ShengdingHu
commited on
Commit
•
23c6635
1
Parent(s):
5f07a6e
Training in progress, step 200
Browse files- all_results.json +14 -14
- eval_results.json +5 -5
- pytorch_model.bin +1 -1
- runs/Jan31_19-25-22_node1/events.out.tfevents.1643628441.node1 +2 -2
- runs/Jan31_19-25-22_node1/events.out.tfevents.1643628866.node1 +3 -0
- runs/Jan31_20-22-53_node1/1643631902.337454/events.out.tfevents.1643631902.node1 +3 -0
- runs/Jan31_20-22-53_node1/events.out.tfevents.1643631902.node1 +3 -0
- test_results.json +5 -5
- tokenizer.json +0 -0
- train_results.json +4 -4
- trainer_state.json +110 -110
- training_args.bin +1 -1
- training_config.json +1 -1
all_results.json
CHANGED
@@ -1,18 +1,18 @@
|
|
1 |
{
|
2 |
"epoch": 20.0,
|
3 |
-
"eval_average_metrics":
|
4 |
-
"eval_loss": 0.
|
5 |
-
"eval_matthews_correlation":
|
6 |
-
"eval_runtime": 0.
|
7 |
-
"eval_samples_per_second":
|
8 |
-
"test_average_metrics":
|
9 |
-
"test_loss": 0.
|
10 |
-
"test_matthews_correlation":
|
11 |
-
"test_runtime": 0.
|
12 |
-
"test_samples_per_second":
|
13 |
-
"train_loss": 0.
|
14 |
-
"train_runtime":
|
15 |
"train_samples": 8551,
|
16 |
-
"train_samples_per_second":
|
17 |
-
"train_steps_per_second":
|
18 |
}
|
|
|
1 |
{
|
2 |
"epoch": 20.0,
|
3 |
+
"eval_average_metrics": 52.359840871035765,
|
4 |
+
"eval_loss": 0.2282242476940155,
|
5 |
+
"eval_matthews_correlation": 52.359840871035765,
|
6 |
+
"eval_runtime": 0.9075,
|
7 |
+
"eval_samples_per_second": 574.107,
|
8 |
+
"test_average_metrics": 63.24519496421077,
|
9 |
+
"test_loss": 0.18458954989910126,
|
10 |
+
"test_matthews_correlation": 63.24519496421077,
|
11 |
+
"test_runtime": 0.8502,
|
12 |
+
"test_samples_per_second": 614.001,
|
13 |
+
"train_loss": 0.5282898814179177,
|
14 |
+
"train_runtime": 424.1274,
|
15 |
"train_samples": 8551,
|
16 |
+
"train_samples_per_second": 403.228,
|
17 |
+
"train_steps_per_second": 4.055
|
18 |
}
|
eval_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 20.0,
|
3 |
-
"eval_average_metrics":
|
4 |
-
"eval_loss": 0.
|
5 |
-
"eval_matthews_correlation":
|
6 |
-
"eval_runtime": 0.
|
7 |
-
"eval_samples_per_second":
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 20.0,
|
3 |
+
"eval_average_metrics": 52.359840871035765,
|
4 |
+
"eval_loss": 0.2282242476940155,
|
5 |
+
"eval_matthews_correlation": 52.359840871035765,
|
6 |
+
"eval_runtime": 0.9075,
|
7 |
+
"eval_samples_per_second": 574.107
|
8 |
}
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1084131
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3a589ded531d3bbb72a7a6e9aeb707c0b45b2aa234b814b877f8e496c28af491
|
3 |
size 1084131
|
runs/Jan31_19-25-22_node1/events.out.tfevents.1643628441.node1
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:334a48cecac110d2e28a68203c3fa53bc97487bcdf69a630c7610b701c5afa43
|
3 |
+
size 11804
|
runs/Jan31_19-25-22_node1/events.out.tfevents.1643628866.node1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2b254740812b3b5b0cf5053822c0316b14a252f438bcb603dd521c2369e286dd
|
3 |
+
size 708
|
runs/Jan31_20-22-53_node1/1643631902.337454/events.out.tfevents.1643631902.node1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:394ea619e40e78d2e08e5492feceb7fc0905a3e73ac274ebea205b9352f84dcb
|
3 |
+
size 5011
|
runs/Jan31_20-22-53_node1/events.out.tfevents.1643631902.node1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2a2143aa1f51e84731a49af5406404480d81036c9e3de46611ba73399f9f75ed
|
3 |
+
size 4304
|
test_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 20.0,
|
3 |
-
"test_average_metrics":
|
4 |
-
"test_loss": 0.
