Emily (Miller) Dorne
Emily (Miller) Dorne
- [ ] add horizontal line rules - [ ] ability to append (deal with `pard`) - [ ] tests
This is harder than anticipated. Consider not including RTF in initial release.
Potentially include options to turn off checklist line summaries with `include_summaries` param
LIME: https://arxiv.org/pdf/1602.04938.pdf
Turns out that `auto_lr_find` was broken in PTL, and has been fixed. However, using a more recent version hasn't yet worked
Moving context from slack into issue: there was a bug where `auto_lr_find` wasn’t resetting to the best learning rate (it just said it was). so it was using a crazy...
Running with an updated version of PTL (1.7.5) yields the following error ``` Finding best initial lr: 81%|████████████████████████▎ | 81/100 [08:13
Confirmed that if we override that `on_fit_start` change, things work as expected. in `zamba.pytorch.finetuning`, add: ``` def on_fit_start(self, trainer, pl_module): """ Raises: MisconfigurationException: If LightningModule has no nn.Module `backbone` attribute....
`yolox` is now pip-installable 🎉 as of release on 2022/01/18: https://github.com/Megvii-BaseDetection/YOLOX/releases
This check may be as simple as just verifying that there is at least one train video and at least one val video. We will need to test if this...