Andrej comments

Results 373 comments of


                                            Andrej

extract all times from all log files isn't working

Hi, strange that this is happening - you're right there should be timestamps following the underscore. Somehow the code is failing to get the timestamp, or compute it sometimes. I...

__LOCKEDSCREEN didn't catch a thing (Yosemite). 'loginwindow' did

Thanks for the report! I haven't written the OSX parts since I work on a Ubuntu machine, so I'm not very familiar with the code there. Is this fixable on...

Update logactivewin.sh

Strange, any idea why this would be different? What ubuntu are you using

Much faster Regex tokenization using c++ and ctypes

Hi, this looks awesome. Maybe make a separate repo and I'm very happy to link to it from this code base in the main Readme under extensions?

`minbpe-rs`: A pure Rust implementation of `minbpe`

submit a PR happy to merge

utilities for mixed-precision tests/benchmarks

This complicates dev/cuda quite a bit, with templates and macros, both a bit scary. What is the problem that it is trying to solve? Isn't it the case that our...

utilities for mixed-precision tests/benchmarks

will avoid for now, closing.

Adding GPT2 model evaluation on WikiText-103 with optional preprocessing in dev/model-eval/

Thank you @joeshmoe0112358 for looking into this, but it looks like we're basically not able to match the paper table. In that case I'd at least try to match Alec's...

Adding GPT2 model evaluation on WikiText-103 with optional preprocessing in dev/model-eval/

We are abandoning WikiText103 because it's a total mess. We'll instead look at one/few of ARC Easy / Challenge, Squad, Hellaswag, TriviaQA, LAMBADA. Closing.

GPU auto-detect capability for kernel builds

any reason we don't do this in main makefile too?