Commit
•
f62d69e
1
Parent(s):
c5d34a6
Updated Commit
Browse files- README.md +28 -6
- all_results.json +10 -10
- eval_results.json +6 -6
- pytorch_model.bin +1 -1
- train_results.json +5 -5
- trainer_state.json +207 -335
- training_args.bin +1 -1
README.md
CHANGED
@@ -5,18 +5,18 @@ tags:
|
|
5 |
datasets:
|
6 |
- pritamdeka/cord-19-abstract
|
7 |
model-index:
|
8 |
-
- name:
|
9 |
results: []
|
10 |
---
|
11 |
|
12 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
13 |
should probably proofread and complete it, then remove this comment. -->
|
14 |
|
15 |
-
#
|
16 |
|
17 |
-
This model is a fine-tuned version of [microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext](https://huggingface.co/microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext) on the
|
18 |
It achieves the following results on the evaluation set:
|
19 |
-
- Loss: 1.
|
20 |
|
21 |
## Model description
|
22 |
|
@@ -42,16 +42,38 @@ The following hyperparameters were used during training:
|
|
42 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
43 |
- lr_scheduler_type: linear
|
44 |
- lr_scheduler_warmup_steps: 10000
|
45 |
-
- num_epochs:
|
46 |
- mixed_precision_training: Native AMP
|
47 |
|
48 |
### Training results
|
49 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
50 |
|
51 |
|
52 |
### Framework versions
|
53 |
|
54 |
- Transformers 4.17.0.dev0
|
55 |
- Pytorch 1.10.0+cu111
|
56 |
-
- Datasets 1.18.
|
57 |
- Tokenizers 0.11.0
|
|
|
5 |
datasets:
|
6 |
- pritamdeka/cord-19-abstract
|
7 |
model-index:
|
8 |
+
- name: pubmedbert-abstract-cord19
|
9 |
results: []
|
10 |
---
|
11 |
|
12 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
13 |
should probably proofread and complete it, then remove this comment. -->
|
14 |
|
15 |
+
# pubmedbert-abstract-cord19
|
16 |
|
17 |
+
This model is a fine-tuned version of [microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext](https://huggingface.co/microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext) on the pritamdeka/cord-19-abstract dataset.
|
18 |
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: 1.3005
|
20 |
|
21 |
## Model description
|
22 |
|
|
|
42 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
43 |
- lr_scheduler_type: linear
|
44 |
- lr_scheduler_warmup_steps: 10000
|
45 |
+
- num_epochs: 3.0
|
46 |
- mixed_precision_training: Native AMP
|
47 |
|
48 |
### Training results
|
49 |
|
50 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
51 |
+
|:-------------:|:-----:|:------:|:---------------:|
|
52 |
+
| 1.3774 | 0.15 | 5000 | 1.3212 |
|
53 |
+
| 1.3937 | 0.29 | 10000 | 1.4059 |
|
54 |
+
| 1.6812 | 0.44 | 15000 | 1.6174 |
|
55 |
+
| 1.4712 | 0.59 | 20000 | 1.4383 |
|
56 |
+
| 1.4293 | 0.73 | 25000 | 1.4356 |
|
57 |
+
| 1.4155 | 0.88 | 30000 | 1.4283 |
|
58 |
+
| 1.3963 | 1.03 | 35000 | 1.4135 |
|
59 |
+
| 1.3718 | 1.18 | 40000 | 1.3948 |
|
60 |
+
| 1.369 | 1.32 | 45000 | 1.3961 |
|
61 |
+
| 1.354 | 1.47 | 50000 | 1.3788 |
|
62 |
+
| 1.3399 | 1.62 | 55000 | 1.3866 |
|
63 |
+
| 1.3289 | 1.76 | 60000 | 1.3630 |
|
64 |
+
| 1.3155 | 1.91 | 65000 | 1.3609 |
|
65 |
+
| 1.2976 | 2.06 | 70000 | 1.3489 |
|
66 |
+
| 1.2783 | 2.2 | 75000 | 1.3333 |
|
67 |
+
| 1.2696 | 2.35 | 80000 | 1.3260 |
|
68 |
+
| 1.2607 | 2.5 | 85000 | 1.3232 |
|
69 |
+
| 1.2547 | 2.64 | 90000 | 1.3034 |
|
70 |
+
| 1.2495 | 2.79 | 95000 | 1.3035 |
|
71 |
+
| 1.2404 | 2.94 | 100000 | 1.3029 |
|
72 |
|
73 |
|
74 |
### Framework versions
|
75 |
|
76 |
- Transformers 4.17.0.dev0
|
77 |
- Pytorch 1.10.0+cu111
|
78 |
+
- Datasets 1.18.3
|
79 |
- Tokenizers 0.11.0
|
all_results.json
CHANGED
@@ -1,14 +1,14 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"eval_loss": 1.
