AirLLM uses layer-wise inference to run 70B models on 4GB GPUs and 405B on 8GB, supporting Llama, Qwen and Mistral, now open source.
Peaked at #2
@oliviscusAI
Oct 11, 12:22
Thread highlights 10 trending free GitHub repos this month, including VoiceStudio and ppt-master, with star counts.
Peaked at #1
@brenzhills
Oct 11, 10:33
Rohan Arun says PR #346 ported Design T to the p=10 bit word, raising the integer multiplication bound by ~2%.
Peaked at #2
@RohanArun
Oct 11, 10:13
Omarchy's next version keeps Caps Lock default, moving compose to Shift+Caps Lock.
Peaked at #2
@omarchy
Oct 11, 10:02
Joshua Warren released metal2vk, an open-source tool translating Metal GPU code to Vulkan, passing 598 of 610 real programs so Mac engines run on Linux.
Peaked at #3
@JoshuaSWarren
Oct 11, 07:47
Jesse Pollak recommends installing shadcn/BaseUI for Base vibe-coded apps so products look non-vibecoded.
Peaked at #3
@jessepollak
Oct 11, 04:34
Discussion 0
Sign up Sign in to join the discussion
No comments yet. Start the conversation.