Update app.py

app.py

@@ -1,5 +1,5 @@
 import streamlit as st
-from transformers import AutoModel, AutoTokenizer
+from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
 from huggingface_hub import login
 import PyPDF2
 import pandas as pd
@@ -17,17 +17,7 @@ st.set_page_config(
 )
 
 # Model name
-MODEL_NAME = "deepseek-ai/DeepSeek-V3-0324"
-
-# Translation prompt template
-TRANSLATION_PROMPT = """
-You are a professional translator specializing in English-to-French translation. Translate the following text accurately and naturally into French, preserving the original meaning and tone:
-
-**Text to translate:**
-{input_text}
-
-**French translation:**
-"""
+MODEL_NAME = "Helsinki-NLP/opus-mt-en-fr"
 
 # Title with rocket emojis
 st.title("🚀 English to French Translator 🚀")
@@ -80,22 +70,19 @@ def load_model(hf_token):
         
         login(token=hf_token)
         
-        # Load tokenizer with trust_remote_code=True
+        # Load tokenizer
         tokenizer = AutoTokenizer.from_pretrained(
             MODEL_NAME,
-            token=hf_token,
-            trust_remote_code=True
+            token=hf_token
         )
         
         # Load the model with appropriate dtype for CPU/GPU compatibility
         dtype = torch.float16 if DEVICE == "cuda" else torch.float32
-        model = AutoModel.from_pretrained(
+        model = AutoModelForSeq2SeqLM.from_pretrained(
             MODEL_NAME,
             token=hf_token,
             torch_dtype=dtype,
-            device_map="auto",  # Automatically maps to CPU or GPU
-            quantization_config=None,  # Disable FP8 quantization
-            trust_remote_code=True  # Allow custom code execution
+            device_map="auto"  # Automatically maps to CPU or GPU
         )
         
         return model, tokenizer
@@ -107,11 +94,8 @@ def load_model(hf_token):
 # Generation function for translation
 def generate_translation(input_text, model, tokenizer):
     try:
-        # Prepare the prompt
-        full_prompt = TRANSLATION_PROMPT.format(input_text=input_text)
-        
-        # Tokenize the input
-        inputs = tokenizer(full_prompt, return_tensors="pt", padding=True, truncation=True, max_length=512)
+        # Tokenize the input (no prompt needed for seq2seq translation models)
+        inputs = tokenizer(input_text, return_tensors="pt", padding=True, truncation=True, max_length=512)
         inputs = inputs.to(DEVICE)
         
         # Generate translation
@@ -120,19 +104,14 @@ def generate_translation(input_text, model, tokenizer):
             outputs = model.generate(
                 input_ids=inputs["input_ids"],
                 attention_mask=inputs["attention_mask"],
-                max_new_tokens=512,
-                temperature=0.7,
-                top_p=0.9,
-                repetition_penalty=1.1,
-                do_sample=True,
-                num_return_sequences=1
+                max_length=512,
+                num_beams=5,
+                length_penalty=1.0,
+                early_stopping=True
             )
         
         # Decode the output
         translation = tokenizer.decode(outputs[0], skip_special_tokens=True)
-        
-        # Extract the French translation part (after the prompt)
-        translation = translation.split("**French translation:**")[-1].strip()
         return translation
         
     except Exception as e: