@pleometric there are no ml ops which are immensely flops bound but not memory bandwidth bound, and people who say image models or video models need this are bad at ml architectures and guessing the near future
2
0
4
48
Just finished implementing the last feature of my game. ☺️
On track for a Nov. 1 launch!
On track for a Nov. 1 launch!
6
0
48
1,072
@pleometric can you make a version of this that also explains the navier-stokes conjecture
2
0
5
111
with some scaffolding, very cheap open models can get the opus 5.5 animation look without much trouble...
now the interesting part will be to see how far you can push them into a polished state
now the interesting part will be to see how far you can push them into a polished state
4
1
53
3,683
@pleometric Kurzgesagt is the gold standard for infographic animations, it’s one of my personal AGI benchmarks. I will be floored the day it can look as good as theirs.
2
0
3
36
@hyrevayn I should note that this is good looking simply because I'm optimizing for speed
0
0
1
36
@pleometric currently on a quest to figure out comfyui with the help of one
1
0
2
67
@kojinglick the question comes by way of people telling me prices are doubling and will soon be tripling
2
0
2
68
Testing Terra's computer use by having it edit pictures in photoshop. It's pretty decent but I keep having to tell it what to do step by step, it's a bit lazy in that respect.
Considering how cheap it is compared to Astra I'd say it's a good use case.
Considering how cheap it is compared to Astra I'd say it's a good use case.
2
0
27
4,807
why would I do such a neolithic thing as editing pixels by hand? Well sometimes I need the tiniest of edits that are faster like this.
0
0
11
5,262
@pleometric If it's for inferencing (especially video models) and you got a few thousand dollars to spend, then I think it's worth it
1
0
1
40
Pleometric retweeted
"haha, look, this neuron makes qwen homicidal, this neuron makes qwen act like-"
Qwen's POV while inferencing:
Qwen's POV while inferencing:
18
125
2,502
41,198
genuinely stupid, how does the platform not detect 5k simultaneous followers from fresh accounts with zero previous interactions?
7
0
86
3,248
@pleometric wow pleo-kun, that's a lot of followers you have there
1
0
3
92
♱ ⋆｡˚ 𝔪𝔞𝔱𝔥 𝔦𝔰 𝔡𝔢𝔞𝔡 ˚｡⋆ ♱ 🥀
♱ sébastien bubeck x terence tao edit ♱
♱ ⋆｡˚ 𝔬𝔭𝔢𝔫𝔞𝔦 𝔨𝔦𝔩𝔩𝔢𝔡 𝔪𝔞𝔱𝔥... 𝔧𝔬𝔦𝔫 𝔪𝔢 𝔦𝔫 𝔡𝔢𝔞𝔱𝔥? ˚｡⋆ ♱
♱ sébastien bubeck x terence tao edit ♱
♱ ⋆｡˚ 𝔬𝔭𝔢𝔫𝔞𝔦 𝔨𝔦𝔩𝔩𝔢𝔡 𝔪𝔞𝔱𝔥... 𝔧𝔬𝔦𝔫 𝔪𝔢 𝔦𝔫 𝔡𝔢𝔞𝔱𝔥? ˚｡⋆ ♱
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math
15
47
436
31,817
@pleometric i have this problem too, wonder why they want to nuke someone else’s reach
1
0
2
236
@lostnlatency I think some people do it just to mess with an account they don't like
0
0
3
80
4,964 bots, all created Oct 26–27, the 4,964 newest followers. They all followed me one hour ago, they are mostly connected to a few crypto products and kpop accounts. Korean and Japanese user handles.
genuinely stupid, how does the platform not detect 5k simultaneous followers from fresh accounts with zero previous interactions?
5
1
56
1,496