Finally catching up on some blog posts I saved for later. This one is VERY interesting. I never really dedicated that much of time to skills, and that was a mistake. Easy to read, well thought out and directly actionable in your everyday tasks! huggingface.co/blog/upskill

we need to get real and move fast about training our own agents. orgs, teams, and individuals all need to be improving their agent. not waiting for the next API to be released (or not). this is a practical hands-on session where I'll show you how to train (and improve) your own agent based on traces. it's supported by tutorials and a repo, starts ofs simple, and works today. over the coming weeks, I'll add more advanced classes working up to RL environments and harnesses. If you're building around training agents, jump on stream, and demo your work.

We now support rich formatting for all chatbots. Tables, nested lists, inline media, formulas, headers and more — right in Telegram messages. 🔨 Start building! Docs: core.telegram.org/bots/api#r…

@grok help me better understand the pi concept of trees in its session management. What happens when you fork a new tree branch from an earlier message? Will that basically create a new session, with all the messages so far in its context? Can you then work of both tree branches independently? Or is it more like a rollback thing where it will undo any tool calls that might have happened before? Also, could you explain what the effect on kv cache would be when creating a new tree branch? Would the model need to keep two separate caches or can the first common part of the cache be reused (is that perhaps prefill cache purpose?) note, im an amateur, be gentle with me