Spaces:
Sleeping
Sleeping
fix runtime error
Browse files- Dockerfile +2 -18
- app.py +16 -56
- requirements.txt +3 -5
Dockerfile
CHANGED
@@ -1,27 +1,11 @@
|
|
1 |
-
# Use lightweight Python image
|
2 |
FROM python:3.10-slim
|
3 |
|
4 |
-
# Prevent interactive prompts during package install
|
5 |
-
ENV DEBIAN_FRONTEND=noninteractive
|
6 |
-
|
7 |
-
# Set working directory
|
8 |
WORKDIR /app
|
9 |
|
10 |
-
# Install basic dependencies
|
11 |
-
RUN apt-get update && apt-get install -y \
|
12 |
-
git \
|
13 |
-
&& rm -rf /var/lib/apt/lists/*
|
14 |
-
|
15 |
-
# Copy requirements and install Python packages
|
16 |
COPY requirements.txt .
|
17 |
-
RUN pip install --no-cache-dir --upgrade pip \
|
18 |
-
&& pip install --no-cache-dir -r requirements.txt
|
19 |
|
20 |
-
|
21 |
-
COPY app.py .
|
22 |
|
23 |
-
|
24 |
-
EXPOSE 7860
|
25 |
|
26 |
-
# Run the app
|
27 |
CMD ["python", "app.py"]
|
|
|
|
|
1 |
FROM python:3.10-slim
|
2 |
|
|
|
|
|
|
|
|
|
3 |
WORKDIR /app
|
4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
COPY requirements.txt .
|
|
|
|
|
6 |
|
7 |
+
RUN pip install --no-cache-dir -r requirements.txt
|
|
|
8 |
|
9 |
+
COPY . .
|
|
|
10 |
|
|
|
11 |
CMD ["python", "app.py"]
|
app.py
CHANGED
@@ -1,64 +1,24 @@
|
|
1 |
import gradio as gr
|
2 |
-
from transformers import
|
3 |
-
import torch
|
4 |
-
from fastapi import FastAPI, Request
|
5 |
-
import uvicorn
|
6 |
|
7 |
-
|
|
|
|
|
8 |
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
15 |
-
model.to(device)
|
16 |
-
|
17 |
-
# -------- Inference function --------
|
18 |
-
def generate_text(prompt, max_tokens=100, temperature=0.9, top_p=0.95):
|
19 |
-
inputs = tokenizer(prompt, return_tensors="pt").to(device)
|
20 |
-
outputs = model.generate(
|
21 |
-
**inputs,
|
22 |
-
max_new_tokens=max_tokens,
|
23 |
do_sample=True,
|
24 |
temperature=temperature,
|
25 |
-
top_p=
|
26 |
-
pad_token_id=tokenizer.eos_token_id,
|
27 |
)
|
28 |
-
return tokenizer.decode(
|
29 |
-
|
30 |
-
# -------- Gradio UI --------
|
31 |
-
def launch_gradio():
|
32 |
-
with gr.Blocks() as demo:
|
33 |
-
gr.Markdown("## 🇻🇳 VietAI GPT-Neo 1.3B - Sinh văn bản tiếng Việt")
|
34 |
-
prompt = gr.Textbox(label="Prompt", placeholder="Nhập đoạn mở đầu văn bản...")
|
35 |
-
max_tokens = gr.Slider(10, 200, value=100, label="Số tokens sinh ra")
|
36 |
-
temperature = gr.Slider(0.1, 1.5, value=0.9, label="Temperature")
|
37 |
-
top_p = gr.Slider(0.1, 1.0, value=0.95, label="Top-p sampling")
|
38 |
-
output = gr.Textbox(label="Kết quả", lines=10)
|
39 |
-
btn = gr.Button("Sinh văn bản")
|
40 |
-
btn.click(fn=generate_text, inputs=[prompt, max_tokens, temperature, top_p], outputs=output)
|
41 |
-
|
42 |
-
demo.launch(server_name="0.0.0.0", server_port=7860, share=False)
|
43 |
-
|
44 |
-
# -------- FastAPI for REST API --------
|
45 |
-
app = FastAPI()
|
46 |
-
|
47 |
-
@app.post("/generate")
|
48 |
-
async def generate(request: Request):
|
49 |
-
body = await request.json()
|
50 |
-
prompt = body.get("prompt", "")
|
51 |
-
max_tokens = body.get("max_tokens", 100)
|
52 |
-
temperature = body.get("temperature", 0.9)
|
53 |
-
top_p = body.get("top_p", 0.95)
|
54 |
-
output = generate_text(prompt, max_tokens, temperature, top_p)
|
55 |
-
return {"response": output}
|
56 |
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
thread.start()
|
62 |
|
63 |
-
|
64 |
-
uvicorn.run(app, host="0.0.0.0", port=8000)
|
|
|
1 |
import gradio as gr
|
2 |
+
from transformers import GPT2Tokenizer, GPT2LMHeadModel
|
|
|
|
|
|
|
3 |
|
4 |
+
model_id = "NlpHUST/gpt2-vietnamese"
|
5 |
+
tokenizer = GPT2Tokenizer.from_pretrained(model_id)
|
6 |
+
model = GPT2LMHeadModel.from_pretrained(model_id)
|
7 |
|
8 |
+
def generate(prompt, max_length=100, temperature=0.8):
|
9 |
+
input_ids = tokenizer.encode(prompt, return_tensors="pt")
|
10 |
+
output = model.generate(
|
11 |
+
input_ids,
|
12 |
+
max_length=max_length,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
do_sample=True,
|
14 |
temperature=temperature,
|
15 |
+
top_p=0.95
|
|
|
16 |
)
|
17 |
+
return tokenizer.decode(output[0], skip_special_tokens=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
|
19 |
+
demo = gr.Interface(fn=generate,
|
20 |
+
inputs=["text", gr.Slider(50, 200, value=100), gr.Slider(0.1, 1.0, value=0.8)],
|
21 |
+
outputs="text",
|
22 |
+
title="Sinh văn bản tiếng Việt bằng GPT-2")
|
|
|
23 |
|
24 |
+
demo.launch()
|
|
requirements.txt
CHANGED
@@ -1,5 +1,3 @@
|
|
1 |
-
transformers
|
2 |
-
torch
|
3 |
-
gradio
|
4 |
-
uvicorn
|
5 |
-
fastapi
|
|
|
1 |
+
transformers==4.39.3
|
2 |
+
torch==2.1.2
|
3 |
+
gradio==4.27.0
|
|
|
|