Swapping CPU model.

app.py

@@ -23,16 +23,20 @@ app.add_middleware(
 print(f"Is CUDA available: {torch.cuda.is_available()}")
 if torch.cuda.is_available():
     print(f"CUDA device: {torch.cuda.get_device_name(torch.cuda.current_device())}")
+    print(f"CUDA version: {torch.version.cuda}")
 
-model_name = "MoritzLaurer/roberta-large-zeroshot-v2.0-c"
-tokenizer_name = "MoritzLaurer/roberta-large-zeroshot-v2.0-c"
+model_name_cpu = "MoritzLaurer/ModernBERT-large-zeroshot-v2.0"
+tokenizer_name_cpu = "MoritzLaurer/ModernBERT-large-zeroshot-v2.0"
 
-classifier_cpu = pipeline(task="zero-shot-classification", model=model_name, tokenizer=tokenizer_name)
-classifier_gpu = pipeline(task="zero-shot-classification", model=model_name, tokenizer=tokenizer_name, device="cuda:0") if torch.cuda.is_available() else classifier_cpu
+model_name_gpu = "MoritzLaurer/roberta-large-zeroshot-v2.0-c"
+tokenizer_name_gpu = "MoritzLaurer/roberta-large-zeroshot-v2.0-c"
+
+classifier_cpu = pipeline(task="zero-shot-classification", model=model_name_cpu, tokenizer=tokenizer_name_cpu, device="cpu", torch_dtype=torch.bfloat16)
+classifier_gpu = pipeline(task="zero-shot-classification", model=model_name_gpu, tokenizer=tokenizer_name_gpu, device="cuda", torch_dtype=torch.bfloat16) if torch.cuda.is_available() else classifier_cpu
 
 def classify(data_string, request: gradio.Request):
     if request:
-        if request.headers["origin"] not in ["https://statosphere-3704059fdd7e.c5v4v4jx6pq5.win", "https://crunchatize-77a78ffcc6a6.c5v4v4jx6pq5.win", "https://crunchatize-2-2b4f5b1479a6.c5v4v4jx6pq5.win", "https://tamabotchi-2dba63df3bf1.c5v4v4jx6pq5.win", "https://ravenok-statosphere-backend.hf.space", "https://lord-raven.github.io"]:
+        if request.headers["origin"] not in ["https://statosphere-3704059fdd7e.c5v4v4jx6pq5.win", "https://crunchatize-77a78ffcc6a6.c5v4v4jx6pq5.win", "https://crunchatize-2-2b4f5b1479a6.c5v4v4jx6pq5.win", "https://tamabotchi-2dba63df3bf1.c5v4v4jx6pq5.win", "https://ravenok-statosphere-backend.hf.space", "https://ravenok-test-space.hf.space",  "https://lord-raven.github.io"]:
             return "{}"
     data = json.loads(data_string)
 

requirements.txt

@@ -1,8 +1,8 @@
 torch==2.4.0
 fastapi==0.88.0
-huggingface_hub==0.26.0
+huggingface_hub==0.33.0
 json5==0.9.25
 numpy
 uvicorn
-optimum[onnxruntime-gpu]==1.24.0
-transformers==4.36
+optimum[onnxruntime-gpu]==1.26.1
+transformers==4.52.4