Upload app.py with huggingface_hub

app.py

@@ -1,64 +1,137 @@
+import os
+import zipfile
+import torch
+import faiss
+import numpy as np
 import gradio as gr
-from huggingface_hub import InferenceClient
-
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
+
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+from sentence_transformers import SentenceTransformer
+from langchain.document_loaders import TextLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.embeddings import HuggingFaceEmbeddings
+from langchain.vectorstores import FAISS as LangChainFAISS
+from langchain.docstore import InMemoryDocstore
+from langchain.schema import Document
+from langchain.llms import HuggingFacePipeline
+from huggingface_hub import login
+from huggingface_hub import upload_file
+
+# Extract the Knowledge Base ZIP
+if os.path.exists("md_knowledge_base.zip"):
+    with zipfile.ZipFile("md_knowledge_base.zip", "r") as zip_ref:
+        zip_ref.extractall("md_knowledge_base")
+    print("✅ Knowledge base extracted.")
+
+# Load Markdown Files
+KB_PATH = "md_knowledge_base"
+files = [os.path.join(dp, f) for dp, _, fn in os.walk(KB_PATH) for f in fn if f.endswith(".md")]
+docs = [doc for f in files for doc in TextLoader(f, encoding="utf-8").load()]
+print(f"✅ Loaded {len(docs)} documents.")
+
+# Chunking
+def get_dynamic_chunk_size(text):
+    if len(text) < 1000:
+        return 300
+    elif len(text) < 5000:
+        return 500
+    else:
+        return 1000
+
+chunks = []
+for doc in docs:
+    chunk_size = get_dynamic_chunk_size(doc.page_content)
+    chunk_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=100)
+    chunks.extend(chunk_splitter.split_documents([doc]))
+texts = [chunk.page_content for chunk in chunks]
+
+# Vectorstore (FAISS)
+embed_model_id = "sentence-transformers/all-MiniLM-L6-v2"
+embedder = SentenceTransformer(embed_model_id)
+embeddings = embedder.encode(texts, show_progress_bar=False)
+
+dim = embeddings.shape[1]
+index = faiss.IndexFlatL2(dim)
+index.add(np.array(embeddings, dtype="float32"))
+
+docs = [Document(page_content=t) for t in texts]
+docstore = InMemoryDocstore({str(i): docs[i] for i in range(len(docs))})
+id_map = {i: str(i) for i in range(len(docs))}
+embed_fn = HuggingFaceEmbeddings(model_name=embed_model_id)
+
+vectorstore = LangChainFAISS(
+    index=index,
+    docstore=docstore,
+    index_to_docstore_id=id_map,
+    embedding_function=embed_fn
 )
 
+print("✅ FAISS vectorstore ready.")
+
+# Load Falcon-e-1B-Instruct
+model_id = "tiiuae/falcon-e-1b-instruct"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    torch_dtype=torch.bfloat16
+).to("cuda" if torch.cuda.is_available() else "cpu")
+
+text_gen_pipeline = pipeline(
+    "text-generation",
+    model=model,
+    tokenizer=tokenizer,
+    torch_dtype=torch.bfloat16,
+    device=0 if torch.cuda.is_available() else -1,
+    return_full_text=False,
+    do_sample=False,
+    max_new_tokens=200,
+    pad_token_id=tokenizer.eos_token_id
+)
+
+llm = HuggingFacePipeline(pipeline=text_gen_pipeline)
+
+def truncate_context(context, max_length=1024):
+    tokens = tokenizer.encode(context)
+    if len(tokens) > max_length:
+        tokens = tokens[:max_length]
+    return tokenizer.decode(tokens, skip_special_tokens=True)
+
+def format_prompt(context, question):
+    return (
+        "You are the Hull University Assistant—a friendly, knowledgeable chatbot dedicated to "
+        "helping students with questions about courses, admissions, tuition fees, and student life. "
+        "Use ONLY the information provided in the context below to answer the question. "
+        "If the answer cannot be found in the context, reply: \"I’m sorry, but I don’t have that "
+        "information available right now.\"\n\n"
+        f"Context:\n{truncate_context(context)}\n\n"
+        f"Student Question: {question}\n"
+        "Assistant Answer:"
+    )
+
+def answer_fn(question):
+    docs = vectorstore.similarity_search(question, k=5)
+    if not docs:
+        return "I'm sorry, I couldn't find any relevant information for your query."
+    context = "\n\n".join(d.page_content for d in docs)
+    prompt = format_prompt(context, question)
+    try:
+        response = llm.invoke(prompt).strip()
+        return response
+    except Exception as e:
+        return f"An error occurred: {e}"
+
+# Gradio Interface
+def chat_fn(user_message, history):
+    bot_response = answer_fn(user_message)
+    history = history + [(user_message, bot_response)]
+    return history, history
+
+with gr.Blocks() as demo:
+    gr.Markdown("## 📘 University of Hull Assistant")
+    chatbot = gr.Chatbot()
+    state = gr.State([])
+    user_input = gr.Textbox(placeholder="Ask a question about University of Hull...", show_label=False)
+
+    user_input.submit(fn=chat_fn, inputs=[user_input, state], outputs=[chatbot, state])
 
-if __name__ == "__main__":
-    demo.launch()
+demo.launch()