Update README.md
Browse files
README.md
CHANGED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
### t5-base finetuned on xsum dataset
|
| 2 |
+
|
| 3 |
+
#### train args<br>
|
| 4 |
+
max_input_length: 512<br>
|
| 5 |
+
max_tgt_length: 128<br>
|
| 6 |
+
epoch: 3<br>
|
| 7 |
+
optimizer: AdamW<br>
|
| 8 |
+
lr: 2e-5<br>
|
| 9 |
+
weight_decay: 1e-3<br>
|
| 10 |
+
fp16: False<br>
|
| 11 |
+
prefix: "summarize: "<br>
|
| 12 |
+
|
| 13 |
+
#### performance<br>
|
| 14 |
+
train_loss 0.5976<br>
|
| 15 |
+
eval_loss: 0.5340<br>
|
| 16 |
+
eval_rouge1: 34.6791<br>
|
| 17 |
+
eval_rouge2: 12.8236<br>
|
| 18 |
+
eval_rougeL: 28.1201<br>
|
| 19 |
+
eval_rougeLsum: 28.1241<br>
|
| 20 |
+
|
| 21 |
+
|
| 22 |
+
#### usage<br>
|
| 23 |
+
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM<br>
|
| 24 |
+
|
| 25 |
+
#### dependency<br>
|
| 26 |
+
trained with transformers==4.24<br>
|
| 27 |
+
compatible with transformers==3.0.2<br>
|