shunxing1234
commited on
Commit
•
e77ef02
1
Parent(s):
265f599
Update README.md
Browse files
README.md
CHANGED
@@ -35,9 +35,32 @@ We will continue to release improved versions of Aquila model as open source. Fo
|
|
35 |
|
36 |
### 1. Inference
|
37 |
|
|
|
|
|
|
|
|
|
38 |
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
|
|
|
40 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
|
42 |
|
43 |
## License
|
|
|
35 |
|
36 |
### 1. Inference
|
37 |
|
38 |
+
```python
|
39 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
40 |
+
import torch
|
41 |
+
from cyg_conversation import covert_prompt_to_input_ids_with_history
|
42 |
|
43 |
+
tokenizer = AutoTokenizer.from_pretrained("BAAI/AquilaChat-7B")
|
44 |
+
model = AutoModelForCausalLM.from_pretrained("BAAI/AquilaChat-7B")
|
45 |
+
model.eval()
|
46 |
+
model.to("cuda:0")
|
47 |
+
vocab = tokenizer.vocab
|
48 |
+
print(len(vocab))
|
49 |
|
50 |
+
text = "请给出10个要到北京旅游的理由。"
|
51 |
|
52 |
+
tokens = covert_prompt_to_input_ids_with_history(text, history=[], tokenizer=tokenizer, max_token=512)
|
53 |
+
|
54 |
+
tokens = torch.tensor(tokens)[None,].to("cuda:0")
|
55 |
+
|
56 |
+
|
57 |
+
with torch.no_grad():
|
58 |
+
out = model.generate(tokens, do_sample=True, max_length=512, eos_token_id=100007)[0]
|
59 |
+
|
60 |
+
out = tokenizer.decode(out.cpu().numpy().tolist())
|
61 |
+
|
62 |
+
print(out)
|
63 |
+
```
|
64 |
|
65 |
|
66 |
## License
|