Upload app.py with huggingface_hub

app.py

@@ -4,7 +4,8 @@ import torch
 import faiss
 import numpy as np
 import gradio as gr
-from transformers import GPT2Tokenizer, AutoTokenizer, AutoModelForCausalLM, pipeline
+
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
 from sentence_transformers import SentenceTransformer
 from langchain.document_loaders import TextLoader
 from langchain.text_splitter import RecursiveCharacterTextSplitter
@@ -13,20 +14,22 @@ 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
 
-# === 1. Extract the Knowledge Base ZIP ===
+# 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.")
 
-# === 2. Load Markdown Files ===
+# 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.")
 
-# === 3. Chunking ===
+# Chunking
 def get_dynamic_chunk_size(text):
     if len(text) < 1000:
         return 300
@@ -42,8 +45,8 @@ for doc in docs:
     chunks.extend(chunk_splitter.split_documents([doc]))
 texts = [chunk.page_content for chunk in chunks]
 
-# === 4. Vectorstore (FAISS) ===
-embed_model_id = "distilbert-base-uncased"
+# Vectorstore (FAISS)
+embed_model_id = "sentence-transformers/all-MiniLM-L6-v2"
 embedder = SentenceTransformer(embed_model_id)
 embeddings = embedder.encode(texts, show_progress_bar=False)
 
@@ -63,11 +66,15 @@ vectorstore = LangChainFAISS(
     embedding_function=embed_fn
 )
 
+print("✅ FAISS vectorstore ready.")
 
-# === 5. Load Falcon-e-1B-Instruct ===
-# model_id = "tiiuae/falcon-e-1b-instruct"
-tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
-model = AutoModelForCausalLM.from_pretrained("gpt2").to("cuda" if torch.cuda.is_available() else "cpu")
+# 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",
@@ -83,7 +90,6 @@ text_gen_pipeline = pipeline(
 
 llm = HuggingFacePipeline(pipeline=text_gen_pipeline)
 
-# === 6. Prompt Format and Q&A ===
 def truncate_context(context, max_length=1024):
     tokens = tokenizer.encode(context)
     if len(tokens) > max_length:
@@ -92,7 +98,7 @@ def truncate_context(context, max_length=1024):
 
 def format_prompt(context, question):
     return (
-        "You are the Cambridge University Assistant—a friendly, knowledgeable chatbot dedicated to "
+        "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 "
@@ -114,17 +120,17 @@ def answer_fn(question):
     except Exception as e:
         return f"An error occurred: {e}"
 
-# === 7. Gradio Interface ===
+# 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 Cambridge Assistant")
+    gr.Markdown("## 📘 University of Hull Assistant")
     chatbot = gr.Chatbot()
     state = gr.State([])
-    user_input = gr.Textbox(placeholder="Ask a question about Cambridge...", show_label=False)
+    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])