Simplifying.

app.py

@@ -1,5 +1,4 @@
 import gradio as gr
-import json
 from transformers import AutoTokenizer, pipeline
 from optimum.onnxruntime import ORTModelForSequenceClassification
 
@@ -10,21 +9,19 @@ model = ORTModelForSequenceClassification.from_pretrained(model_id, file_name=fi
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 
 classifier = pipeline(
-    task="text-classification",
-    model=model,
-    tokenizer=tokenizer,
-    top_k=None,
-    function_to_apply="sigmoid",
-    binary_output=True,
+    task = "text-classification",
+    model = model,
+    tokenizer = tokenizer,
+    top_k = None,
+    function_to_apply = "sigmoid",
 )
 
 def predict(param_0):
     results = classifier(param_0)
-    response = {
+    return {
         "label": results[0][0]["label"],
         "confidences": results[0]
     }
-    return response
 
 demo = gr.Interface(
     fn = predict,

requirements.txt

@@ -1,5 +1,4 @@
 torch==2.4.0
 huggingface_hub==0.26.0
 transformers==4.36
-optimum[exporters,onnxruntime]==1.21.3
-json5==0.9.25
+optimum[exporters,onnxruntime]==1.21.3