fuel
fuel copied to clipboard
mila-iqia
A data pipeline framework for machine learning
Right now it's somewhat hard to write a class that inherits from IndexableDataset and doesn't duplicate a bit of annoying logic. IIRC some inheriting classes have forgone `start` and `stop`...
Just sent them an e-mail about it: > I recently ran into what seems to be a bug in Scrutinizer's static code analysis. Consider any recent analysis log of the...
At my best knowledge it is not currently possible to compute statistics on the whole dataset, e.g., per-pixel mean and standard deviation. This wouldn't fit well in the `Transformer` class...
This is something that should be simple but isn't. I'd like to calculate the mean and standard deviation of my data _as returned by a Transformer chain_, then stick another...
It seems very weird to me that a `Dataset` object returns a `state` through `Dataset.open` and then manipolates the state itself with, e.g., `Dataset.get_data(state)`. It seems much better to have...
I know it was discussed before but I wasn't able to find an open issue on this. It would be useful to have a module that takes one or more...
The `AbstractDataStream` and the `Dataset` classes are both abstract, but the naming is not consistent. Is there a reason for this that I am not seeing? I like "Abstract" to...
Currently (correct) job resumption and the "server process" are an either/or. It would be nice to at least think about fixing this. @bartvm Have you thought at all about how...
Scripts like `fuel-download` used to be outside of the `fuel` package, and not tested for coverage. Now, we need them to be importable, so #211 moved them to `fuel.bin`, and...
Why not just generate both and put them in the same file under different dataset/source names? This dataset is _tiny_, there's really no reason not to do this.