Spaces:
Sleeping
Sleeping
Delete chat-qwen.py
Browse files- chat-qwen.py +0 -33
chat-qwen.py
DELETED
@@ -1,33 +0,0 @@
|
|
1 |
-
from transformers import AutoModelForCausalLM, AutoTokenizer
|
2 |
-
|
3 |
-
def get_response(prompt: str):
|
4 |
-
model = AutoModelForCausalLM.from_pretrained(
|
5 |
-
"Qwen/Qwen2-7B-Instruct-GPTQ-Int4",
|
6 |
-
torch_dtype="auto",
|
7 |
-
device_map="auto",
|
8 |
-
)
|
9 |
-
|
10 |
-
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-7B-Instruct-GPTQ-Int4")
|
11 |
-
|
12 |
-
prompt = "Give me a short introduction to large language model."
|
13 |
-
messages = [
|
14 |
-
{"role": "system", "content": "You are a helpful assistant."},
|
15 |
-
{"role": "user", "content": prompt},
|
16 |
-
]
|
17 |
-
text = tokenizer.apply_chat_template(
|
18 |
-
messages,
|
19 |
-
tokenize=False,
|
20 |
-
add_generation_prompt=True,
|
21 |
-
)
|
22 |
-
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
|
23 |
-
|
24 |
-
generated_ids = model.generate(
|
25 |
-
**model_inputs,
|
26 |
-
max_new_tokens=512,
|
27 |
-
)
|
28 |
-
generated_ids = [
|
29 |
-
output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
|
30 |
-
]
|
31 |
-
|
32 |
-
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
33 |
-
return response
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|