|
4 |
-
"eval_runtime":
|
5 |
"eval_samples": 14605,
|
6 |
-
"eval_samples_per_second":
|
7 |
-
"eval_steps_per_second": 10.
|
8 |
-
"perplexity": 3.
|
9 |
-
"train_loss": 1.
|
10 |
-
"train_runtime":
|
11 |
"train_samples": 272217,
|
12 |
-
"train_samples_per_second":
|
13 |
-
"train_steps_per_second": 3.
|
14 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_loss": 1.3005212545394897,
|
4 |
+
"eval_runtime": 168.9513,
|
5 |
"eval_samples": 14605,
|
6 |
+
"eval_samples_per_second": 86.445,
|
7 |
+
"eval_steps_per_second": 10.808,
|
8 |
+
"perplexity": 3.6712098037361005,
|
9 |
+
"train_loss": 1.35220570928797,
|
10 |
+
"train_runtime": 32542.0864,
|
11 |
"train_samples": 272217,
|
12 |
+
"train_samples_per_second": 25.095,
|
13 |
+
"train_steps_per_second": 3.137
|
14 |
}
|
eval_results.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"eval_loss": 1.
|
4 |
-
"eval_runtime":
|
5 |
"eval_samples": 14605,
|
6 |
-
"eval_samples_per_second":
|
7 |
-
"eval_steps_per_second": 10.
|
8 |
-
"perplexity": 3.
|
9 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_loss": 1.3005212545394897,
|
4 |
+
"eval_runtime": 168.9513,
|
5 |
"eval_samples": 14605,
|
6 |
+
"eval_samples_per_second": 86.445,
|
7 |
+
"eval_steps_per_second": 10.808,
|
8 |
+
"perplexity": 3.6712098037361005
|
9 |
}
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 438141995
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ce02810425330e4b0a62825f9aa6bb5dfd3394c5a5ae82a04b7169a7bdb17e60
|
3 |
size 438141995
|
train_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"train_loss": 1.
|
4 |
-
"train_runtime":
|
5 |
"train_samples": 272217,
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second": 3.
|
8 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"train_loss": 1.35220570928797,
|
4 |
+
"train_runtime": 32542.0864,
|
5 |
"train_samples": 272217,
|
6 |
+
"train_samples_per_second": 25.095,
|
7 |
+
"train_steps_per_second": 3.137
|
8 |
}
|
trainer_state.json
CHANGED
@@ -1,433 +1,305 @@
|
|
1 |
{
|
2 |
-
"best_metric":
|
3 |
-
"best_model_checkpoint":
|
4 |
-
"epoch":
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
9 |
"log_history": [
|
10 |
-
{
|
11 |
-
"epoch": 0.01,
|
12 |
-
"learning_rate": 2.46e-06,
|
13 |
-
"loss": 1.4943,
|
14 |
-
"step": 500
|
15 |
-
},
|
16 |
-
{
|
17 |
-
"epoch": 0.03,
|
18 |
-
"learning_rate": 4.96e-06,
|
19 |
-
"loss": 1.4172,
|
20 |
-
"step": 1000
|
21 |
-
},
|
22 |
-
{
|
23 |
-
"epoch": 0.04,
|
24 |
-
"learning_rate": 7.4600000000000006e-06,
|
25 |
-
"loss": 1.3694,
|
26 |
-
"step": 1500
|
27 |
-
},
|
28 |
-
{
|
29 |
-
"epoch": 0.06,
|
30 |
-
"learning_rate": 9.96e-06,
|
31 |
-
"loss": 1.3653,
|
32 |
-
"step": 2000
|
33 |
-
},
|
34 |
-
{
|
35 |
-
"epoch": 0.07,
|
36 |
-
"learning_rate": 1.2460000000000001e-05,
|
37 |
-
"loss": 1.3572,
|
38 |
-
"step": 2500
|
39 |
-
},
|
40 |
-
{
|
41 |
-
"epoch": 0.09,
|
42 |
-
"learning_rate": 1.4960000000000002e-05,
|
43 |
-
"loss": 1.3499,
|
44 |
-
"step": 3000
|
45 |
-
},
|
46 |
-
{
|
47 |
-
"epoch": 0.1,
|
48 |
-
"learning_rate": 1.7460000000000002e-05,
|
49 |
-
"loss": 1.347,
|
50 |
-
"step": 3500
|
51 |
-
},
|
52 |
-
{
|
53 |
-
"epoch": 0.12,
|
54 |
-
"learning_rate": 1.9960000000000002e-05,
|
55 |
-
"loss": 1.3345,
|
56 |
-
"step": 4000
|
57 |
-
},
|
58 |
-
{
|
59 |
-
"epoch": 0.13,
|
60 |
-
"learning_rate": 2.2460000000000002e-05,
|
61 |
-
"loss": 1.3558,
|
62 |
-
"step": 4500
|
63 |
-
},
|
64 |
{
|
65 |
"epoch": 0.15,
|
66 |
"learning_rate": 2.496e-05,
|
67 |
-
"loss": 1.
