Update app.py

app.py

@@ -1,39 +1,39 @@
 import os
-from transformers import pipeline
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForCausalLM
 from sentence_transformers import SentenceTransformer
 from sklearn.metrics.pairwise import cosine_similarity
 import numpy as np
 import pandas as pd
-import gradio as gr
 
 # Load pre-trained Sentence Transformer model
-model = SentenceTransformer('LaBSE')
+model_sentence_transformer = SentenceTransformer('LaBSE')
 
 # Load questions and answers from the CSV file
 df = pd.read_csv('combined_questions_and_answers.csv')
 
 # Encode all questions in the dataset
-question_embeddings = model.encode(df['Question'].tolist())
+question_embeddings = model_sentence_transformer.encode(df['Question'].tolist())
 
 # Hugging Face API details for Meta-Llama 3B
-api_key = os.getenv("HUGGINGFACE_API_KEY")
-if not api_key:
+HF_TOKEN = os.environ.get("HUGGINGFACE_API_KEY", None)
+if not HF_TOKEN:
     raise ValueError("Hugging Face API key not found in environment variables. Please set the HUGGINGFACE_API_KEY environment variable.")
 
-pipe = pipeline("text-generation", model="meta-llama/Meta-Llama-3-8B-Instruct", token=api_key)
+# Load the tokenizer and model
+tokenizer = AutoTokenizer.from_pretrained("meta-llama/Meta-Llama-3-8B-Instruct")
+model = AutoModelForCausalLM.from_pretrained("meta-llama/Meta-Llama-3-8B-Instruct", device_map="auto")
 
 # Function to refine and translate text using Meta-Llama 3B
 def refine_text(prompt):
-    messages = [
-        {"role": "user", "content": prompt},
-    ]
-    response = pipe(messages)
-    return response[0]['generated_text']
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    outputs = model.generate(**inputs, max_new_tokens=50)
+    return tokenizer.decode(outputs[0], skip_special_tokens=True)
 
 # Function to find the most similar question and provide the answer
 def get_answer(user_question, threshold=0.30):
     # Encode the user question
-    user_embedding = model.encode(user_question)
+    user_embedding = model_sentence_transformer.encode(user_question)
 
     # Calculate cosine similarities
     similarities = cosine_similarity([user_embedding], question_embeddings)