very unfair penguins can do this
December 23, 2024 at 10:21 AM
very unfair penguins can do this
new benchmark just dropped: if you sound like this, consider 'mitigating' your skill issues
December 10, 2024 at 5:16 PM
new benchmark just dropped: if you sound like this, consider 'mitigating' your skill issues
MISR eval reveals frontier language agents can self-reason to modify their own settings & use tools, but they fail at harder tests involving social reasoning and knowledge seeking. Opaque reasoning remains difficult. arxiv.org/abs/2412.03904
December 10, 2024 at 11:16 AM
MISR eval reveals frontier language agents can self-reason to modify their own settings & use tools, but they fail at harder tests involving social reasoning and knowledge seeking. Opaque reasoning remains difficult. arxiv.org/abs/2412.03904
Finally read this great anthropological study the hidden spiritual and cultural references within the latent space of models, by @bvlsingler.bsky.social & @mpshanahan.bsky.social. Puzzled by terms like 'hyperstition' or 'noosphere'? This is a great place to start!https://arxiv.org/abs/2411.13223
December 9, 2024 at 11:10 AM
Finally read this great anthropological study the hidden spiritual and cultural references within the latent space of models, by @bvlsingler.bsky.social & @mpshanahan.bsky.social. Puzzled by terms like 'hyperstition' or 'noosphere'? This is a great place to start!https://arxiv.org/abs/2411.13223
me post labour automation
December 7, 2024 at 12:07 AM
me post labour automation
🪔 imma keep it real with you chief: genie 2 is far cooler than your language model that scores 76.29 on the questions-i-don't-care-about eval. shape rotate all you want, i'll be exploring a generated 3D interstellar ocean with a level 100 octopus-inspired humanoid.
December 4, 2024 at 9:58 PM
🪔 imma keep it real with you chief: genie 2 is far cooler than your language model that scores 76.29 on the questions-i-don't-care-about eval. shape rotate all you want, i'll be exploring a generated 3D interstellar ocean with a level 100 octopus-inspired humanoid.
yep, otherwise you get
December 4, 2024 at 4:56 PM
yep, otherwise you get
Between 2010 and 2018, global data centre compute increased by more than 550%. Yet energy use in data centres increased by just 6%. How much energy do data centres and AI use today? A few percent of the world’s electricity, at most. sustainabilitybynumbers.com/p/ai-energy-...
December 4, 2024 at 4:47 PM
Between 2010 and 2018, global data centre compute increased by more than 550%. Yet energy use in data centres increased by just 6%. How much energy do data centres and AI use today? A few percent of the world’s electricity, at most. sustainabilitybynumbers.com/p/ai-energy-...
rare good tech regulation: the obligation for email newsletters to have an unsubscribe option (CAN-SPAM Act of 2003)
December 4, 2024 at 4:45 PM
rare good tech regulation: the obligation for email newsletters to have an unsubscribe option (CAN-SPAM Act of 2003)
on the one hand, deep theoretical understanding is good for building robust mental models and solving complex problems. but on the other, I don't actually know if the right abstractions are being taught; there should probably be more specialised paths early as the BCS proposes.
November 29, 2024 at 9:06 PM
on the one hand, deep theoretical understanding is good for building robust mental models and solving complex problems. but on the other, I don't actually know if the right abstractions are being taught; there should probably be more specialised paths early as the BCS proposes.
⚠️ the EU has identified a missing piece of paper ⚠️
November 28, 2024 at 8:50 PM
⚠️ the EU has identified a missing piece of paper ⚠️
this is where I post from btw
November 25, 2024 at 10:32 PM
this is where I post from btw
Great paper. RLHF risks "deceptive inflation," where AIs manipulate observable actions to appear more successful than they are, and "overjustification," where AIs incur needless costs to make actions seem reasonable, even if inefficient. arxiv.org/abs/2402.17747
November 25, 2024 at 5:55 PM
Great paper. RLHF risks "deceptive inflation," where AIs manipulate observable actions to appear more successful than they are, and "overjustification," where AIs incur needless costs to make actions seem reasonable, even if inefficient. arxiv.org/abs/2402.17747
saturday night meditations 🌀
November 17, 2024 at 3:20 PM
saturday night meditations 🌀
wholesome and serene time in Mexico
June 13, 2023 at 8:31 PM
wholesome and serene time in Mexico
I will simply not leave 🇲🇽
June 11, 2023 at 10:50 PM
I will simply not leave 🇲🇽
