Messing with configuration.

app.py

@@ -31,16 +31,16 @@ print(f"CUDA device: {torch.cuda.get_device_name(torch.cuda.current_device())}")
 # model_name = "MoritzLaurer/deberta-v3-base-zeroshot-v2.0"
 # file_name = "onnx/model.onnx"
 # tokenizer_name = "MoritzLaurer/deberta-v3-base-zeroshot-v2.0"
-# model = ORTModelForSequenceClassification.from_pretrained(model_name, file_name=file_name, provider="CUDAExecutionProvider")
-# tokenizer = AutoTokenizer.from_pretrained(tokenizer_name, model_max_length=512)
+model = ORTModelForSequenceClassification.from_pretrained(model_name, file_name=file_name, export=True, provider="CUDAExecutionProvider")
+tokenizer = AutoTokenizer.from_pretrained(tokenizer_name, model_max_length=512)
 
-model = ORTModelForSequenceClassification.from_pretrained(
-    "philschmid/tiny-bert-sst2-distilled",
-    export=True,
-    provider="CUDAExecutionProvider",
-)
+# model = ORTModelForSequenceClassification.from_pretrained(
+#     "distilbert-base-uncased-finetuned-sst-2-english",
+#     export=True,
+#     provider="CUDAExecutionProvider",
+# )
 
-tokenizer = AutoTokenizer.from_pretrained("philschmid/tiny-bert-sst2-distilled")
+# tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased-finetuned-sst-2-english")
 
 classifier = pipeline(task="zero-shot-classification", model=model, tokenizer=tokenizer, device="cuda:0")
 
@@ -54,6 +54,7 @@ def classify(data_string, request: gradio.Request):
     # else:
     return zero_shot_classification(data)
 
+@spaces.GPU()
 def zero_shot_classification(data):
     results = classifier(data['sequence'], candidate_labels=data['candidate_labels'], hypothesis_template=data['hypothesis_template'], multi_label=data['multi_label'])
     response_string = json.dumps(results)