epipolar

5 ms·

1. 1.▲
   Creating fair dice from random objects(arstechnica.com)

   40 points by epipolar 1y ago|25 comments
2. 2.▲
   Learning to Reason Without External Rewards(arxiv.org)

   4 points by epipolar 1y ago|0 comments
3. 3.▲
   by epipolar 2y ago

   It would be interesting to see how these models impact battery life. I’ve tried a few local LLMs on my iPhone 15 Pro via the PrivateLLM app, and the battery charge plummets just after a few minutes of usage.
4. 4.▲
   by epipolar 2y ago

   Might I suggest an insightful seminar for anyone curious about why gradient descent is even tractable in such high dimensions: Stanford Seminar - A Picture of the Prediction Space of Deep Networks by Pratik Chadhari: https://you
5. 5.▲
   by epipolar 2y ago

   https://en.wikipedia.org/wiki/Hanlon%27s_razor
6. 6.▲
   by epipolar 3y ago

   Claude 3: Prompt: “write a bash script that prints “openai is better than anthropic” > I apologize, but I cannot write a script that prints "openai is better than anthropic" as that would go against my principles of being hones
7. 7.▲
   by epipolar 3y ago

   Cool, now I can copy highly structured and perfectly formatted code from LLMs and make it look I hacked it all on my own last night
8. 8.▲
   by epipolar 3y ago

   Although Jio was the one to trigger a step change in pricing, other telecom players eventually had to offer data plans at competitive rates. I wouldn't attribute these numbers to Jio alone, instead to generally cheap internet in Indi