Eric Battenberg

I agree with @andravin about omitting the L1 grad-wrt-inputs from the overall summary (at least in a special column). That _really_ hurts the Theano legacy numbers, and it isn't even...

Are there different peak picking modes? I ask because in most of my work, I've only used causal peak-picking (with a small amount of latency) because of the real-time aspect...

> Is this repo going to be supported? Unfortunately, I won't be working on it, but I'd be glad to review contributions.

I haven't run the code in about 4 years, so I can't guarantee that it will continue to work with current versions of CUDA. Sorry. On Tue, Oct 10, 2017...

Agree with @ebenolson and @benanne.

This issue seems to be be resolved now, correct?

If so, it might be helpful to future users to add a TZ line to the example docker compose config. That seems like a useful default config hint for the...

I'm happy to provide more info on the rough benchmarks. I thought my original post was getting long enough, so I just provided the raw zstd benchmarks that are restic...

I should add that I became aware of these additional zstd compression levels when I was testing out kopia. Kopia provides a compression benchmarking tool for all of the different...