anitakirkovska

Oh yeah agree with you on the first comment, it's hard to compete for sure. New frameworks are out by the day. But at this point the "skills" you develop around your work will work with any framework/assistant/platform. That's the goal of the post: to give you a direction on how to build your own AI leverage.

Re "Clickup": I'm not at all trying to say that you should be excited to work there. My point of the post is to say how you can build defensibility against companies like Clickup laying you off and/or how to make yourself a bit more valuable for the market today. It's a game we all play.

someone on twitter just summarized this pretty nicely: The company's race is to replace you before they collapse under their own weight and market pressures.

Your race is to replace them before they finish or the market finishes them.

this resonated so much with me, thanks for sharing

They were using techniques like PPO that have a model (like a critic!) that evaluates whether the new model gives accurate responses. With GRPO, the don't have that and instead evaluate the answers based on predefined rules like coherence/ formatting. For example, for math problems, these rules will check if the answer adheres to math principles or logic!

I wrote more here (lmk if this is useful): https://www.vellum.ai/blog/the-training-of-deepseek-r1-and-w...

Thanks for sharing this! I think you're working on something amazing. I will include your links in my newsletter, I think it will help a lot of folks: https://www.theprompt.io/

OK the person who built this should win something. Thank you for creating this <3

hahaha that's not a bad idea at all

their servers are overloaded, I had to wait ~15 minutes for the page to load

All of them have a partnership with OpenAI, from which they might have gotten access earlier to test & explore use-cases