pytorch
7e32e604 - Fix linspace step computation for large integral types (#40132)

Commit
4 years ago
Fix linspace step computation for large integral types (#40132) Summary: Convert start and end to `step_t` before computing the difference Should fix `torch.linspace(-2147483647, 2147483647, 10, dtype=torch.int32)` Closes https://github.com/pytorch/pytorch/issues/40118 Pull Request resolved: https://github.com/pytorch/pytorch/pull/40132 Differential Revision: D22190095 Pulled By: malfet fbshipit-source-id: 01cb158a30c505191df663d021804d411b697871
Author
Parents
Loading