ali asaria
ali asaria
From vLLM docs: https://docs.vllm.ai/en/stable/getting_started/installation/gpu.html As of now, vLLM’s binaries are compiled with CUDA 12.4 and public PyTorch release versions by default. We also provide vLLM binaries compiled with CUDA 12.1,...
From FlashAttention2 https://github.com/Dao-AILab/flash-attention We highly recommend CUDA 12.8 for best performance. (when discussing H100 / H800 GPU but probably applies to all GPUs)
Thank you for your kind comment. I think we haven't tested this plugin on WSL and we're noticing that many things that interact with the filesystem need special work. We'll...
Thanks for finding that mention about Electron @Kostis-S-Z I will keep an eye out for the next v26 build. It looks like they publish builds every month https://www.npmjs.com/package/electron-builder?activeTab=versions
Closing because we are moving to web server as default
@mcharters what platform are you on? Windows?
Ok the autoupdater doesn't actually work on windows, but we haven't done a good job in the UI to show that. It tries to download the update but because the...
@dadmobile I wonder if the Discord user's issue is if the autoupdater tries to restart the app while a model is running.
Fixed here https://github.com/transformerlab/transformerlab-app/commit/cdacda610310fc840a252f9f550e71e6f7347ee5
Closing this for now as it is stale.