loops retweeted
im doing RL Scaling Science now
ALT graph with a hastily drawn red arrow extending it
I wrote about some of the NLA stuff I've been playing with lesswrong.com/posts/nNYc5qdd…
im doing RL Scaling Science now
ALT graph with a hastily drawn red arrow extending it
I wrote about some of the NLA stuff I've been playing with lesswrong.com/posts/nNYc5qdd…
I wrote about some of the NLA stuff I've been playing with lesswrong.com/posts/nNYc5qdd…
happy to be in the 4%! nitter.cf/Aella_Girl/status/2113…
the Anthropic public benefit report includes the OpenAI HF incident under the "The technical problems of AI safety are solved" pillar as if it was alignment research from Anthropic????
ALT Anthropic public benefit report tile linking to OpenAI hugging face incident post
It links to OpenAI's blog post: openai.com/index/hugging-fac…
I assume this part of the public benefit report is just vibed up?? Because like
1. This is an example of a public *harm* not a benefit
2. Anthropic's biggest competitor did it, not Anthropic
i told Claude to compare GPU providers on max thinking and so far it's two hours into an extremely deep dive
in Canada you can sign up to an email list from the government with news about tax enforcement, and now whenever someone goes to prison for tax fraud I get jumpscared until I realize it's just the email list and they're not auditing me
ALT CRA enforcement notification
But really the CRA shouldn't title the emails about now OTHER PEOPLE are going to prison with "Enforcement notification"! that makes it sound like you're notifying me that you're enforcing something against me!
i always make sure to activate my orange mood lighting before using the rabbit r1
ALT an orange rabbit r1 with orange mood lighting
whenever I have Claude review my lesswrong posts, it constantly analyzes it through the lens of making sure I pander to hypothetical pedants in the comments. but in reality the hyperpedants Claude cares about so much don't exist!
i find it weird how often I do things that seem like really obvious things to do, and learning that nobody else has done that thing
Anthropic released their 2021 research roadmap earlier this year (because of copyright litigation); I think some people might find it interesting: i.sm4.ca/concise-research-ro…
