Update app.py

app.py

@@ -1,64 +1,143 @@
+# app.py - Complete Chatbot with Fine-tuning and Deployment
 import gradio as gr
-from huggingface_hub import InferenceClient
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline, Trainer, TrainingArguments
+from datasets import load_dataset, Dataset
+import torch
+import pandas as pd
+from huggingface_hub import notebook_login, Repository
 
-"""
-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")
+# Configuration
+MODEL_NAME = "t5-small"  # Lightweight model good for chatbots
+DATASET_NAME = "AmazonQA"
+FINETUNED_MODEL_NAME = "MujtabaShopifyChatbot"
+HF_TOKEN = "your_huggingface_token"  # Replace with your token
 
+# --- Step 1: Load and Prepare Dataset ---
+def load_and_preprocess_data():
+    print("Loading AmazonQA dataset...")
+    dataset = load_dataset(DATASET_NAME)
+    
+    # Convert to pandas for easier processing
+    df = pd.DataFrame(dataset['train'])
+    
+    # Preprocessing - create consistent Q&A pairs
+    df = df[['question', 'answer']].dropna()
+    df = df[:5000]  # Use subset for faster training
+    
+    # Convert back to Hugging Face Dataset
+    processed_dataset = Dataset.from_pandas(df)
+    
+    # Split into train and eval
+    split_dataset = processed_dataset.train_test_split(test_size=0.1)
+    return split_dataset
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+# --- Step 2: Tokenization ---
+def tokenize_data(dataset):
+    print("Tokenizing data...")
+    tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+    
+    def preprocess_function(examples):
+        inputs = [f"question: {q} answer:" for q in examples["question"]]
+        targets = examples["answer"]
+        
+        model_inputs = tokenizer(inputs, max_length=128, truncation=True)
+        labels = tokenizer(targets, max_length=128, truncation=True)
+        
+        model_inputs["labels"] = labels["input_ids"]
+        return model_inputs
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+    tokenized_dataset = dataset.map(preprocess_function, batched=True)
+    return tokenized_dataset
 
-    messages.append({"role": "user", "content": message})
+# --- Step 3: Fine-tuning ---
+def fine_tune_model(tokenized_dataset):
+    print("Fine-tuning model...")
+    model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME)
+    
+    training_args = TrainingArguments(
+        output_dir="./results",
+        evaluation_strategy="epoch",
+        learning_rate=2e-5,
+        per_device_train_batch_size=8,
+        per_device_eval_batch_size=8,
+        num_train_epochs=3,
+        weight_decay=0.01,
+        save_total_limit=3,
+        fp16=torch.cuda.is_available(),
+        push_to_hub=True,
+        hub_model_id=FINETUNED_MODEL_NAME,
+        hub_token=HF_TOKEN,
+    )
+    
+    trainer = Trainer(
+        model=model,
+        args=training_args,
+        train_dataset=tokenized_dataset["train"],
+        eval_dataset=tokenized_dataset["test"],
+    )
+    
+    trainer.train()
+    trainer.push_to_hub()
+    return model
 
-    response = ""
+# --- Step 4: Chatbot Interface ---
+def initialize_chatbot():
+    print("Loading chatbot...")
+    try:
+        # Try loading fine-tuned model first
+        model = AutoModelForSeq2SeqLM.from_pretrained(FINETUNED_MODEL_NAME)
+        tokenizer = AutoTokenizer.from_pretrained(FINETUNED_MODEL_NAME)
+    except:
+        # Fallback to pre-trained model
+        model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME)
+        tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+    
+    chatbot_pipe = pipeline("text2text-generation", model=model, tokenizer=tokenizer)
+    return chatbot_pipe
 
-    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)",
-        ),
-    ],
-)
+def generate_response(message, history):
+    # Format the input for the model
+    input_text = f"question: {message} answer:"
+    
+    # Generate response
+    response = chatbot_pipe(input_text, max_length=128, do_sample=True)[0]['generated_text']
+    
+    # Clean up the response
+    if "answer:" in response:
+        response = response.split("answer:")[-1].strip()
+    return response
 
+# --- Step 5: Deployment ---
+def deploy_chatbot():
+    print("Launching chatbot interface...")
+    demo = gr.ChatInterface(
+        fn=generate_response,
+        title="Mujtaba's Shopify Chatbot",
+        description="Ask me anything about products, shipping, or returns!",
+        examples=[
+            "What's the return policy?",
+            "How long does shipping take to Karachi?",
+            "Do you have size charts for kurtas?"
+        ],
+        theme="soft"
+    )
+    return demo
 
+# --- Main Execution ---
 if __name__ == "__main__":
-    demo.launch()
+    # Login to Hugging Face Hub
+    notebook_login()
+    
+    # Dataset preparation
+    dataset = load_and_preprocess_data()
+    tokenized_dataset = tokenize_data(dataset)
+    
+    # Fine-tuning (uncomment to run)
+    # fine_tune_model(tokenized_dataset)
+    
+    # Initialize chatbot
+    chatbot_pipe = initialize_chatbot()
+    
+    # Launch interface
+    demo = deploy_chatbot()
+    demo.launch(share=True)