Issues

[Action Needed] This repo is inactive
microsoft/PyMarlin#96 · opened on Jun 15, 2025
Enable distributed training over multiple node in compliant detonation chamber
microsoft/PyMarlin#93 · opened on May 25, 2022
1
Restrict scope of modification of root logger to logging_utils
microsoft/PyMarlin#88 · opened on May 18, 2022
1
Gracefully handle hang issue when training with uneven batches via throwing an exception
microsoft/PyMarlin#82 · opened on Jan 11, 2022
Add support for uneven number of batches on different GPUs
microsoft/PyMarlin#81 · opened on Jan 11, 2022
DDP Hangs with uneven batches
microsoft/PyMarlin#80 · opened on Jan 10, 2022
Link the blogposts?
microsoft/PyMarlin#79 · opened on Jan 10, 2022
log_graph is pretty restrictive
microsoft/PyMarlin#78 · opened on Dec 23, 2021
tb_log_hist_steps is misplaced in the Tensorboard writer
microsoft/PyMarlin#77 · opened on Dec 23, 2021
这个真没必要
microsoft/PyMarlin#76 · opened on Dec 10, 2021
Running validation more often than once an epoch.
microsoft/PyMarlin#74 · opened on Dec 3, 2021
Easy self.log() and other QoL improvements such as sane defaults for module_interface
microsoft/PyMarlin#71 · opened on Oct 22, 2021
Dockerfile and CI pipeline for Docker image
microsoft/PyMarlin#70 · opened on Oct 21, 2021
Add integration test for DDP
microsoft/PyMarlin#67 · opened on Oct 7, 2021
1
DeepSpeed Trainer Backend
microsoft/PyMarlin#63 · opened on Sep 16, 2021
user defined save_best_checkpoint() hook
microsoft/PyMarlin#58 · opened on Aug 18, 2021
1
CustomArgParser doesn't support the yaml_file_arg_key functionality
microsoft/PyMarlin#57 · opened on Aug 12, 2021
refactor to units of steps instead of epochs
microsoft/PyMarlin#56 · opened on Aug 10, 2021
DDP validation: All gather for flattened 1D tensors taking long time to complete
microsoft/PyMarlin#55 · opened on Aug 10, 2021
4