 4400
|
117 |
-
},
|
118 |
-
{
|
119 |
-
"epoch": 5.11,
|
120 |
-
"learning_rate": 9.772727272727273e-06,
|
121 |
-
"loss": 0.0809,
|
122 |
-
"step": 4500
|
123 |
-
},
|
124 |
-
{
|
125 |
-
"epoch": 5.68,
|
126 |
-
"learning_rate": 8.636363636363637e-06,
|
127 |
-
"loss": 0.0679,
|
128 |
-
"step": 5000
|
129 |
-
},
|
130 |
-
{
|
131 |
-
"epoch": 6.0,
|
132 |
-
"eval_accuracy": 0.952673578751165,
|
133 |
-
"eval_f1": 0.89337822671156,
|
134 |
-
"eval_loss": 0.1982535719871521,
|
135 |
-
"eval_precision": 0.8779032048786817,
|
136 |
-
"eval_recall": 0.9094086021505376,
|
137 |
-
"eval_runtime": 4.536,
|
138 |
-
"eval_samples_per_second": 207.892,
|
139 |
-
"eval_steps_per_second": 41.667,
|
140 |
-
"step": 5280
|
141 |
-
},
|
142 |
-
{
|
143 |
-
"epoch": 6.25,
|
144 |
-
"learning_rate": 7.500000000000001e-06,
|
145 |
-
"loss": 0.0582,
|
146 |
-
"step": 5500
|
147 |
-
},
|
148 |
-
{
|
149 |
-
"epoch": 6.82,
|
150 |
-
"learning_rate": 6.363636363636364e-06,
|
151 |
-
"loss": 0.0468,
|
152 |
-
"step": 6000
|
153 |
-
},
|
154 |
-
{
|
155 |
-
"epoch": 7.0,
|
156 |
-
"eval_accuracy": 0.9559354613233924,
|
157 |
-
"eval_f1": 0.9001193792280144,
|
158 |
-
"eval_loss": 0.20358432829380035,
|
159 |
-
"eval_precision": 0.8884524744697565,
|
160 |
-
"eval_recall": 0.9120967741935484,
|
161 |
-
"eval_runtime": 4.9544,
|
162 |
-
"eval_samples_per_second": 190.337,
|
163 |
-
"eval_steps_per_second": 38.148,
|
164 |
-
"step": 6160
|
165 |
-
},
|
166 |
-
{
|
167 |
-
"epoch": 7.39,
|
168 |
-
"learning_rate": 5.2272727272727274e-06,
|
169 |
-
"loss": 0.0436,
|
170 |
-
"step": 6500
|
171 |
-
},
|
172 |
-
{
|
173 |
-
"epoch": 7.95,
|
174 |
-
"learning_rate": 4.0909090909090915e-06,
|
175 |
-
"loss": 0.0403,
|
176 |
-
"step": 7000
|
177 |
-
},
|
178 |
-
{
|
179 |
-
"epoch": 8.0,
|
180 |
-
"eval_accuracy": 0.955527726001864,
|
181 |
-
"eval_f1": 0.8995240613432047,
|
182 |
-
"eval_loss": 0.20606616139411926,
|
183 |
-
"eval_precision": 0.8850156087408949,
|
184 |
-
"eval_recall": 0.9145161290322581,
|
185 |
-
"eval_runtime": 3.9742,
|
186 |
-
"eval_samples_per_second": 237.282,
|
187 |
-
"eval_steps_per_second": 47.557,
|
188 |
-
"step": 7040
|
189 |
-
},
|
190 |
-
{
|
191 |
-
"epoch": 8.52,
|
192 |
-
"learning_rate": 2.954545454545455e-06,
|
193 |
-
"loss": 0.0336,
|
194 |
-
"step": 7500
|
195 |
-
},
|
196 |
-
{
|
197 |
-
"epoch": 9.0,
|
198 |
-
"eval_accuracy": 0.9554694780987885,
|
199 |
-
"eval_f1": 0.9019062748212867,
|
200 |
-
"eval_loss": 0.21168170869350433,
|
201 |
-
"eval_precision": 0.8884976525821596,
|
202 |
-
"eval_recall": 0.915725806451613,
|
203 |
-
"eval_runtime": 2.986,
|
204 |
-
"eval_samples_per_second": 315.802,
|
205 |
-
"eval_steps_per_second": 63.294,
|
206 |
-
"step": 7920
|
207 |
-
},
|
208 |
-
{
|
209 |
-
"epoch": 9.09,
|
210 |
-
"learning_rate": 1.8181818181818183e-06,
|
211 |
-
"loss": 0.0328,
|
212 |
-
"step": 8000
|
213 |
-
},
|
214 |
-
{
|
215 |
-
"epoch": 9.66,
|
216 |
-
"learning_rate": 6.818181818181818e-07,
|
217 |
-
"loss": 0.0312,
|
218 |
-
"step": 8500
|
219 |
-
}
|
220 |
-
],
|
221 |
-
"max_steps": 8800,
|
222 |
-
"num_train_epochs": 10,
|
223 |
-
"total_flos": 408432261397380.0,
|
224 |
-
"trial_name": null,
|
225 |
-
"trial_params": null
|
226 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
training_args.bin
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:b551fbf29c3318236ff0e19876fb0afff99641683390db79faf1b4fadab8b042
|
3 |
-
size 4088
|
|
|
|
|
|
|
|