|
5 |
-
"test_matthews_correlation":
|
6 |
-
"test_runtime": 0.
|
7 |
-
"test_samples_per_second":
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 20.0,
|
3 |
+
"test_average_metrics": 63.24519496421077,
|
4 |
+
"test_loss": 0.18458954989910126,
|
5 |
+
"test_matthews_correlation": 63.24519496421077,
|
6 |
+
"test_runtime": 0.8502,
|
7 |
+
"test_samples_per_second": 614.001
|
8 |
}
|
tokenizer.json
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
train_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 20.0,
|
3 |
-
"train_loss": 0.
|
4 |
-
"train_runtime":
|
5 |
"train_samples": 8551,
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second":
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 20.0,
|
3 |
+
"train_loss": 0.5282898814179177,
|
4 |
+
"train_runtime": 424.1274,
|
5 |
"train_samples": 8551,
|
6 |
+
"train_samples_per_second": 403.228,
|
7 |
+
"train_steps_per_second": 4.055
|
8 |
}
|
trainer_state.json
CHANGED
@@ -1,6 +1,6 @@
|
|
1 |
{
|
2 |
-
"best_metric":
|
3 |
-
"best_model_checkpoint": "outputs/bitfit/t5-base/cola/checkpoint-
|
4 |
"epoch": 20.0,
|
5 |
"global_step": 1720,
|
6 |
"is_hyper_param_search": false,
|
@@ -10,209 +10,209 @@
|
|
10 |
{
|
11 |
"epoch": 1.0,
|
12 |
"eval_average_metrics": 0.0,
|
13 |
-
"eval_loss":
|
14 |
"eval_matthews_correlation": 0.0,
|
15 |
-
"eval_runtime": 0.
|
16 |
-
"eval_samples_per_second":
|
17 |
"step": 86
|
18 |
},
|
19 |
{
|
20 |
"epoch": 2.0,
|
21 |
-
"eval_average_metrics":
|
22 |
-
"eval_loss": 0.
|
23 |
-
"eval_matthews_correlation":
|
24 |
-
"eval_runtime": 0.
|
25 |
-
"eval_samples_per_second":
|
26 |
"step": 172
|
27 |
},
|
28 |
{
|
29 |
"epoch": 3.0,
|
30 |
-
"eval_average_metrics":
|
31 |
-
"eval_loss": 0.
|
32 |
-
"eval_matthews_correlation":
|
33 |
-
"eval_runtime": 0.
|
34 |
-
"eval_samples_per_second":
|
35 |
"step": 258
|
36 |
},
|
37 |
{
|
38 |
"epoch": 4.0,
|
39 |
-
"eval_average_metrics":
|
40 |
-
"eval_loss": 0.
|
41 |
-
"eval_matthews_correlation":
|
42 |
-
"eval_runtime": 0.
|
43 |
-
"eval_samples_per_second":
|
44 |
"step": 344
|
45 |
},
|
46 |
{
|
47 |
"epoch": 5.0,
|
48 |
-
"eval_average_metrics":
|
49 |
-
"eval_loss": 0.
|
50 |
-
"eval_matthews_correlation":
|
51 |
-
"eval_runtime": 0.
|
52 |
-
"eval_samples_per_second":
|
53 |
"step": 430
|
54 |
},
|
55 |
{
|
56 |
"epoch": 5.81,
|
57 |
-
"learning_rate": 0.
|
58 |
-
"loss":
|
59 |
"step": 500
|
60 |
},
|
61 |
{
|
62 |
"epoch": 6.0,
|
63 |
-
"eval_average_metrics":
|
64 |
-
"eval_loss": 0.
|
65 |
-
"eval_matthews_correlation":
|
66 |
-
"eval_runtime": 0.
|
67 |
-
"eval_samples_per_second":
|
68 |
"step": 516
|
69 |
},
|
70 |
{
|
71 |
"epoch": 7.0,
|
72 |
-
"eval_average_metrics":
|
73 |
-
"eval_loss": 0.
|
74 |
-
"eval_matthews_correlation":
|
75 |
-
"eval_runtime": 0.
