Update app.py

app.py

@@ -8,7 +8,7 @@ import os
 import nltk
 nltk.download('punkt')
 from nltk.tokenize import sent_tokenize
-
+import os
 # Additions for file processing
 import fitz  # PyMuPDF for PDF
 import docx
@@ -18,21 +18,21 @@ import chardet
 
 # --- Device selection ---
 device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
-
+HF_TOKEN = os.getenv("HF_TOKEN")
 # --- Load translation models ---
 def load_models():
     en_dar_model_path = "LocaleNLP/english_hausa"
     en_wol_model_path = "LocaleNLP/eng_wolof"
     en_hau_model_path = "LocaleNLP/english_darija"
 
-    en_dar_model = AutoModelForSeq2SeqLM.from_pretrained(en_dar_model_path).to(device)
-    en_dar_tokenizer = MarianTokenizer.from_pretrained(en_dar_model_path)
+    en_dar_model = AutoModelForSeq2SeqLM.from_pretrained(en_dar_model_path, token=HF_TOKEN).to(device)
+    en_dar_tokenizer = MarianTokenizer.from_pretrained(en_dar_model_path, token=HF_TOKEN)
     
-    en_wol_model = AutoModelForSeq2SeqLM.from_pretrained(en_wol_model_path).to(device)
-    en_wol_tokenizer = MarianTokenizer.from_pretrained(en_wol_model_path)
+    en_wol_model = AutoModelForSeq2SeqLM.from_pretrained(en_wol_model_path, token=HF_TOKEN).to(device)
+    en_wol_tokenizer = MarianTokenizer.from_pretrained(en_wol_model_path, token=HF_TOKEN)
     
-    en_hau_model = AutoModelForSeq2SeqLM.from_pretrained(en_hau_model_path).to(device)
-    en_hau_tokenizer = MarianTokenizer.from_pretrained(en_hau_model_path)
+    en_hau_model = AutoModelForSeq2SeqLM.from_pretrained(en_hau_model_path, token=HF_TOKEN).to(device)
+    en_hau_tokenizer = MarianTokenizer.from_pretrained(en_hau_model_path, token=HF_TOKEN)
 
     en_dar_translator = pipeline("translation", model=en_dar_model, tokenizer=en_dar_tokenizer, device=0 if device.type == 'cuda' else -1)
     en_wol_translator = pipeline("translation", model=en_wol_model, tokenizer=en_wol_tokenizer, device=0 if device.type == 'cuda' else -1)