Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,36 @@
|
|
1 |
-
---
|
2 |
-
license: apache-2.0
|
3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
---
|
4 |
+
```python
|
5 |
+
from transformers import AutoTokenizer,GenerationConfig,AutoConfig,GenerationConfig,AutoModelForCausalLM
|
6 |
+
import torch
|
7 |
+
from transformers import AutoTokenizer
|
8 |
+
model_path = "model_path"
|
9 |
+
model = AutoModelForCausalLM.from_pretrained(model_path,trust_remote_code = True)
|
10 |
+
tokenizer = AutoTokenizer.from_pretrained(model_path)
|
11 |
+
|
12 |
+
_inputs = """your input"""
|
13 |
+
|
14 |
+
gen_conf = GenerationConfig(
|
15 |
+
num_beams=1,
|
16 |
+
do_sample=False,
|
17 |
+
max_length=128,
|
18 |
+
max_new_tokens=128,
|
19 |
+
no_repeat_ngram_size=4,
|
20 |
+
eos_token_id=tokenizer.eos_token_id,
|
21 |
+
pad_token_id=tokenizer.pad_token_id,
|
22 |
+
)
|
23 |
+
|
24 |
+
tokend = tokenizer.encode_plus(text=_inputs, add_special_tokens=False)
|
25 |
+
input_ids_lengths = len(tokend.input_ids)
|
26 |
+
print(_inputs)
|
27 |
+
input_ids = torch.LongTensor([tokend.input_ids])
|
28 |
+
|
29 |
+
outputs = model.generate(
|
30 |
+
inputs=input_ids,
|
31 |
+
generation_config=gen_conf,
|
32 |
+
)
|
33 |
+
|
34 |
+
outs = tokenizer.decode(outputs[0][input_ids_lengths:],skip_special_tokens=False)
|
35 |
+
print(outs)
|
36 |
+
|