Update app.py

app.py

@@ -6,9 +6,8 @@ import torch
 from huggingface_hub import login
 import os
 
-
-# Retrieve the Hugging Face token from secrets (replace 'HUGGINGFACE_TOKEN' with your secret key)
-hf_token = os.getenv('My_Token')
+# Directly assign your Hugging Face token here
+hf_token = "your_hugging_face_api_token"
 
 # Log in to Hugging Face
 login(token=hf_token)
@@ -101,10 +100,18 @@ def google_search(term, num_results=5, lang="en", timeout=5, safe="active", ssl_
     print(f"Total results fetched: {len(all_results)}")
     return all_results
 
-# Load the Mixtral-8x7B-Instruct model and tokenizer
+# Load the Mixtral-8x7B-Instruct model and tokenizer with authorization header
 model_name = 'mistralai/Mistral-7B-Instruct-v0.3'
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
+headers = {"Authorization": f"Bearer {hf_token}"}
+
+# Ensure sentencepiece is installed
+try:
+    import sentencepiece
+except ImportError:
+    raise ImportError("The sentencepiece library is required for this tokenizer. Please install it with `pip install sentencepiece`.")
+
+tokenizer = AutoTokenizer.from_pretrained(model_name, use_auth_token=hf_token)
+model = AutoModelForCausalLM.from_pretrained(model_name, use_auth_token=hf_token)
 
 # Check if a GPU is available and if not, fall back to CPU
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
@@ -118,7 +125,7 @@ search_results = google_search(search_term, num_results=3)
 combined_text = "\n\n".join(result['text'] for result in search_results if result['text'])
 
 # Tokenize the input text
-inputs = tokenizer(combined_text, return_tensors="pt")
+inputs = tokenizer(combined_text, return_tensors="pt").to(device)
 
 # Generate a response
 outputs = model.generate(**inputs, max_length=150, temperature=0.7, top_p=0.9, top_k=50)