Another try

app/main.py

@@ -10,11 +10,11 @@ torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
 model_id = "openai/whisper-large-v3"
 
 model = AutoModelForSpeechSeq2Seq.from_pretrained(
-    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
+    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True, cache_dir="./app.cache"
 )
 model.to(device)
 
-processor = AutoProcessor.from_pretrained(model_id)
+processor = AutoProcessor.from_pretrained(model_id, cache_dir="./app.cache")
 
 pipe = pipeline(
     "automatic-speech-recognition",