on deepseek's r1

[Quinn Dougherty]

By now you've seen the DeepSeek model, which makes the cost of finetuning and running a SOTA model way cheaper. While the dual use / proliferation implications are scary, I'm writing to quickly flag that the opportunity space for GSAI is huge right now. Reach out if you want to talk about r1 finetuning for formal verification.

[Jacques Thibodeau]

If anyone is looking to do this, I’d love to talk. Not my current focus, but I want to eventually help automate mathematical formalization as described in Davidad’s programme. I’ve been eye-ing DeepSeekMath and DeepSeek-Prover since they came out for this reason.

From a LessWrong shortform:

I'm currently in the Catalyze Impact AI safety incubator program. I'm working on creating infrastructure for automating AI safety research. This startup is attempting to fill a gap in the alignment ecosystem and looking to build with the expectation of under 3 years left to automated AI R&D. This is my short timelines plan.

If you're interested in talking and getting more info, send me a direct email.

Thanks!

On Fri, Jan 24, 2025 at 9:05 PM Quinn Dougherty <quinn.do...@gmail.com> wrote:

By now you've seen the DeepSeek model, which makes the cost of finetuning and running a SOTA model way cheaper. While the dual use / proliferation implications are scary, I'm writing to quickly flag that the opportunity space for GSAI is huge right now. Reach out if you want to talk about r1 finetuning for formal verification.

--
You received this message because you are subscribed to the Google Groups "guaranteed-safe-ai" group.
To unsubscribe from this group and stop receiving emails from it, send an email to guaranteed-safe...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/guaranteed-safe-ai/CAGaOgARzfy69pJ2gCiD0g1bR9%3Dz3WKVc0JVsETtxiNH0HRwj3A%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

[Kris Carlson]

I'm very interested. Thanks Quinn. - Kris

To view this discussion visit https://groups.google.com/d/msgid/guaranteed-safe-ai/CANbxZb5dL45gtqyJ2m%3DLM1Hm6pGFx-gPWyvt%3DeyHXWKesWe8sA%40mail.gmail.com.