|
68 |
"step": 5000
|
69 |
},
|
70 |
{
|
71 |
-
"epoch": 0.
|
72 |
-
"
|
73 |
-
"
|
74 |
-
"
|
75 |
-
|
76 |
-
|
77 |
-
"epoch": 0.18,
|
78 |
-
"learning_rate": 2.9959999999999998e-05,
|
79 |
-
"loss": 1.3708,
|
80 |
-
"step": 6000
|
81 |
-
},
|
82 |
-
{
|
83 |
-
"epoch": 0.19,
|
84 |
-
"learning_rate": 3.2460000000000004e-05,
|
85 |
-
"loss": 1.3762,
|
86 |
-
"step": 6500
|
87 |
-
},
|
88 |
-
{
|
89 |
-
"epoch": 0.21,
|
90 |
-
"learning_rate": 3.4960000000000004e-05,
|
91 |
-
"loss": 1.3771,
|
92 |
-
"step": 7000
|
93 |
-
},
|
94 |
-
{
|
95 |
-
"epoch": 0.22,
|
96 |
-
"learning_rate": 3.7460000000000004e-05,
|
97 |
-
"loss": 1.391,
|
98 |
-
"step": 7500
|
99 |
-
},
|
100 |
-
{
|
101 |
-
"epoch": 0.24,
|
102 |
-
"learning_rate": 3.9960000000000004e-05,
|
103 |
-
"loss": 1.3875,
|
104 |
-
"step": 8000
|
105 |
-
},
|
106 |
-
{
|
107 |
-
"epoch": 0.25,
|
108 |
-
"learning_rate": 4.246e-05,
|
109 |
-
"loss": 1.3885,
|
110 |
-
"step": 8500
|
111 |
-
},
|
112 |
-
{
|
113 |
-
"epoch": 0.26,
|
114 |
-
"learning_rate": 4.496e-05,
|
115 |
-
"loss": 1.4058,
|
116 |
-
"step": 9000
|
117 |
-
},
|
118 |
-
{
|
119 |
-
"epoch": 0.28,
|
120 |
-
"learning_rate": 4.746e-05,
|
121 |
-
"loss": 1.4011,
|
122 |
-
"step": 9500
|
123 |
},
|
124 |
{
|
125 |
"epoch": 0.29,
|
126 |
-
"learning_rate": 4.
|
127 |
-
"loss": 1.
|
128 |
"step": 10000
|
129 |
},
|
130 |
{
|
131 |
-
"epoch": 0.
