Update app.py

app.py

@@ -1,56 +1,46 @@
 import gradio as gr
-from transformers import AutoModelForCausalLM, AutoTokenizer
 import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-# Load the fine-tuned Llama-3-8B model and tokenizer for ubiodee/plutus_llm
-model_name = "ubiodee/plutus_llm"
-tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False)  # Safeguard against fast tokenizer issues
-model = AutoModelForCausalLM.from_pretrained(
-    model_name, 
-    torch_dtype=torch.float16, 
-    device_map="auto",
-    load_in_8bit=True  # Enable 8-bit quantization as per model specs
-)
+# Load model & tokenizer
+MODEL_NAME = "ubiodee/plutus_llm"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+model = AutoModelForCausalLM.from_pretrained(MODEL_NAME)
+model.eval()
+
+if torch.cuda.is_available():
+    model.to("cuda")
+
+# Response function
+def generate_response(prompt):
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+
+    with torch.no_grad():
+        outputs = model.generate(
+            **inputs,
+            max_new_tokens=200,
+            temperature=0.7,
+            top_p=0.9,
+            do_sample=True,
+            eos_token_id=tokenizer.eos_token_id,
+            pad_token_id=tokenizer.pad_token_id,
+        )
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+    # Remove the prompt from the output to return only the answer
+    if response.startswith(prompt):
+        response = response[len(prompt):].strip()
+
+    return response
 
-# Set padding token if not already set
-if tokenizer.pad_token is None:
-    tokenizer.pad_token = tokenizer.eos_token
-
-def generate_text(prompt, max_length=200, temperature=0.7, top_p=0.9):
-    # Tokenize the input prompt
-    inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True).to("cuda" if torch.cuda.is_available() else "cpu")
-    
-    # Generate text
-    outputs = model.generate(
-        inputs["input_ids"],
-        attention_mask=inputs["attention_mask"],
-        max_length=max_length,
-        temperature=temperature,
-        top_p=top_p,
-        do_sample=True,
-        num_return_sequences=1,
-        pad_token_id=tokenizer.eos_token_id
-    )
-    
-    # Decode the generated text
-    generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    # Remove the input prompt from the output for cleaner response
-    generated_text = generated_text[len(prompt):].strip()
-    return generated_text
-
-# Create Gradio interface
+# Gradio UI
 demo = gr.Interface(
-    fn=generate_text,
-    inputs=[
-        gr.Textbox(label="Input Prompt", placeholder="Enter your prompt here...", lines=3),
-        gr.Slider(label="Max Length", minimum=50, maximum=500, value=200, step=10),
-        gr.Slider(label="Temperature", minimum=0.1, maximum=2.0, value=0.7, step=0.1),
-        gr.Slider(label="Top P", minimum=0.1, maximum=1.0, value=0.9, step=0.05)
-    ],
-    outputs=gr.Textbox(label="Generated Text", lines=10),
-    title="Plutus LLM Demo (ubiodee/plutus_llm)",
-    description="Interact with the fine-tuned Llama-3-8B model using LoRA and 8-bit quantization. This is based on ubiodee/plutus_llm."
+    fn=generate_response,
+    inputs=gr.Textbox(label="Enter your prompt", lines=4, placeholder="Ask about Plutus..."),
+    outputs=gr.Textbox(label="Model Response"),
+    title="Cardano Plutus AI Assistant",
+    description="Ask questions about Plutus smart contracts or Cardano blockchain."
 )
 
-if __name__ == "__main__":
-    demo.launch()
+demo.launch()