HuggingFace Space deployment

Dockerfile

@@ -1,21 +1,17 @@
+# 使用官方 Python 镜像
 FROM python:3.9-slim
 
+# 设置工作目录
 WORKDIR /app
 
-RUN apt-get update && apt-get install -y \
-    build-essential \
-    curl \
-    software-properties-common \
-    git \
-    && rm -rf /var/lib/apt/lists/*
+# 拷贝依赖和代码
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
 
-COPY requirements.txt ./
-COPY src/ ./src/
+COPY . .
 
-RUN pip3 install -r requirements.txt
+# 设置默认端口
+ENV PORT=7860
 
-EXPOSE 8501
-
-HEALTHCHECK CMD curl --fail http://localhost:8501/_stcore/health
-
-ENTRYPOINT ["streamlit", "run", "src/streamlit_app.py", "--server.port=8501", "--server.address=0.0.0.0"]
+# 启动 FastAPI 服务
+CMD ["./start.sh"]

app.py

@@ -0,0 +1,47 @@
+import os
+from transformers import AutoTokenizer, AutoModelForCausalLM, T5Tokenizer, T5ForConditionalGeneration
+from peft import PeftModel
+from fastapi import FastAPI
+from pydantic import BaseModel
+from huggingface_hub import login
+
+login(token=os.getenv("HF_TOKEN"))
+print("Hugging Face Successfully Login!")
+
+app = FastAPI()
+
+# Load fine-tuned model and tokenizer
+# tokenizer = AutoTokenizer.from_pretrained("./llama2-7b", local_files_only=True)
+# base_model = AutoModelForCausalLM.from_pretrained("./llama2-7b", local_files_only=True)
+# model = PeftModel.from_pretrained(base_model, "./checkpoint-5400", local_files_only=True)
+
+
+tokenizer = T5Tokenizer.from_pretrained("google/flan-t5-small")
+model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-small")
+
+# Define data structure of parameters
+class PromptInput(BaseModel):
+    prompt: str
+
+# define API interface
+# @app.post("/generate")
+# def generate_script(input: PromptInput):
+#     print("Starts Generating!")
+#     inputs = tokenizer(input.prompt, return_tensors="pt")
+#     print("Inputs Tokenized! Generating Begins~")
+#     outputs = model.generate(**inputs, max_new_tokens=200)
+#     print("Generating Succeed!")
+#     result = tokenizer.decode(outputs[0], skip_special_tokens=True)
+#     print("Results formed!")
+#     return {"generated_script": result}
+
+@app.post("/generate")
+def generate_script(input: PromptInput):
+    print("Starts Generating!")
+    inputs = tokenizer(input.prompt, return_tensors="pt").input_ids
+    print("Inputs Tokenized! Generating Begins~")
+    outputs = model.generate(inputs)
+    print("Generating Succeed!")
+    result = tokenizer.decode(outputs[0])
+    print("Results formed!")
+    return {"generated_script": result}

requirements.txt

@@ -1,3 +1,45 @@
-altair
-pandas
+accelerate==1.6.0
+annotated-types==0.7.0
+anyio==4.9.0
+certifi==2025.4.26
+charset-normalizer==3.4.2
+click==8.1.8
+exceptiongroup==1.2.2
+fastapi==0.115.12
+filelock==3.18.0
+fsspec==2025.3.2
+h11==0.16.0
+httptools==0.6.4
+huggingface-hub==0.30.2
+idna==3.10
+Jinja2==3.1.6
+MarkupSafe==3.0.2
+mpmath==1.3.0
+networkx==3.2.1
+numpy==2.0.2
+packaging==25.0
+peft==0.15.2
+psutil==7.0.0
+pydantic==2.11.4
+pydantic_core==2.33.2
+python-dotenv==1.1.0
+PyYAML==6.0.2
+regex==2024.11.6
+requests==2.32.3
+safetensors==0.5.3
+sniffio==1.3.1
+starlette==0.46.2
+sympy==1.14.0
+tokenizers==0.21.1
+torch==2.7.0
+tqdm==4.67.1
+transformers==4.51.3
+typing-inspection==0.4.0
+typing_extensions==4.13.2
+urllib3==2.4.0
+uvicorn==0.34.2
+uvloop==0.21.0
+watchfiles==1.0.5
+websockets==15.0.1
+sentencepiece
 streamlit

start.sh

@@ -0,0 +1,5 @@
+#!/bin/zsh
+
+uvicorn app:app --host 0.0.0.0 --port 8000 &
+
+streamlit run ./streamlit_app.py --server.port 7860 --server.enableCORS false

streamlit_app.py

@@ -0,0 +1,18 @@
+import streamlit as st
+import requests
+
+st.title("LLaMA2 TV Script Generator")
+prompt = st.text_area("Enter your prompt:")
+
+if st.button("Generate"):
+    with st.spinner("Generating..."):
+        response = requests.post(
+            "http://localhost:8000/generate",
+            json={"prompt": prompt}
+        )
+
+        if response.ok:
+            st.markdown("### Output")
+            st.write(response.json()["generated_script"])
+        else:
+            st.error("Something went wrong.")