|
132 |
-
"
|
133 |
-
"
|
134 |
-
"
|
135 |
-
|
136 |
-
|
137 |
-
"epoch": 0.32,
|
138 |
-
"learning_rate": 4.793782254036957e-05,
|
139 |
-
"loss": 1.4302,
|
140 |
-
"step": 11000
|
141 |
-
},
|
142 |
-
{
|
143 |
-
"epoch": 0.34,
|
144 |
-
"learning_rate": 4.689736973530881e-05,
|
145 |
-
"loss": 1.4189,
|
146 |
-
"step": 11500
|
147 |
-
},
|
148 |
-
{
|
149 |
-
"epoch": 0.35,
|
150 |
-
"learning_rate": 4.585899783585817e-05,
|
151 |
-
"loss": 1.4264,
|
152 |
-
"step": 12000
|
153 |
-
},
|
154 |
-
{
|
155 |
-
"epoch": 0.37,
|
156 |
-
"learning_rate": 4.482062593640753e-05,
|
157 |
-
"loss": 1.4302,
|
158 |
-
"step": 12500
|
159 |
-
},
|
160 |
-
{
|
161 |
-
"epoch": 0.38,
|
162 |
-
"learning_rate": 4.378017313134676e-05,
|
163 |
-
"loss": 1.4285,
|
164 |
-
"step": 13000
|
165 |
-
},
|
166 |
-
{
|
167 |
-
"epoch": 0.4,
|
168 |
-
"learning_rate": 4.2739720326286004e-05,
|
169 |
-
"loss": 1.4212,
|
170 |
-
"step": 13500
|
171 |
-
},
|
172 |
-
{
|
173 |
-
"epoch": 0.41,
|
174 |
-
"learning_rate": 4.169926752122524e-05,
|
175 |
-
"loss": 1.4084,
|
176 |
-
"step": 14000
|
177 |
-
},
|
178 |
-
{
|
179 |
-
"epoch": 0.43,
|
180 |
-
"learning_rate": 4.0662976527384723e-05,
|
181 |
-
"loss": 1.4326,
|
182 |
-
"step": 14500
|
183 |
},
|
184 |
{
|
185 |
"epoch": 0.44,
|
186 |
-
"learning_rate":
|
187 |
-
"loss": 1.
|
188 |
"step": 15000
|
189 |
},
|
190 |
{
|
191 |
-
"epoch": 0.
|
192 |
-
"
|
193 |
-
"
|
194 |
-
"
|
195 |
-
|
196 |
-
|
197 |
-
"epoch": 0.47,
|
198 |
-
"learning_rate": 3.754161811220243e-05,
|
199 |
-
"loss": 1.5707,
|
200 |
-
"step": 16000
|
201 |
-
},
|
202 |
-
{
|
203 |
-
"epoch": 0.48,
|
204 |
-
"learning_rate": 3.650324621275179e-05,
|
205 |
-
"loss": 2.1501,
|
206 |
-
"step": 16500
|
207 |
-
},
|
208 |
-
{
|
209 |
-
"epoch": 0.5,
|
210 |
-
"learning_rate": 3.546279340769103e-05,
|
211 |
-
"loss": 1.8142,
|
212 |
-
"step": 17000
|
213 |
},
|
214 |
{
|
215 |
-
"epoch": 0.
|
216 |
-
"learning_rate":
|
217 |
-
"loss": 1.
|
218 |
-
"step":
|
219 |
},
|
220 |
{
|
221 |
-
"epoch": 0.
|
222 |
-
"
|
223 |
-
"
|
224 |
-
"
|
|
|
|
|
225 |
},
|
226 |
{
|
227 |
-
"epoch": 0.
|
228 |
-
"learning_rate":
|
229 |
-
"loss": 1.
|
230 |
-
"step":
|
231 |
},
|
232 |
{
|
233 |
-
"epoch": 0.
|
234 |
-
"
|
235 |
-
"
|
236 |
-
"
|
|
|
|
|
237 |
},
|
238 |
{
|
239 |
-
"epoch": 0.
|
240 |
-
"learning_rate": 3.
|
241 |
-
"loss": 1.
|
242 |
-
"step":
|
243 |
},
|
244 |
{
|
245 |
-
"epoch": 0.
|
246 |
-
"
|
247 |
-
"
|
248 |
-
"
|
|
|
|
|
249 |
},
|
250 |
{
|
251 |
-
"epoch":
|
252 |
-
"learning_rate":
|
253 |
-
"loss": 1.
|
254 |
-
"step":
|
255 |
},
|
256 |
{
|
257 |
-
"epoch":
|
258 |
-
"
|
259 |
-
"
|
260 |
-
"
|
|
|
|
|
261 |
},
|
262 |
{
|
263 |
-
"epoch":
|
264 |
-
"learning_rate":
|
265 |
-
"loss": 1.
|
266 |
-
"step":
|
267 |
},
|
268 |
{
|
269 |
-
"epoch":
|
270 |
-
"
|
271 |
-
"
|
272 |
-
"
|
|
|
|
|
273 |
},
|
274 |
{
|
275 |
-
"epoch":
|
276 |
-
"learning_rate":
|
277 |
-
"loss": 1.
|
278 |
-
"step":
|
279 |
},
|
280 |
{
|
281 |
-
"epoch":
|
282 |
-
"
|
283 |
-
"
|
284 |
-
"
|
|
|
|
|
285 |
},
|
286 |
{
|
287 |
-
"epoch":
|
288 |
-
"learning_rate": 2.
