Rupesh K Srivastava
Rupesh K Srivastava
Since some papers have made available pre-trained Caffe convnets, it'd be nice to be able to use them in Brainstorm.
For double buffering to work properly, the new strategy of copying data to device in `provide_external_data` instead of the iterators requires some changes to how InputLayer works. This is a...
We should have a `context_reset_rate` parameter (subject to renaming) in the trainer which is set by the `train` function. Using this, the context should be reset (cleared) if `current_update_nr` %...
From what I can see, the mechanism for generating descriptions and serializing networks etc. does not fully work yet. @Qwlouse, you were working on this. Any comments on what else...
How about using `blacken-docs` as a pre-commit hook as I did here? If it's too annoying we can remove it but still keep the formatted code, and manually control formatting...
When defining a `Problem`, if the user specifies `device="cuda", num_actors=2` (as an example), but does not specify `num_gpus_per_actor`, the following error results from each actor which is not fully clear...
This is related to scientific comparisons but also checkpointing/resuming a search, as pointed out by user DTrescher on Slack. Even if full reproducibility is not possible, we can clarify to...