Update app.py

app.py

@@ -8,26 +8,26 @@ from sentence_transformers import SentenceTransformer
 from transformers import AutoTokenizer, AutoModelForSequenceClassification
 from torch.nn.functional import softmax
 
-# ========== Model Paths ==========
-MODEL_DIR = "model"
-SBERT_PATH = os.path.join(MODEL_DIR, "sbert_model")
-SBERT_CLF_PATH = os.path.join(MODEL_DIR, "sbert_mlp.pkl")
-DEBERTA_PATH = os.path.join(MODEL_DIR, "deberta_model")
+from huggingface_hub import hf_hub_download
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
+
 # ========== Load Models ==========
-# Load SBERT sentence embedding model
-sbert = SentenceTransformer(SBERT_PATH, device=device)
 
-# Load trained SBERT classifier (MLP)
-clf = joblib.load(SBERT_CLF_PATH)
+# Load SBERT encoder
+sbert = SentenceTransformer("Sbhatti33/sbert-stance-model")
+
+# Load SBERT classifier
+clf_path = hf_hub_download(repo_id="Sbhatti33/sbert-stance-model", filename="sbert_mlp.pkl")
+clf = joblib.load(clf_path)
 
 # Load DeBERTa model and tokenizer
-deberta_tokenizer = AutoTokenizer.from_pretrained(DEBERTA_PATH)
-deberta_model = AutoModelForSequenceClassification.from_pretrained(DEBERTA_PATH).to(device)
+deberta_tokenizer = AutoTokenizer.from_pretrained("Sbhatti33/deberta-stance-model")
+deberta_model = AutoModelForSequenceClassification.from_pretrained("Sbhatti33/deberta-stance-model").to(device)
 deberta_model.eval()
 
+
 # ========== Class Names ==========
 label_map = {0: "believer", 1: "neutral", 2: "denier"}
 class_names = ["believer", "neutral", "denier"]