Update app.py

app.py

@@ -16,9 +16,9 @@ summ = pipeline("summarization", model=SUM_MODEL)
 # === REQUEST SCHEMAS ===
 class ZSCReq(BaseModel):
     text: str
-    labels: list[str]
+    labels: list[str] : []
     multi_label: bool = True
-    template: str = "This text is about {}."
+    template: str = "This feedback is primarily about {}."
 
 class SAReq(BaseModel):
     text: str
@@ -29,6 +29,16 @@ class SumReq(BaseModel):
     min_length: int = 20
     do_sample: bool = False
 
+DEFAULT_LABELS = [
+    "Usability", "Performance", "Visual Design",
+    "Feedback", "Navigation", "Responsiveness",
+]
+
+@app.on_event("startup")
+def warmup():
+    # tiny warmup so first real call isn’t cold
+    _ = zsc("warmup", candidate_labels=DEFAULT_LABELS, multi_label=False)
+    
 # === ROUTES ===
 @app.get("/")
 def health():
@@ -38,18 +48,22 @@ def health():
 # in your HF Space app.py
 @app.post("/predict")
 def predict(req: ZSCReq):
+    labels = [l for l in (req.labels or DEFAULT_LABELS) if l in DEFAULT_LABELS]
+    if not labels:
+        labels = DEFAULT_LABELS
+        
     out = zsc(
         req.text,
         candidate_labels=req.labels,
-        multi_label=req.multi_label,
-        hypothesis_template=req.template,
+        multi_label=False,
+        hypothesis_template=req.template or "This feedback is primarily about {}.",
     )
     return {"labels": out["labels"], "scores": out["scores"]}
 
 @app.post("/sa")
 def sentiment(req: SAReq):
     result = sa(req.text)[0]
-    return {"label": result["label"], "score": float(result["score"])}
+    return {"label": result["label"], "scores": [float(s) for s in out["scores"]],}
 
 @app.post("/sum")
 def summarize(req: SumReq):