zza

Hi, I used a few different datasets, but imagenet is the main one. To train you just need a reasonably varied set of images, even only 100 photos. Unfortunately sometime...

I've updated the dependancies, so you might want to rebase. Changing FACTOR should be all that is required, although it means that the built-in res/*.rsr files have to be removed...

Regarding nothing happening, that's worrying. I've had one other report of that happening on OSX: https://github.com/millardjn/rusty_sr/issues/3. Unfortunately I can't test on OSX. Could you tell me what gets printed to...

Thank you very much for doing this testing, its good to know where the problem is! I've updated `matrixmultiply_mt` to use the `parking_lot` crate which supplies alternative Condvar and Mutex....

Crates.io "Last Updated" is wrong, not sure why. Version 0.1.4 is the new one. That training setup looks correct :). The PARAMETER_FILE argument is where the weights learned by the...

Hi! The intended license is MIT though I have only put that in the cargo.toml file. I'll get a LICENSE file up shortly but please feel free to consider it...

Hi, If you want to change the training objective then `CrossEntropy` should be swapped with `Mse`. The `Prediction` loss is only used for counting how many correct guesses are made...

This is super cool, I should have a chance to review it soon.

Thanks, I've made those changes. I ran into issues when removing `"use_1200bps_touch": true` without also removing `"wait_for_upload_port": true`. Seems like either both or neither were required so I've removed both....

I think the problem in @kbillings's example is due to the `ContentLength` header is being set to `0` by `set_headers(&state, &mut response, Some(::mime::TEXT_PLAIN), None);`, preventing continued streaming of the response....