• [Blog] Thinking Machines Lab: LoRA is better to be applied to all weights, including MLP and MoE. On RL, LoRA performs on par with full training even with small ranks.

  • [Blog] Thinking Machines Lab’s LoRA API. Waitlist required.

  • [Announcement] ChatGPT has affiliate earnings now. Accompanying Agentic Commerce Protocol.

  • [Announcement] AgentKit. Also Agents SDK (Website) and other tools for agent building. Notably, among other things, this includes automatic prompt tuning via an autorater.

  • [GitHub] pip install weightwatcher.

Have a great week!

Updated:

Comments