|
289 |
-
"loss": 1.
|
290 |
-
"step":
|
291 |
},
|
292 |
{
|
293 |
-
"epoch":
|
294 |
-
"
|
295 |
-
"
|
296 |
-
"
|
|
|
|
|
297 |
},
|
298 |
{
|
299 |
-
"epoch":
|
300 |
-
"learning_rate":
|
301 |
-
"loss": 1.
|
302 |
-
"step":
|
303 |
},
|
304 |
{
|
305 |
-
"epoch":
|
306 |
-
"
|
307 |
-
"
|
308 |
-
"
|
|
|
|
|
309 |
},
|
310 |
{
|
311 |
-
"epoch":
|
312 |
-
"learning_rate":
|
313 |
-
"loss": 1.
|
314 |
-
"step":
|
315 |
},
|
316 |
{
|
317 |
-
"epoch":
|
318 |
-
"
|
319 |
-
"
|
320 |
-
"
|
|
|
|
|
321 |
},
|
322 |
{
|
323 |
-
"epoch":
|
324 |
-
"learning_rate":
|
325 |
-
"loss": 1.
|
326 |
-
"step":
|
327 |
},
|
328 |
{
|
329 |
-
"epoch":
|
330 |
-
"
|
331 |
-
"
|
332 |
-
"
|
|
|
|
|
333 |
},
|
334 |
{
|
335 |
-
"epoch":
|
336 |
-
"learning_rate": 1.
|
337 |
-
"loss": 1.
|
338 |
-
"step":
|
339 |
},
|
340 |
{
|
341 |
-
"epoch":
|
342 |
-
"
|
343 |
-
"
|
344 |
-
"
|
|
|
|
|
345 |
},
|
346 |
{
|
347 |
-
"epoch":
|
348 |
-
"learning_rate": 1.
|
349 |
-
"loss": 1.
|
350 |
-
"step":
|
351 |
},
|
352 |
{
|
353 |
-
"epoch":
|
354 |
-
"
|
355 |
-
"
|
356 |
-
"
|
|
|
|
|
357 |
},
|
358 |
{
|
359 |
-
"epoch":
|
360 |
-
"learning_rate":
|
361 |
-
"loss": 1.
|
362 |
-
"step":
|
363 |
},
|
364 |
{
|
365 |
-
"epoch":
|
366 |
-
"
|
367 |
-
"
|
368 |
-
"
|
|
|
|
|
369 |
},
|
370 |
{
|
371 |
-
"epoch":
|
372 |
-
"learning_rate":
|
373 |
-
"loss": 1.
|
374 |
-
"step":
|
375 |
},
|
376 |
{
|
377 |
-
"epoch":
|
378 |
-
"
|
379 |
-
"
|
380 |
-
"
|
|
|
|
|
381 |
},
|
382 |
{
|
383 |
-
"epoch":
|
384 |
-
"learning_rate":
|
385 |
-
"loss": 1.
|
386 |
-
"step":
|
387 |
},
|
388 |
{
|
389 |
-
"epoch":
|
390 |
-
"
|
391 |
-
"
|
392 |
-
"
|
|
|
|
|
393 |
},
|
394 |
{
|
395 |
-
"epoch":
|
396 |
-
"learning_rate": 3.
|
397 |
-
"loss": 1.
|
398 |
-
"step":
|
399 |
},
|
400 |
{
|
401 |
-
"epoch":
|
402 |
-
"
|
403 |
-
"
|
404 |
-
"
|
|
|
|
|
405 |
},
|
406 |
{
|
407 |
-
"epoch":
|
408 |
-
"learning_rate": 1.
|
409 |
-
"loss": 1.
|
410 |
-
"step":
|
411 |
},
|
412 |
{
|
413 |
-
"epoch":
|
414 |
-
"
|
415 |
-
"
|
416 |
-
"
|
|
|
|
|
417 |
},
|
418 |
{
|
419 |
-
"epoch":
|
420 |
-
"step":
|
421 |
-
"total_flos":
|
422 |
-
"train_loss": 1.
|
423 |
-
"train_runtime":
|
424 |
-
"train_samples_per_second":
|
425 |
-
"train_steps_per_second": 3.
