fix ipex+fp32 jit trace error in ipex 1.13 (#20504)

error show like: “Currently the auto_kernel_selection does not support the grad mode! Please add torch.no_grad() before the inference runtime..”
since jit mode only work in inference mode, it's safe to add such logic.

src/transformers/trainer.py

@@ -1305,6 +1305,7 @@ class Trainer:
                         jit_inputs = tuple(jit_inputs)
                         jit_model = torch.jit.trace(jit_model, jit_inputs, strict=False)
                 jit_model = torch.jit.freeze(jit_model)
+                with torch.no_grad():
                     jit_model(**example_batch)
                     jit_model(**example_batch)
                 model = jit_model