|
76 |
-
"eval_samples_per_second":
|
77 |
"step": 602
|
78 |
},
|
79 |
{
|
80 |
"epoch": 8.0,
|
81 |
-
"eval_average_metrics":
|
82 |
-
"eval_loss": 0.
|
83 |
-
"eval_matthews_correlation":
|
84 |
-
"eval_runtime": 0.
|
85 |
-
"eval_samples_per_second":
|
86 |
"step": 688
|
87 |
},
|
88 |
{
|
89 |
"epoch": 9.0,
|
90 |
-
"eval_average_metrics":
|
91 |
-
"eval_loss": 0.
|
92 |
-
"eval_matthews_correlation":
|
93 |
-
"eval_runtime": 0.
|
94 |
-
"eval_samples_per_second":
|
95 |
"step": 774
|
96 |
},
|
97 |
{
|
98 |
"epoch": 10.0,
|
99 |
-
"eval_average_metrics":
|
100 |
-
"eval_loss": 0.
|
101 |
-
"eval_matthews_correlation":
|
102 |
-
"eval_runtime": 0.
|
103 |
-
"eval_samples_per_second":
|
104 |
"step": 860
|
105 |
},
|
106 |
{
|
107 |
"epoch": 11.0,
|
108 |
-
"eval_average_metrics":
|
109 |
-
"eval_loss": 0.
|
110 |
-
"eval_matthews_correlation":
|
111 |
-
"eval_runtime": 0.
|
112 |
-
"eval_samples_per_second":
|
113 |
"step": 946
|
114 |
},
|
115 |
{
|
116 |
"epoch": 11.63,
|
117 |
-
"learning_rate": 0.
|
118 |
-
"loss": 0.
|
119 |
"step": 1000
|
120 |
},
|
121 |
{
|
122 |
"epoch": 12.0,
|
123 |
-
"eval_average_metrics":
|
124 |
-
"eval_loss": 0.
|
125 |
-
"eval_matthews_correlation":
|
126 |
-
"eval_runtime": 0.
|
127 |
-
"eval_samples_per_second":
|
128 |
"step": 1032
|
129 |
},
|
130 |
{
|
131 |
"epoch": 13.0,
|
132 |
-
"eval_average_metrics":
|
133 |
-
"eval_loss": 0.
|
134 |
-
"eval_matthews_correlation":
|
135 |
-
"eval_runtime": 0.
|
136 |
-
"eval_samples_per_second":
|
137 |
"step": 1118
|
138 |
},
|
139 |
{
|
140 |
"epoch": 14.0,
|
141 |
-
"eval_average_metrics":
|
142 |
-
"eval_loss": 0.
|
143 |
-
"eval_matthews_correlation":
|
144 |
-
"eval_runtime": 0.
|
145 |
-
"eval_samples_per_second":
|
146 |
"step": 1204
|
147 |
},
|
148 |
{
|
149 |
"epoch": 15.0,
|
150 |
-
"eval_average_metrics":
|
151 |
-
"eval_loss": 0.
|
152 |
-
"eval_matthews_correlation":
|
153 |
-
"eval_runtime": 0.
|
154 |
-
"eval_samples_per_second":
|
155 |
"step": 1290
|
156 |
},
|
157 |
{
|
158 |
"epoch": 16.0,
|
159 |
-
"eval_average_metrics":
|
160 |
-
"eval_loss": 0.
|
161 |
-
"eval_matthews_correlation":
|
162 |
-
"eval_runtime": 0.
|
163 |
-
"eval_samples_per_second":
|
164 |
"step": 1376
|
165 |
},
|
166 |
{
|
167 |
"epoch": 17.0,
|
168 |
-
"eval_average_metrics":
|
169 |
-
"eval_loss": 0.
|
170 |
-
"eval_matthews_correlation":
|
171 |
-
"eval_runtime": 0.
|
172 |
-
"eval_samples_per_second":
|
173 |
"step": 1462
|
174 |
},
|
175 |
{
|
176 |
"epoch": 17.44,
|
177 |
-
"learning_rate":
|
178 |
-
"loss": 0.
|
179 |
"step": 1500
|
180 |
},
|
181 |
{
|
182 |
"epoch": 18.0,
|
183 |
-
"eval_average_metrics":
|
184 |
-
"eval_loss": 0.
|
185 |
-
"eval_matthews_correlation":
|
186 |
-
"eval_runtime": 0.