|
426 |
}
|
427 |
],
|
428 |
-
"max_steps":
|
429 |
-
"num_train_epochs":
|
430 |
-
"total_flos":
|
431 |
"trial_name": null,
|
432 |
"trial_params": null
|
433 |
}
|
|
|
1 |
{
|
2 |
+
"best_metric": 1.3029065132141113,
|
3 |
+
"best_model_checkpoint": "/content/pubmedbert-abstract-cord19/checkpoint-100000",
|
4 |
+
"epoch": 3.0,
|
5 |
+
"global_step": 102084,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
9 |
"log_history": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
{
|
11 |
"epoch": 0.15,
|
12 |
"learning_rate": 2.496e-05,
|
13 |
+
"loss": 1.3774,
|
14 |
"step": 5000
|
15 |
},
|
16 |
{
|
17 |
+
"epoch": 0.15,
|
18 |
+
"eval_loss": 1.3211560249328613,
|
19 |
+
"eval_runtime": 169.3169,
|
20 |
+
"eval_samples_per_second": 86.258,
|
21 |
+
"eval_steps_per_second": 10.785,
|
22 |
+
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
23 |
},
|
24 |
{
|
25 |
"epoch": 0.29,
|
26 |
+
"learning_rate": 4.9945000000000004e-05,
|
27 |
+
"loss": 1.3937,
|
28 |
"step": 10000
|
29 |
},
|
30 |
{
|
31 |
+
"epoch": 0.29,
|
32 |
+
"eval_loss": 1.4059454202651978,
|
33 |
+
"eval_runtime": 169.3069,
|
34 |
+
"eval_samples_per_second": 86.263,
|
35 |
+
"eval_steps_per_second": 10.785,
|
36 |
+
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
37 |
},
|
38 |
{
|
39 |
"epoch": 0.44,
|
40 |
+
"learning_rate": 4.7291603318709e-05,
|
41 |
+
"loss": 1.6812,
|
42 |
"step": 15000
|
43 |
},
|
44 |
{
|
45 |
+
"epoch": 0.44,
|
46 |
+
"eval_loss": 1.6174367666244507,
|
47 |
+
"eval_runtime": 168.9751,
|
48 |
+
"eval_samples_per_second": 86.433,
|
49 |
+
"eval_steps_per_second": 10.806,
|
50 |
+
"step": 15000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
},
|
52 |
{
|
53 |
+
"epoch": 0.59,
|
54 |
+
"learning_rate": 4.457723382998132e-05,
|
55 |
+
"loss": 1.4712,
|
56 |
+
"step": 20000
|
57 |
},
|
58 |
{
|
59 |
+
"epoch": 0.59,
|
60 |
+
"eval_loss": 1.4382678270339966,
|
61 |
+
"eval_runtime": 168.9508,
|
62 |
+
"eval_samples_per_second": 86.445,
|
63 |
+
"eval_steps_per_second": 10.808,
|
64 |
+
"step": 20000
|
65 |
},
|
66 |
{
|
67 |
+
"epoch": 0.73,
|
68 |
+
"learning_rate": 4.186340732374788e-05,
|
69 |
+
"loss": 1.4293,
|
70 |
+
"step": 25000
|
71 |
},
|
72 |
{
|
73 |
+
"epoch": 0.73,
|
74 |
+
"eval_loss": 1.4356446266174316,
|
75 |
+
"eval_runtime": 168.8828,
|
76 |
+
"eval_samples_per_second": 86.48,
|
77 |
+
"eval_steps_per_second": 10.812,
|
78 |
+
"step": 25000
|
79 |
},
|
80 |
{
|
81 |
+
"epoch": 0.88,
|
82 |
+
"learning_rate": 3.914849485252596e-05,
|
83 |
+
"loss": 1.4155,
|
84 |
+
"step": 30000
|
85 |
},
|
86 |
{
|
87 |
+
"epoch": 0.88,
|
88 |
+
"eval_loss": 1.4283361434936523,
|
89 |
+
"eval_runtime": 169.0831,
|
90 |
+
"eval_samples_per_second": 86.378,
|
91 |
+
"eval_steps_per_second": 10.799,
|
92 |
+
"step": 30000
|
93 |
},
|
94 |
{
|
95 |
+
"epoch": 1.03,
|
96 |
+
"learning_rate": 3.6435754311281005e-05,
|
97 |
+
"loss": 1.3963,
|
98 |
+
"step": 35000
|
99 |
},
|
100 |
