Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,31 @@
|
|
1 |
---
|
2 |
license: mit
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: mit
|
3 |
---
|
4 |
+
|
5 |
+
# How to use
|
6 |
+
```python3
|
7 |
+
from transformers import MT5Tokenizer, MT5ForConditionalGeneration
|
8 |
+
|
9 |
+
tokenizer = MT5Tokenizer.from_pretrained('/Users/siwaboon/Desktop/Artifacts/thaisum/checkpoint-21000')
|
10 |
+
model = MT5ForConditionalGeneration.from_pretrained('/Users/siwaboon/Desktop/Artifacts/thaisum/checkpoint-21000')
|
11 |
+
|
12 |
+
text = "some news with head line"
|
13 |
+
|
14 |
+
tokenized_text = tokenizer(text, truncation=True, padding=True, return_tensors='pt')
|
15 |
+
|
16 |
+
source_ids = tokenized_text['input_ids'].to("cpu", dtype = torch.long)
|
17 |
+
source_mask = tokenized_text['attention_mask'].to("cpu", dtype = torch.long)
|
18 |
+
|
19 |
+
generated_ids = model.generate(
|
20 |
+
input_ids = source_ids,
|
21 |
+
attention_mask = source_mask,
|
22 |
+
max_length=512,
|
23 |
+
num_beams=5,
|
24 |
+
repetition_penalty=1,
|
25 |
+
length_penalty=1,
|
26 |
+
early_stopping=True,
|
27 |
+
no_repeat_ngram_size=2
|
28 |
+
)
|
29 |
+
|
30 |
+
pred = tokenizer.decode(generated_ids[0], skip_special_tokens=True, clean_up_tokenization_spaces=True)
|
31 |
+
```
|