|
187 |
-
"eval_samples_per_second":
|
188 |
"step": 1548
|
189 |
},
|
190 |
{
|
191 |
"epoch": 19.0,
|
192 |
-
"eval_average_metrics":
|
193 |
-
"eval_loss": 0.
|
194 |
-
"eval_matthews_correlation":
|
195 |
-
"eval_runtime": 0.
|
196 |
-
"eval_samples_per_second":
|
197 |
"step": 1634
|
198 |
},
|
199 |
{
|
200 |
"epoch": 20.0,
|
201 |
-
"eval_average_metrics":
|
202 |
-
"eval_loss": 0.
|
203 |
-
"eval_matthews_correlation":
|
204 |
-
"eval_runtime": 0.
|
205 |
-
"eval_samples_per_second":
|
206 |
"step": 1720
|
207 |
},
|
208 |
{
|
209 |
"epoch": 20.0,
|
210 |
"step": 1720,
|
211 |
"total_flos": 6646762904679216.0,
|
212 |
-
"train_loss": 0.
|
213 |
-
"train_runtime":
|
214 |
-
"train_samples_per_second":
|
215 |
-
"train_steps_per_second":
|
216 |
}
|
217 |
],
|
218 |
"max_steps": 1720,
|
|
|
1 |
{
|
2 |
+
"best_metric": 52.359840871035765,
|
3 |
+
"best_model_checkpoint": "outputs/bitfit/t5-base/cola/checkpoint-1290",
|
4 |
"epoch": 20.0,
|
5 |
"global_step": 1720,
|
6 |
"is_hyper_param_search": false,
|
|
|
10 |
{
|
11 |
"epoch": 1.0,
|
12 |
"eval_average_metrics": 0.0,
|
13 |
+
"eval_loss": 4.8104567527771,
|
14 |
"eval_matthews_correlation": 0.0,
|
15 |
+
"eval_runtime": 0.763,
|
16 |
+
"eval_samples_per_second": 682.828,
|
17 |
"step": 86
|
18 |
},
|
19 |
{
|
20 |
"epoch": 2.0,
|
21 |
+
"eval_average_metrics": 8.869450752831899,
|
22 |
+
"eval_loss": 0.27582260966300964,
|
23 |
+
"eval_matthews_correlation": 8.869450752831899,
|
24 |
+
"eval_runtime": 0.8417,
|
25 |
+
"eval_samples_per_second": 619.016,
|
26 |
"step": 172
|
27 |
},
|
28 |
{
|
29 |
"epoch": 3.0,
|
30 |
+
"eval_average_metrics": 43.97683870357735,
|
31 |
+
"eval_loss": 0.21786993741989136,
|
32 |
+
"eval_matthews_correlation": 43.97683870357735,
|
33 |
+
"eval_runtime": 0.8315,
|
34 |
+
"eval_samples_per_second": 626.599,
|
35 |
"step": 258
|
36 |
},
|
37 |
{
|
38 |
"epoch": 4.0,
|
39 |
+
"eval_average_metrics": 46.78131759250157,
|
40 |
+
"eval_loss": 0.22430144250392914,
|
41 |
+
"eval_matthews_correlation": 46.78131759250157,
|
42 |
+
"eval_runtime": 0.8615,
|
43 |
+
"eval_samples_per_second": 604.782,
|
44 |
"step": 344
|
45 |
},
|
46 |
{
|
47 |
"epoch": 5.0,
|
48 |
+
"eval_average_metrics": 49.53306157776009,
|
49 |
+
"eval_loss": 0.21720413863658905,
|
50 |
+
"eval_matthews_correlation": 49.53306157776009,
|
51 |
+
"eval_runtime": 0.8515,
|
52 |
+
"eval_samples_per_second": 611.84,
|
53 |
"step": 430
|
54 |
},
|
55 |
{
|
56 |
"epoch": 5.81,
|
57 |
+
"learning_rate": 0.0003,
|
58 |
+
"loss": 1.3817,
|
59 |
"step": 500
|
60 |
},
|
61 |
{
|
62 |
"epoch": 6.0,
|
63 |
+
"eval_average_metrics": 48.16295515684713,
|
64 |
+
"eval_loss": 0.22913119196891785,
|
65 |
+
"eval_matthews_correlation": 48.16295515684713,
|
66 |
+
"eval_runtime": 0.8751,
|
67 |
+
"eval_samples_per_second": 595.357,
|
68 |
"step": 516
|
69 |
},
|
70 |
{
|
71 |
"epoch": 7.0,
|
72 |
+
"eval_average_metrics": 47.85938832793908,
|
73 |
+
"eval_loss": 0.23619267344474792,
|
74 |
+
