Update app.py

app.py

@@ -1,55 +1,47 @@
+import os
+#os.environ['HF_HOME'] = 'E:/huggingface_cache'
+
 import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
-from typing import List, Tuple
 import torch
 
-
-model_name = "Hawoly18/Adia_Llama3.1"
-
-# Vérifier si un GPU est disponible
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-
-
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
-
-def respond(
-    message: str,
-    history: List[Tuple[str, str]],
-    system_message: str,
-    max_tokens: int,
-    temperature: float,
-    top_p: float,
-) -> str:
-    
-    prompt = system_message
-    for user_msg, assistant_msg in history:
-        prompt += f"\nUser: {user_msg}\nAssistant: {assistant_msg}"
-    prompt += f"\nUser: {message}\nAssistant:"
-
-    
-    inputs = tokenizer(prompt, return_tensors="pt")
-    outputs = model.generate(
-        **inputs,
-        max_length=max_tokens,
-        temperature=temperature,
-        top_p=top_p,
-        do_sample=True,
-    )
-    response = tokenizer.decode(outputs[0], skip_special_tokens=True).split("Assistant:")[-1].strip()
+# Load the model and tokenizer
+tokenizer = AutoTokenizer.from_pretrained("Hawoly18/Adia_Llama3.1")
+model = AutoModelForCausalLM.from_pretrained("Hawoly18/Adia_Llama3.1")
+
+if tokenizer.pad_token is None:
+  tokenizer.pad_token = tokenizer.eos_token
+
+# Function to generate responses
+def generate_response(question, max_length=512):
+    input_text = f"Question: {question}\nRéponse:"
+    input_ids = tokenizer.encode(input_text, return_tensors='pt', padding=True, truncation=True)
+    attention_mask = input_ids != tokenizer.pad_token_id
+
+    with torch.no_grad():
+        output_ids = model.generate(
+            input_ids,
+            max_length=max_length,
+            attention_mask=attention_mask,
+            pad_token_id=tokenizer.eos_token_id,
+            eos_token_id=tokenizer.eos_token_id,
+            num_beams=5,  # Beam search for better quality
+            no_repeat_ngram_size=2,  # Prevent n-gram repetition
+            early_stopping=True
+        )
+    response = tokenizer.decode(output_ids[0], skip_special_tokens=True)
+    response = response.replace(input_text, "").strip()
     return response
 
-
-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"),  # Fixed syntax error
-        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
-    ],
-    title="Chatbot Interface"
+# Define the Gradio interface
+interface = gr.Interface(
+    fn=generate_response,
+    inputs="text",
+    outputs="text",
+    title="Model Q&A Interface",
+    description="Ask a question related to BSE and entrepreneurship!",
+    
 )
 
-if __name__ == "__main__":
-    demo.launch()
+# Launch the interface
+interface.launch(share=True)