Akira Kawata

I'm working on test_pareto_front.py.

Thank you. But it doesn't work for me. ``` julia> using FFMPEG.FFMPEG_jll julia> run(`$(ffmpeg()) -v 0 -framerate 15 -i /home/akira/materials-for-FFMPEG/%06d.png -i /home/akira/matrials-for-FFMPEG/palette.bmp -lavfi paletteuse=dither=sierra2_4a -loop 0 -y anim_fps15.gif`) ERROR: failed...

Further explanation: - Calculations of bias_correction are different - The timings of adding epsilon are different. So we cannot use the same epsilon for Pytorch and ONNX.

Hi, > 1. Is there a missing implementation that needs to be included? Umm... I confirmed `sudo docker build . --network=host` still working. Which OS are you using? Can you...

I moved unused script to misc directory in https://github.com/akawashiro/sloader/pull/42/files.

I'm sorry for being late. I'm debugging on this branch using debian 12.2 image. https://github.com/akawashiro/sloader/pull/43

The error is reproduced https://gist.github.com/akawashiro/c7ad091abe2fef5cdc1d35c60b4dada8.

> Should I re-run the generate script on my machine? No, I think the cause of this error is the difference of libc of ubuntu and debian.

Sorry, I cannot fix this problem soon. Just tweaking libc_mapping.cc doesn't fix this. I'm approaching this problem from completely different way which uses custom built glibc.

I confirmed this is due to the difference of glibc (glibc-2.36 and glibc-2.35).