mirror of
https://github.com/saymrwulf/pytorch.git
synced 2026-05-14 20:57:59 +00:00
fixes https://github.com/pytorch/pytorch/issues/139404. The input grads get saved in a new `self.bwd_cache` container and get popped off after they are used in `backward_one_chunk` `python test/distributed/pipelining/test_schedule_multiproc.py -k test_pipeline_schedule_runtime_custom_sched` Pull Request resolved: https://github.com/pytorch/pytorch/pull/140981 Approved by: https://github.com/wconstab |
||
|---|---|---|
| .. | ||
| artifacts | ||
| __init__.py | ||
| model_registry.py | ||
| schedule_registry.py | ||
| test_backward.py | ||
| test_microbatch.py | ||
| test_pipe.py | ||
| test_schedule.py | ||
| test_schedule_multiproc.py | ||
| test_stage.py | ||
| test_transformer.py | ||
| test_unflatten.py | ||