Update app.py

app.py

@@ -123,11 +123,12 @@ def GenerateMIDI(num_tok, idrums, iinstr):
         inp = torch.LongTensor([outy]).cuda()
 
         with torch.amp.autocast(device_type='cuda', dtype=torch.bfloat16):
-            out = model.module.generate(inp,
-                                        1,
-                                        temperature=0.9,
-                                        return_prime=False,
-                                        verbose=False)
+            with torch.inference_mode():
+                out = model.module.generate(inp,
+                                            1,
+                                            temperature=0.9,
+                                            return_prime=False,
+                                            verbose=False)
 
         out0 = out[0].tolist()
         outy.extend(out0)