Kurman Karabukaev a60b566d37 [TorchElastic] Support for overprovisioning in C10 based rendezvous (#117066) ... Summary: Allow TorchElastic to manage more nodes than a maximum nnodes specifed in a job. It will be used as a spare capacity/warm nodes for schedulers that support elasticity. RFC: https://github.com/pytorch/pytorch/issues/114097 Test Plan: Integration tests Differential Revision: D52343874 Pull Request resolved: https://github.com/pytorch/pytorch/pull/117066 Approved by: https://github.com/zdevito		2024-01-18 01:16:55 +00:00
..
agent_diagram.jpg
agent.rst
customization.rst
errors.rst
etcd_rdzv_diagram.png
events.rst
examples.rst
kubernetes.rst	Fix typo under docs directory (#92762)	2023-01-23 18:07:22 +00:00
metrics.rst
multiprocessing.rst
quickstart.rst	[BE] Prefer dash over underscore in command-line options (#94505)	2023-02-09 20:16:49 +00:00
rendezvous.rst	[TorchElastic] Support for overprovisioning in C10 based rendezvous (#117066)	2024-01-18 01:16:55 +00:00
run.rst
timer.rst
train_script.rst	[BE] Prefer dash over underscore in command-line options (#94505)	2023-02-09 20:16:49 +00:00