aoeusnth1

8 ms·

121.
by aoeusnth1 2y ago
Well, if physicalism is true then consciousness is a phenomenon of quantum fields, which span the universe. So yes, stretching the definition of creature, this could be interpreted as literally true.
122.
by aoeusnth1 2y ago
If you're skimming a text to answer a specific question, you can go a lot faster than if you have to memorize the text well enough to answer an unknown question after the fact.
123.
by aoeusnth1 2y ago
Can you give an example where you would need to adjust tools dynamically based on context? Is that for all tools or just for some? For example, why does a “Google search” tool need to change from context to context?
124.
by aoeusnth1 2y ago
If you are a tool provider, you need a standard protocol for the AI agent frontends to be able to connect to your tool.
125.
by aoeusnth1 2y ago
Does aider have support for prompt caching? That can reduce the cost significantly for many-turn sequences.
126.
by aoeusnth1 2y ago
One of the most surprising things about this announcement is that it didn't happen during business hours where the insiders could buy call options before hand.
127.
by aoeusnth1 2y ago
As far as you know, AI labs are doing E2E RL training with running code in the loop to advance the model's capability to act as an agent (for cursor et al).
128.
by aoeusnth1 2y ago
There is 0% of the exam which is trivially answerable. The entire point of USAMO problems is that they demand novel insight and rigorous, original proofs. They are intentionally designed not to be variations of things you can just look up.
129.
by aoeusnth1 2y ago
2.5 pro nails each of these calculations. I don’t agree with Google’s decision to use a weak model in its search queries, but you can’t say progress on LLMs in bullshit as evidenced by a weak model no one thinks is close to SOTA.
130.
by aoeusnth1 2y ago
2.5 pro scores 25%. It’s just a much harder math benchmark which will fall by the end of next year just like all the others. You won’t be vindicated.
131.
by aoeusnth1 2y ago
I am continuously surprised by the neverending parade of white cars on the roads.
132.
by aoeusnth1 2y ago
One more N=54 trial makes absolutely no difference to the mountain of evidence in favor of creatine. The news article should reference meta studies, not chasing headlines from the drip drip drip of meaningless small studies
133.
by aoeusnth1 2y ago
Of course, this is flagged by the reactionary contingent on this site.
134.
by aoeusnth1 2y ago
They said "all proprietary" models, not "all but 1 proprietary, non-thinking" models. It doesn't beat all the models! It's pretty good, especially nice since it's open source, but it's not going to be
135.
by aoeusnth1 2y ago
No it doesn’t.
136.
by aoeusnth1 2y ago
How about fog and rain?
137.
by aoeusnth1 2y ago
I think if they could have been on top earlier, they would have been. They’ve been struggling to catch anthropic and OpenAI’s lead and they finally did it (for now), probably due to TPU superiority plus some secret sauce of some kind. Good!
138.
by aoeusnth1 2y ago
Ok, thanks for clearing that up.
139.
by aoeusnth1 2y ago
Am I correct in assuming that accuracy < using correct edit format? i.e. it made mistakes in 27% of the problems, 11% of which were due to (at least) messing up the diff format? In which case, google should be working on achieving better
140.
by aoeusnth1 2y ago
It should be in the "reasoning" category, right? (still topping the charts there)
141.
by aoeusnth1 2y ago
I'm guessing it should be enabled eventually. @logankilpatrick thoughts?
142.
by aoeusnth1 2y ago
I think it's because of the big jump in coding benchmarks. 74% on aider is just much, much better than before and worthy of a .5 upgrade.
143.
by aoeusnth1 2y ago
I agree, but GPS for knowledge work does sound like something that will pan out to be a big industry, right?
144.
by aoeusnth1 2y ago
I guess they don’t do ultras anymore, but where was the announcement for it? What other announcement was vaporware?
145.
by aoeusnth1 2y ago
Because ai is following a predicable trend of exponentially increasing task length @50% probability: https://x.com/METR_Evals/status/1902384481111322929 If we are the top of an s-curve, the recent samples on that
146.
by aoeusnth1 2y ago
Well, is your claim that's it's isnt working? Perhaps after Tesla is bankrupt 6 years from now, tycoons with political ambitions will be less likely to adopt extremist political stances. It doesn't need to make sense to you,
147.
by aoeusnth1 2y ago
What kind of overhead? I’m curious if there’s data about this because I hadn’t heard that 1.1 was better for the data center.
148.
by aoeusnth1 2y ago
"Weights available"
149.
by aoeusnth1 2y ago
Looking at every other benchmark, it's significantly behind typical big models from a year ago (Claude 3.0, Gemini 1.5, GPT 4.0). I think Google must have extensive LMArena-focused RLHF tuning for their models to juice their scores.
150.
by aoeusnth1 2y ago
Where on the statcounter page does it show that people are using their default browser? You're making a lot of claims about Chrome commonly being the default browser without linking to any evidence of your correction of this "comm

More ›