Nikolay Bogoychev

@dzarukin we are doing `8-bit` neural network inference with intermediate `fp32` results. We find that the bias term compresses poorly and this is why we keep it in `fp32` format....

@mgouicem the heaviest computation for any neural network inference is the output layer which involves multiplication of the type `1x512 * 512x32000 + 1x32000`. However the vast majority of those...

@mgouicem Just to clarify, the above case where the leading dimension M is `1`, is an idealised case where we care the most of about latency (mini-batch 1, beam size...

@rezad1393 could you open this issue on https://github.com/XapaJIaMnu/translateLocally/issues Could you also do an `ldd translateLocally`? Also can you access https://translatelocally.com/models.json from Iran? If not could you paste the output of...

I assume you're building and using archlinux? The only difference between my build and your build is that mine also links to `/usr/lib/libQt6PrintSupport.so.6`, which is part to `qt6-base`. Not sure...

@rezad1393 I'm Dheart from arch forums, you are talking to the same person. I use arch, btw.

Hi, We support external repos. You can add an extra repository through the settings menu and one such repository is OpusMT: `https://object.pouta.csc.fi/OPUS-MT-models/app/models.json` In terms of Farsi, we have Farsi-English, but...

@sshleifer What version of MacOs are you using? I have tested the compilation with: ```bash xapajiamnu@dhcp-91-025 marian-dev % sw_vers -productVersion 10.15.3 xapajiamnu@dhcp-91-025 marian-dev % clang++ --version Apple clang version 11.0.3...

> I upgrade to 10.15.4 and it didn't work -> identical error message. > First offending line: > > ```shell > [ 94%] Linking CXX executable ../marian-conv > cd /Users/shleifer/marian/build/src...

@ugermann can we add those to ignore list?