David Abrahams

I had a similar question recently and did some experimentation. I didn't get much improvement in my model, but maybe yours will benefit more. In [my branch of segmentation_models](https://github.com/AllWashedOut/segmentation_models), unet...

I agree, there are really two behaviors being controlled by one setting. My ideal would be something like: Alarms [repeating, once, off] Menu beep [on, off]

Ok, I've done the much harder change of separating beeps and alerts. Menu beep [on, off] Alerts [repeat, once, off] https://photos.app.goo.gl/5KcHWpXttWe7o8Uh9

BTW I've been running this mod on my MK3s for weeks now. Things are working normally. I have muted the menu but left alarms repeating. When my filament runs out,...

Yes if you want this to move forward, someone will need to rebase and get a code review. I don't know the process to get a official code review so...

I just noticed that this param isn't even present in lightgbm 3.3.x which comes with synapseml 1.0.8. It must have been added to lightgbm in the 4.x era. So there's...

@Vonatzki did you ever find a work around for this issue? I have the same problem using LightGBMRanker on recent SynapseML builds. (old version 0.11.3 appears to be unaffected, or...

I believe that using bulk mode instead of streaming mode is a workaround.

I have also been struggling with vague crashes for the past week. How big is your training data? Millions of rows? If so, check whether training succeeds on a much...

Oh yes, the `limit` thing is real. I habitually call `train = train.limit(train.count())` before starting large training passes. It looks so dumb, but it legitimately prevents some undiagnosed Synapse bug....