Update app.py
Browse files
app.py
CHANGED
@@ -1,5 +1,6 @@
|
|
1 |
import gradio as gr
|
2 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|
|
3 |
|
4 |
model_name = "Qwen/Qwen2.5-0.5B-Instruct"
|
5 |
|
@@ -8,24 +9,36 @@ model = AutoModelForCausalLM.from_pretrained(
|
|
8 |
model_name, low_cpu_mem_usage=True, device_map="auto", torch_dtype="auto"
|
9 |
)
|
10 |
|
11 |
-
def predict(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
text = tokenizer.apply_chat_template(
|
13 |
messages, tokenize=False, add_generation_prompt=True
|
14 |
)
|
15 |
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
|
16 |
|
17 |
-
logger.info(f"Model generation process started at - {process_id}")
|
18 |
generated_ids = model.generate(**model_inputs, max_new_tokens=512)
|
19 |
generated_ids = [
|
20 |
-
output_ids[len(input_ids)
|
21 |
for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
|
22 |
]
|
23 |
-
logger.info(f"Model generation process completed [{process_id}]")
|
24 |
|
25 |
reply = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
26 |
-
|
|
|
27 |
|
28 |
-
|
|
|
|
|
|
|
29 |
|
30 |
-
|
31 |
-
iface.launch(server_name="0.0.0.0", server_port=7860, share=False)
|
|
|
1 |
import gradio as gr
|
2 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
3 |
+
import torch
|
4 |
|
5 |
model_name = "Qwen/Qwen2.5-0.5B-Instruct"
|
6 |
|
|
|
9 |
model_name, low_cpu_mem_usage=True, device_map="auto", torch_dtype="auto"
|
10 |
)
|
11 |
|
12 |
+
def predict(history):
|
13 |
+
"""
|
14 |
+
history: list of [user, bot] message pairs from the Chatbot
|
15 |
+
"""
|
16 |
+
# Convert history into the 'messages' format for chat template
|
17 |
+
messages = []
|
18 |
+
for human, bot in history:
|
19 |
+
if human:
|
20 |
+
messages.append({"role": "user", "content": human})
|
21 |
+
if bot:
|
22 |
+
messages.append({"role": "assistant", "content": bot})
|
23 |
+
|
24 |
text = tokenizer.apply_chat_template(
|
25 |
messages, tokenize=False, add_generation_prompt=True
|
26 |
)
|
27 |
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
|
28 |
|
|
|
29 |
generated_ids = model.generate(**model_inputs, max_new_tokens=512)
|
30 |
generated_ids = [
|
31 |
+
output_ids[len(input_ids):]
|
32 |
for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
|
33 |
]
|
|
|
34 |
|
35 |
reply = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
36 |
+
history.append((messages[-1]["content"] if messages else "", reply))
|
37 |
+
return history
|
38 |
|
39 |
+
with gr.Blocks() as server:
|
40 |
+
chatbot = gr.Chatbot()
|
41 |
+
msg = gr.Textbox(placeholder="Type your message here...")
|
42 |
+
msg.submit(predict, [chatbot], chatbot)
|
43 |
|
44 |
+
server.launch(server_name="0.0.0.0", server_port=7860, share=False)
|
|