on the new device (android 11), the "load" button is not active after selecting a network for loading. I tried versions 1.18.0 and the latest 1.18.3 (baduk_ai__armeabi-v7a). what are the possible reasons? it works fine on android 7 devices
https://imgur.com/a/p4N6EW8

.bin is not a loadable model extension. Should be .bin.gz

thank you very much!
in this phone when downloading ".bin.gz" turns into ".bin"
after renaming it works

KataGo 1.13.0 is released, can BadukAI update it?

@lovego

Update to KataGo 1.13.2 is here: https://github.com/aki65/aki65.github.io/releases/tag/v1.18.7

wonderful! will it be for 32 bits?

I added a 32 bit version to release 1.18.7.

Be aware that .dlc files can't be loaded with the 32 bit version (Qualcomm doesn't support that).

I made a new version (1.19) available on play store https://play.google.com/store/apps/details?id=net.kir.baduk_ai and github https://github.com/aki65/aki65.github.io/releases/tag/v1.19.0 which wraps up all the changes in the pre-releases on github of the past months.

kata1-b18c384nbt-s7405796608-d3635431045 is very strong! Can you support it

@lovego

I put an optimized version of kata1-b18c384nbt-s7405796608-d3635431045 here: https://github.com/aki65/aki65.github.io/releases/tag/v1.19.1

By the way, in version 1.19 I have implemented "parallel evaluation" for the 18b networks, but it depends on the chipset if that really yields an increase in performance. Could you please try on Snapdragon 8 gen 2, if there is any speed difference when you tick "parallel evaluation" in settings ?

On my OPPO Find X6 Pro, the speed of this 18b weight is about 400n/s. After turning on parallel evaluation, the speed remains unchanged.

Thanks a lot!

On the latest Mi14Pro (8Gen3 processor) phone, the 18b weight cannot be loaded.
Looking forward to your good news~

There is a new version at https://github.com/aki65/aki65.github.io/releases/tag/v1.19.3 that works on SD 8 gen 3.

Sayuri will soon be stronger than LZ. and sees ladders better. can add Sayuri in BadukAI?

I have taken a look at Sayuri. Seems interesting, but porting it to android and optimizing the networks will be a lot of work. I will try to find the time to do so, but I can't promise anything.

KataGo v1.14.0 is released! This version adds support for a new larger and hopefully stronger "b28" neural net that will likely be ready within the next months. It also has a large number of more minor bugfixes and changes.

The latest version on github includes KataGo 1.14: https://github.com/aki65/aki65.github.io/releases/tag/v1.19.7

Kata1-b18c384nbt-s8526915840-d3929217702 is very strong, can you update it~

I optimized kata1-b18c384nbt-s8040575488-d3801933292 just a few weeks ago (available at https://github.com/aki65/aki65.github.io/releases/tag/v1.19.6) and it's only 10 ELO weaker than kata1-b18c384nbt-s8526915840-d3929217702, so that one should be sufficient.

Can you optimize these two weights? I want to test them on my mobile phone

kata1-b18c384nbt-s9131461376-d4087399203
b28c512nbt-s5034669312-d4191783685 (https://drive.google.com/drive/folders/ ... sp=sharing)