"eval_matthews_correlation": 47.85938832793908,
|
75 |
+
"eval_runtime": 0.8864,
|
76 |
+
"eval_samples_per_second": 587.749,
|
77 |
"step": 602
|
78 |
},
|
79 |
{
|
80 |
"epoch": 8.0,
|
81 |
+
"eval_average_metrics": 49.724093419139464,
|
82 |
+
"eval_loss": 0.2390584498643875,
|
83 |
+
"eval_matthews_correlation": 49.724093419139464,
|
84 |
+
"eval_runtime": 0.8668,
|
85 |
+
"eval_samples_per_second": 601.078,
|
86 |
"step": 688
|
87 |
},
|
88 |
{
|
89 |
"epoch": 9.0,
|
90 |
+
"eval_average_metrics": 48.86384938654592,
|
91 |
+
"eval_loss": 0.22121600806713104,
|
92 |
+
"eval_matthews_correlation": 48.86384938654592,
|
93 |
+
"eval_runtime": 0.8908,
|
94 |
+
"eval_samples_per_second": 584.879,
|
95 |
"step": 774
|
96 |
},
|
97 |
{
|
98 |
"epoch": 10.0,
|
99 |
+
"eval_average_metrics": 48.08237616875362,
|
100 |
+
"eval_loss": 0.2251322716474533,
|
101 |
+
"eval_matthews_correlation": 48.08237616875362,
|
102 |
+
"eval_runtime": 0.8537,
|
103 |
+
"eval_samples_per_second": 610.303,
|
104 |
"step": 860
|
105 |
},
|
106 |
{
|
107 |
"epoch": 11.0,
|
108 |
+
"eval_average_metrics": 47.083822335710934,
|
109 |
+
"eval_loss": 0.23612065613269806,
|
110 |
+
"eval_matthews_correlation": 47.083822335710934,
|
111 |
+
"eval_runtime": 0.852,
|
112 |
+
"eval_samples_per_second": 611.521,
|
113 |
"step": 946
|
114 |
},
|
115 |
{
|
116 |
"epoch": 11.63,
|
117 |
+
"learning_rate": 0.00017704918032786883,
|
118 |
+
"loss": 0.1859,
|
119 |
"step": 1000
|
120 |
},
|
121 |
{
|
122 |
"epoch": 12.0,
|
123 |
+
"eval_average_metrics": 49.76615268960096,
|
124 |
+
"eval_loss": 0.2268751710653305,
|
125 |
+
"eval_matthews_correlation": 49.76615268960096,
|
126 |
+
"eval_runtime": 0.8867,
|
127 |
+
"eval_samples_per_second": 587.54,
|
128 |
"step": 1032
|
129 |
},
|
130 |
{
|
131 |
"epoch": 13.0,
|
132 |
+
"eval_average_metrics": 50.45736339822007,
|
133 |
+
"eval_loss": 0.22725746035575867,
|
134 |
+
"eval_matthews_correlation": 50.45736339822007,
|
135 |
+
"eval_runtime": 0.848,
|
136 |
+
"eval_samples_per_second": 614.383,
|
137 |
"step": 1118
|
138 |
},
|
139 |
{
|
140 |
"epoch": 14.0,
|
141 |
+
"eval_average_metrics": 51.11915736464516,
|
142 |
+
"eval_loss": 0.22515951097011566,
|
143 |
+
"eval_matthews_correlation": 51.11915736464516,
|
144 |
+
"eval_runtime": 0.8495,
|
145 |
+
"eval_samples_per_second": 613.291,
|
146 |
"step": 1204
|
147 |
},
|
148 |
{
|
149 |
"epoch": 15.0,
|
150 |
+
"eval_average_metrics": 52.359840871035765,
|
151 |
+
"eval_loss": 0.2282242476940155,
|
152 |
+
"eval_matthews_correlation": 52.359840871035765,
|
153 |
+
"eval_runtime": 0.9084,
|
154 |
+
"eval_samples_per_second": 573.516,
|
155 |
"step": 1290
|
156 |
},
|
157 |
{
|
158 |
"epoch": 16.0,
|
159 |
+
"eval_average_metrics": 50.78390137665081,
|
160 |
+
"eval_loss": 0.24213995039463043,
|
161 |
+
"eval_matthews_correlation": 50.78390137665081,
|
162 |
+
"eval_runtime": 0.7976,
|
163 |
+
"eval_samples_per_second": 653.201,
|
164 |
"step": 1376
|
165 |
},
|
166 |
{
|
167 |
"epoch": 17.0,
|
168 |
