Upload 2 files

app.py

@@ -0,0 +1,131 @@
+import logging
+import gradio as gr
+import torch
+from langchain_chroma import Chroma
+from langchain_huggingface import HuggingFaceEmbeddings
+from langchain_community.llms import HuggingFacePipeline
+from langchain.chains import RetrievalQA
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+
+logging.basicConfig(level=logging.INFO)
+
+# 1. Load vector store
+embedding_model = HuggingFaceEmbeddings(
+    model_name="sentence-transformers/paraphrase-multilingual-mpnet-base-v2"
+)
+vector_store = Chroma(
+    persist_directory="vector_store",
+    embedding_function=embedding_model,
+)
+
+# 2. Load lightweight LLM (Phi-2)
+model_id = "microsoft/phi-2"
+tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+    device_map="auto",
+)
+gen_pipe = pipeline(
+    task="text-generation",
+    model=model,
+    tokenizer=tokenizer,
+    max_new_tokens=256,
+    temperature=0.5,
+    top_p=0.9,
+    do_sample=True,
+)
+llm = HuggingFacePipeline(pipeline=gen_pipe)
+
+# 3. Build RAG QA chain
+retriever = vector_store.as_retriever(search_kwargs={"k": 3})
+qa_chain = RetrievalQA.from_chain_type(
+    llm=llm,
+    chain_type="stuff",
+    retriever=retriever,
+)
+
+# 4. Business functions
+def simple_qa(user_query: str) -> str:
+    if not user_query.strip():
+        return "⚠️ 请输入学习问题，例如：什么是定积分？"
+    try:
+        return qa_chain.run(user_query)
+    except Exception as e:
+        logging.exception("问答失败：%s", e)
+        return f"⚠️ 问答失败，请稍后再试。\n[调试信息] {e}"
+
+def generate_outline(topic: str):
+    if not topic.strip():
+        yield "⚠️ 请输入章节或主题，例如：高等数学 第六章 定积分", ""
+        return
+
+    yield "⌛ 正在检索/生成，请稍候…", ""
+
+    try:
+        docs = retriever.get_relevant_documents(topic)
+        if not docs:
+            yield "⚠️ 没有找到相关内容，请换个关键词试试。", ""
+            return
+
+        snippet = "\n".join(d.page_content for d in docs)
+        prompt = (
+            f"根据以下内容，为“{topic}”生成大学本科层次的结构化学习大纲，格式示例：\n"
+            f"一、章节标题\n  1. 节标题\n    （1）要点描述\n...\n\n"
+            f"文档内容：\n{snippet}\n\n学习大纲："
+        )
+        raw = gen_pipe(prompt, max_new_tokens=512)[0]["generated_text"]
+        outline = raw.split("学习大纲：")[-1].strip()
+        yield outline, snippet
+    except Exception as e:
+        logging.exception("大纲生成失败：%s", e)
+        yield "⚠️ 抱歉，生成失败，请稍后再试。", ""
+
+def placeholder_fn(*args, **kwargs):
+    return "功能尚未实现，请等待后续更新。"
+
+# 5. Gradio UI
+with gr.Blocks(title="智能学习助手", theme=gr.themes.Base()) as demo:
+    gr.Markdown("# 📚 智能学习助手 v2.0\n— 专业课向量问答与大纲生成 —")
+
+    with gr.Tabs():
+        # Chat tab
+        with gr.TabItem("💬 智能问答"):
+            chatbot = gr.Chatbot(show_label=False, height=400)
+            user_msg = gr.Textbox(placeholder="输入学习问题", show_label=False)
+            send_btn = gr.Button("发送", variant="primary")
+
+            def chat_flow(message, history):
+                history.append((message, "🤔 正在思考中，请稍后…"))
+                yield "", history
+                ans = simple_qa(message)
+                history[-1] = (message, ans)
+                yield "", history
+
+            send_btn.click(chat_flow, [user_msg, chatbot], [user_msg, chatbot])
+            user_msg.submit(chat_flow, [user_msg, chatbot], [user_msg, chatbot])
+
+        # Outline tab
+        with gr.TabItem("📝 生成学习大纲"):
+            topic_in = gr.Textbox(label="章节主题", placeholder="例如：定积分")
+            outline_out = gr.Textbox(label="系统生成的大纲", lines=12)
+            snippet_out = gr.Textbox(label="[调试] 检索片段", lines=6, visible=False)
+            gen_btn = gr.Button("生成大纲", variant="primary")
+            gen_btn.click(generate_outline, inputs=topic_in, outputs=[outline_out, snippet_out])
+
+        # Placeholder tabs
+        with gr.TabItem("❓ 自动出题"):
+            gr.Textbox(label="知识点").render()
+            gr.Dropdown(["简单", "中等", "困难"], label="难度").render()
+            gr.Slider(1, 10, step=1, label="题目数量").render()
+            gr.Button("开始出题").click(placeholder_fn, [], [])
+
+        with gr.TabItem("✅ 答案批改"):
+            gr.Textbox(label="标准答案", lines=4).render()
+            gr.Textbox(label="学生答案", lines=4).render()
+            gr.Button("开始批改").click(placeholder_fn, [], [])
+
+    gr.Markdown("---\n模型：Phi-2 + 向量库检索 | Powered by Hugging Face Spaces")
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,9 @@
+langchain>=0.2.9
+langchain-huggingface>=0.0.6
+langchain-chroma>=0.0.6
+chromadb>=0.4.24
+transformers>=4.40.0
+accelerate
+torch>=2.1.0
+gradio>=4.24.0
+sentence-transformers