Update embedder.py

embedder.py

@@ -1,8 +1,19 @@
 # embedder.py
 
-from sentence_transformers import SentenceTransformer
+from transformers import AutoTokenizer, AutoModel
+import torch
 
-model = SentenceTransformer("all-MiniLM-L6-v2")
+# Use a model with PyTorch weights available
+MODEL_NAME = "thenlper/gte-small"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+model = AutoModel.from_pretrained(MODEL_NAME)
 
 def get_embeddings(texts):
-    return model.encode(texts)
+    inputs = tokenizer(texts, padding=True, truncation=True, return_tensors="pt")
+    with torch.no_grad():
+        model_output = model(**inputs)
+
+    # Mean Pooling
+    embeddings = model_output.last_hidden_state.mean(dim=1)
+    return embeddings.numpy()