{
|
101 |
+
"epoch": 1.03,
|
102 |
+
"eval_loss": 1.4134665727615356,
|
103 |
+
"eval_runtime": 169.088,
|
104 |
+
"eval_samples_per_second": 86.375,
|
105 |
+
"eval_steps_per_second": 10.799,
|
106 |
+
"step": 35000
|
107 |
},
|
108 |
{
|
109 |
+
"epoch": 1.18,
|
110 |
+
"learning_rate": 3.372138482255332e-05,
|
111 |
+
"loss": 1.3718,
|
112 |
+
"step": 40000
|
113 |
},
|
114 |
{
|
115 |
+
"epoch": 1.18,
|
116 |
+
"eval_loss": 1.3948187828063965,
|
117 |
+
"eval_runtime": 169.0271,
|
118 |
+
"eval_samples_per_second": 86.406,
|
119 |
+
"eval_steps_per_second": 10.803,
|
120 |
+
"step": 40000
|
121 |
},
|
122 |
{
|
123 |
+
"epoch": 1.32,
|
124 |
+
"learning_rate": 3.1006472351331396e-05,
|
125 |
+
"loss": 1.369,
|
126 |
+
"step": 45000
|
127 |
},
|
128 |
{
|
129 |
+
"epoch": 1.32,
|
130 |
+
"eval_loss": 1.3961154222488403,
|
131 |
+
"eval_runtime": 169.0308,
|
132 |
+
"eval_samples_per_second": 86.404,
|
133 |
+
"eval_steps_per_second": 10.803,
|
134 |
+
"step": 45000
|
135 |
},
|
136 |
{
|
137 |
+
"epoch": 1.47,
|
138 |
+
"learning_rate": 2.8292645845097955e-05,
|
139 |
+
"loss": 1.354,
|
140 |
+
"step": 50000
|
141 |
},
|
142 |
{
|
143 |
+
"epoch": 1.47,
|
144 |
+
"eval_loss": 1.378829836845398,
|
145 |
+
"eval_runtime": 169.1643,
|
146 |
+
"eval_samples_per_second": 86.336,
|
147 |
+
"eval_steps_per_second": 10.794,
|
148 |
+
"step": 50000
|
149 |
},
|
150 |
{
|
151 |
+
"epoch": 1.62,
|
152 |
+
"learning_rate": 2.5579362321358764e-05,
|
153 |
+
"loss": 1.3399,
|
154 |
+
"step": 55000
|
155 |
},
|
156 |
{
|
157 |
+
"epoch": 1.62,
|
158 |
+
"eval_loss": 1.3865987062454224,
|
159 |
+
"eval_runtime": 169.0688,
|
160 |
+
"eval_samples_per_second": 86.385,
|
161 |
+
"eval_steps_per_second": 10.8,
|
162 |
+
"step": 55000
|
163 |
},
|
164 |
{
|
165 |
+
"epoch": 1.76,
|
166 |
+
"learning_rate": 2.2866078797619567e-05,
|
167 |
+
"loss": 1.3289,
|
168 |
+
"step": 60000
|
169 |
},
|
170 |
{
|
171 |
+
"epoch": 1.76,
|
172 |
+
"eval_loss": 1.362976312637329,
|
173 |
+
"eval_runtime": 169.1136,
|
174 |
+
"eval_samples_per_second": 86.362,
|
175 |
+
"eval_steps_per_second": 10.797,
|
176 |
+
"step": 60000
|
177 |
},
|
178 |
{
|
179 |
+
"epoch": 1.91,
|
180 |
+
"learning_rate": 2.0151709308891882e-05,
|
181 |
+
"loss": 1.3155,
|
182 |
+
"step": 65000
|
183 |
},
|
184 |
{
|
185 |
+
"epoch": 1.91,
|
186 |
+
"eval_loss": 1.3609150648117065,
|
187 |
+
"eval_runtime": 169.0556,
|
188 |
+
"eval_samples_per_second": 86.392,
|
189 |
+
"eval_steps_per_second": 10.801,
|
190 |
+
"step": 65000
|
191 |
},
|
192 |
{
|
193 |
+
"epoch": 2.06,
|
194 |
+
"learning_rate": 1.743788280265844e-05,
|
195 |
+
"loss": 1.2976,
|
196 |
+
"step": 70000
|
197 |
},
|
198 |
{
|
199 |
+
"epoch": 2.06,
|
200 |
+
"eval_loss": 1.3489034175872803,
|
201 |
+
"eval_runtime": 169.1114,
|
202 |
+
"eval_samples_per_second": 86.363,
|
203 |
+
"eval_steps_per_second": 10.798,
|
204 |
+
"step": 70000
|
205 |
},
|
206 |
{
|
207 |
+
