Update handler.py

handler.py

@@ -6,32 +6,37 @@ import open_clip
 class EndpointHandler:
     """
     Zero‑shot classifier for MobileCLIP‑B (OpenCLIP).
-    Request:
-      {
-        "image": "<base64‑png/jpeg>",
+
+    Expected client JSON *to the endpoint*:
+    {
+      "inputs": {
+        "image": "<base64 PNG/JPEG>",
         "candidate_labels": ["cat", "dog", ...]
       }
-    Response: list[{"label": str, "score": float}]
+    }
     """
 
     def __init__(self, path: str = ""):
-        weights = f"{path}/mobileclip_b.pt"           # ckpt in your repo
+        weights = f"{path}/mobileclip_b.pt"
         self.model, _, self.preprocess = open_clip.create_model_and_transforms(
             "MobileCLIP-B", pretrained=weights
         )
         self.model.eval()
 
         self.tokenizer = open_clip.get_tokenizer("MobileCLIP-B")
-        self.device    = "cuda" if torch.cuda.is_available() else "cpu"
+        self.device = "cuda" if torch.cuda.is_available() else "cpu"
         self.model.to(self.device)
 
     def __call__(self, data):
-        img_b64 = data["image"]
-        labels  = data.get("candidate_labels", [])
+        # ── unwrap Hugging Face's `inputs` envelope ───────────
+        payload = data.get("inputs", data)
+
+        img_b64 = payload["image"]
+        labels  = payload.get("candidate_labels", [])
         if not labels:
             return {"error": "candidate_labels list is empty"}
 
-        # Decode + preprocess image
+        # Decode & preprocess image
         image = Image.open(io.BytesIO(base64.b64decode(img_b64))).convert("RGB")
         img_tensor = self.preprocess(image).unsqueeze(0).to(self.device)
 
@@ -44,11 +49,10 @@ class EndpointHandler:
             txt_feat = self.model.encode_text(text_tokens)
             img_feat = img_feat / img_feat.norm(dim=-1, keepdim=True)
             txt_feat = txt_feat / txt_feat.norm(dim=-1, keepdim=True)
-            probs    = (100 * img_feat @ txt_feat.T).softmax(dim=-1)[0].tolist()
+            probs = (100 * img_feat @ txt_feat.T).softmax(dim=-1)[0].tolist()
 
-        # Return sorted results
+        # Sorted output
         return [
             {"label": l, "score": float(p)}
             for l, p in sorted(zip(labels, probs), key=lambda x: x[1], reverse=True)
         ]
-