buptzyb
buptzyb
oh, an easier way is to remove the two `make_weak_ref` calls in `graph.py`, like replacing the `per_callable_static_outputs[per_callable_bwd_idx] = make_weak_ref(static_outputs)` with `per_callable_static_outputs[per_callable_bwd_idx] = static_outputs`. This makes you fully get rid of...
/ok to test c04b798921215af6a3b8d17853e4e09dce23c90a
/ok to test c08df6609d7ab236d950deb25e1e22f2f25880d2
> This PR introduces significant logic changes but no corresponding unit tests. Please consider add tests for _get_sample_arguments() Added to test_cuda_graphs.py
/ok to test c3a590ff650824fba8ba3944a9194aee7ed25dd2
/ok to test c8ecbb436ea9fe30e6d784506f83d88b26a59b29
/ok to test 3a082d0db1e37b63f1d9a7903d2003685cd85a91
/ok to test 3b85592a59cc24785baa41ae8d343fec87311344
/ok to test 97e09403b38049f2007fed74e8f168dd6d8e984a
/ok to test 770b510b59662610321525996ebbda4c1f12af41