"epoch": 2.2,
|
208 |
+
"learning_rate": 1.4724599278919249e-05,
|
209 |
+
"loss": 1.2783,
|
210 |
+
"step": 75000
|
211 |
},
|
212 |
{
|
213 |
+
"epoch": 2.2,
|
214 |
+
"eval_loss": 1.3333380222320557,
|
215 |
+
"eval_runtime": 169.0576,
|
216 |
+
"eval_samples_per_second": 86.391,
|
217 |
+
"eval_steps_per_second": 10.801,
|
218 |
+
"step": 75000
|
219 |
},
|
220 |
{
|
221 |
+
"epoch": 2.35,
|
222 |
+
"learning_rate": 1.2009686807697321e-05,
|
223 |
+
"loss": 1.2696,
|
224 |
+
"step": 80000
|
225 |
},
|
226 |
{
|
227 |
+
"epoch": 2.35,
|
228 |
+
"eval_loss": 1.3259785175323486,
|
229 |
+
"eval_runtime": 168.9798,
|
230 |
+
"eval_samples_per_second": 86.43,
|
231 |
+
"eval_steps_per_second": 10.806,
|
232 |
+
"step": 80000
|
233 |
},
|
234 |
{
|
235 |
+
"epoch": 2.5,
|
236 |
+
"learning_rate": 9.29586030146388e-06,
|
237 |
+
"loss": 1.2607,
|
238 |
+
"step": 85000
|
239 |
},
|
240 |
{
|
241 |
+
"epoch": 2.5,
|
242 |
+
"eval_loss": 1.3232313394546509,
|
243 |
+
"eval_runtime": 168.7484,
|
244 |
+
"eval_samples_per_second": 86.549,
|
245 |
+
"eval_steps_per_second": 10.821,
|
246 |
+
"step": 85000
|
247 |
},
|
248 |
{
|
249 |
+
"epoch": 2.64,
|
250 |
+
"learning_rate": 6.582033795230443e-06,
|
251 |
+
"loss": 1.2547,
|
252 |
+
"step": 90000
|
253 |
},
|
254 |
{
|
255 |
+
"epoch": 2.64,
|
256 |
+
"eval_loss": 1.3033902645111084,
|
257 |
+
"eval_runtime": 168.7446,
|
258 |
+
"eval_samples_per_second": 86.551,
|
259 |
+
"eval_steps_per_second": 10.821,
|
260 |
+
"step": 90000
|
261 |
},
|
262 |
{
|
263 |
+
"epoch": 2.79,
|
264 |
+
"learning_rate": 3.867664306502758e-06,
|
265 |
+
"loss": 1.2495,
|
266 |
+
"step": 95000
|
267 |
},
|
268 |
{
|
269 |
+
"epoch": 2.79,
|
270 |
+
"eval_loss": 1.3035175800323486,
|
271 |
+
"eval_runtime": 168.7573,
|
272 |
+
"eval_samples_per_second": 86.544,
|
273 |
+
"eval_steps_per_second": 10.82,
|
274 |
+
"step": 95000
|
275 |
},
|
276 |
{
|
277 |
+
"epoch": 2.94,
|
278 |
+
"learning_rate": 1.1538378002693194e-06,
|
279 |
+
"loss": 1.2404,
|
280 |
+
"step": 100000
|
281 |
},
|
282 |
{
|
283 |
+
"epoch": 2.94,
|
284 |
+
"eval_loss": 1.3029065132141113,
|
285 |
+
"eval_runtime": 168.8016,
|
286 |
+
"eval_samples_per_second": 86.522,
|
287 |
+
"eval_steps_per_second": 10.817,
|
288 |
+
"step": 100000
|
289 |
},
|
290 |
{
|
291 |
+
"epoch": 3.0,
|
292 |
+
"step": 102084,
|
293 |
+
"total_flos": 1.04954332801725e+17,
|
294 |
+
"train_loss": 1.35220570928797,
|
295 |
+
"train_runtime": 32542.0864,
|
296 |
+
"train_samples_per_second": 25.095,
|
297 |
+
"train_steps_per_second": 3.137
|
298 |
}
|
299 |
],
|
300 |
+
"max_steps": 102084,
|
301 |
+
"num_train_epochs": 3,
|
302 |
+
"total_flos": 1.04954332801725e+17,
|
303 |
"trial_name": null,
|
304 |
"trial_params": null
|
305 |
}
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3055
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b2b18b7fbbb222cc58a7279e9f20f061015602730e5f2e94504e9367dc4156be
|
3 |
size 3055
|