+
"eval_average_metrics": 51.452415683714314,
|
169 |
+
"eval_loss": 0.22793160378932953,
|
170 |
+
"eval_matthews_correlation": 51.452415683714314,
|
171 |
+
"eval_runtime": 0.8964,
|
172 |
+
"eval_samples_per_second": 581.215,
|
173 |
"step": 1462
|
174 |
},
|
175 |
{
|
176 |
"epoch": 17.44,
|
177 |
+
"learning_rate": 5.40983606557377e-05,
|
178 |
+
"loss": 0.1748,
|
179 |
"step": 1500
|
180 |
},
|
181 |
{
|
182 |
"epoch": 18.0,
|
183 |
+
"eval_average_metrics": 51.452415683714314,
|
184 |
+
"eval_loss": 0.23017793893814087,
|
185 |
+
"eval_matthews_correlation": 51.452415683714314,
|
186 |
+
"eval_runtime": 0.8935,
|
187 |
+
"eval_samples_per_second": 583.099,
|
188 |
"step": 1548
|
189 |
},
|
190 |
{
|
191 |
"epoch": 19.0,
|
192 |
+
"eval_average_metrics": 50.9850664467699,
|
193 |
+
"eval_loss": 0.2308470457792282,
|
194 |
+
"eval_matthews_correlation": 50.9850664467699,
|
195 |
+
"eval_runtime": 0.8691,
|
196 |
+
"eval_samples_per_second": 599.475,
|
197 |
"step": 1634
|
198 |
},
|
199 |
{
|
200 |
"epoch": 20.0,
|
201 |
+
"eval_average_metrics": 50.9850664467699,
|
202 |
+
"eval_loss": 0.22891275584697723,
|
203 |
+
"eval_matthews_correlation": 50.9850664467699,
|
204 |
+
"eval_runtime": 0.8366,
|
205 |
+
"eval_samples_per_second": 622.759,
|
206 |
"step": 1720
|
207 |
},
|
208 |
{
|
209 |
"epoch": 20.0,
|
210 |
"step": 1720,
|
211 |
"total_flos": 6646762904679216.0,
|
212 |
+
"train_loss": 0.5282898814179177,
|
213 |
+
"train_runtime": 424.1274,
|
214 |
+
"train_samples_per_second": 403.228,
|
215 |
+
"train_steps_per_second": 4.055
|
216 |
}
|
217 |
],
|
218 |
"max_steps": 1720,
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3183
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5a646a3fb66d1560b545d08f84e93ab16cd3153ff19e20982a22b554a34ec127
|
3 |
size 3183
|
training_config.json
CHANGED
@@ -1 +1 @@
|
|
1 |
-
{"dataset_config_name": ["en"], "delta_type": "bitfit", "do_eval": true, "do_test": true, "do_train": true, "eval_dataset_config_name": ["en"], "eval_dataset_name": "cola", "evaluation_strategy": "
|
|
|
1 |
+
{"dataset_config_name": ["en"], "delta_type": "bitfit", "do_eval": true, "do_test": true, "do_train": true, "eval_dataset_config_name": ["en"], "eval_dataset_name": "cola", "eval_steps": 200, "evaluation_strategy": "steps", "greater_is_better": true, "learning_rate": 0.0003, "load_best_model_at_end": true, "max_source_length": 128, "metric_for_best_model": "average_metrics", "model_name_or_path": "../../../../plm_cache/t5-base", "num_train_epochs": 20, "output_dir": "outputs/bitfit/t5-base/cola", "overwrite_output_dir": true, "per_device_eval_batch_size": 100, "per_device_train_batch_size": 100, "predict_with_generate": true, "push_to_hub": true, "save_steps": 200, "save_strategy": "steps", "save_total_limit": 1, "seed": 42, "split_validation_test": true, "task_name": "cola", "test_dataset_config_name": ["en"], "test_dataset_name": "cola", "tokenizer_name": "../../../../plm_cache/t5-base", "warmup_steps": 500}
|