Messing with configuration.

app.py

@@ -36,12 +36,13 @@ print(f"CUDA device: {torch.cuda.get_device_name(torch.cuda.current_device())}")
 # model = ORTModelForSequenceClassification.from_pretrained(model_name, export=True, provider="CUDAExecutionProvider")
 # tokenizer = AutoTokenizer.from_pretrained(tokenizer_name, model_max_length=512)
 
+print(f"ORTModelForSequenceClassification.from_pretrained")
 model = ORTModelForSequenceClassification.from_pretrained(
     "philschmid/tiny-bert-sst2-distilled",
     export=True,
     provider="CUDAExecutionProvider",
 )
-
+print(f"AutoTokenizer.from_pretrained")
 tokenizer = AutoTokenizer.from_pretrained("philschmid/tiny-bert-sst2-distilled")
 
 # classifier = pipeline(task="zero-shot-classification", model=model, tokenizer=tokenizer, device="cuda:0")

requirements.txt

@@ -2,8 +2,8 @@ torch==2.4.0
 fastapi==0.88.0
 huggingface_hub==0.23.5
 json5==0.9.25
-numpy<2.0
-optimum[exporters,onnxruntime-gpu]==1.21.4
-transformers
+numpy
+optimum[exporters,onnxruntime-gpu]==1.24.0
+transformers==4.50.3
 sentence-transformers==3.0.1
 safetensors