markisus

5 ms·

1.
by markisus 2mo ago
I have no problem with using agents for chess engine dev but I find this offer from the author unsatisfactory. > if there is a particular optimization of yours that we've used unfarily and/or not credited, tell me which one … I
2.
by markisus 2mo ago
The Anthropic employee typed “Make the next version of Claude. It must score better on all our benchmarks. Make no mistakes. Don’t exceed the training budget.” She made sure to turn on —-dangerously-allow-all and pressed enter. The Agent sp
3.
by markisus 2mo ago
The post makes it seem like ONNX is CPU only. I've used ONNX runtime to run models on Nvidia GPUs. The runtime can even dispatch to TensorRT. I'm not sure what the performance is on Apple hardware so maybe that was the motivation
4.
by markisus 2mo ago
This is every cool research. Do you have any idea why the authors chose Z80 as the program language? I have seen other studies in the same spirit that use simpler toy languages like Brainfuck ( https://arxiv.org/abs/2406
5.
by markisus 2mo ago
It's funny that the machine intelligences that arose out of the bitter lesson cracked this benchmark by meticulously modeling each individual case with rule based approaches. When humans tried making AI through rule based approaches, m
6.
by markisus 5mo ago
Claude code has some basic security features like asking for user confirmation for bash commands, or restricting commands to the current directory. If these features are not being code reviewed, what assurances do we have that they actually
7.
by markisus 5mo ago
The article mentions Triton for this purpose. I don’t think you will get maxed out performance on the hardware though because abstraction layers won’t let you access the fastest possible path.
8.
by markisus 5mo ago
Don’t get hung up on the Claude part. We already know an algorithm that produces AGI: evolution. We don’t have the compute to run this algorithm because it requires simulating the whole Earth. But with enough compute, it becomes possible.
9.
by markisus 6mo ago
But this ruling will surely set precedent for other cases where AI is used. It may cover the case of AI summaries as well.
10.
by markisus 6mo ago
If you sell and then rebuy isn’t that considered a wash trade and therefore exempt from taxes?
11.
by markisus 6mo ago
Can you explain what asymmetric means? Are you saying that this forced rebalancing will be large enough to cause a large price drop on other stocks? Let’s just think about any stock in particular, eg stock ABC. If I am an active investor, I
12.
by markisus 6mo ago
QQQ is problematic because it’s influenced by strange back room dealings with Space X, if the article is to be believed. VTI is different. It literally tracks all public stocks, weighted by market cap so no such manipulation is possible. If
13.
by markisus 6mo ago
I’m trying to understand the mechanics here. I get that SpaceX and Nasdaq are in cahoots to get SpaceX bundled with a bunch of other stocks (and that bundle is called QQQ?) But why must retail investors hold this bundle? If I’m holding now,
14.
by markisus 6mo ago
Like what you mean when you say “frames per second” in a video game. The image that is finally presented to the user, and then quickly thrown away and rebuilt, again and again to give the illusion of motion.
15.
by markisus 6mo ago
I also seem to be developing an immune response to several slopisms. But the actual content is useful for outlining tradeoffs if you’re needing to make your Python code go faster.
16.
by markisus 6mo ago
I wish there were more details on this part. > Missing @cython.cdivision(True) inserts a zero-division check before every floating-point divide in the inner loop. Millions of branches that are never taken. I thought never taken branches
17.
by markisus 7mo ago
Yeah if your program has a natural notion of a "frame" (eg most video games), you can do memory management by simply incrementing an integer (bump allocation). At the end of your frame, you reset the integer to zero. You can'
18.
by markisus 7mo ago
I'm not sure everyone knew that gaps reflected incorrect reasoning. It would have been natural to assume that all infinite sets were qualitatively the same size, since uncountable infinity was not an idea that had been discovered yet.
19.
by markisus 7mo ago
> Before their papers, mathematicians had assumed that even though the number line might look like a continuous object, if you zoomed in far enough, you’d eventually find gaps. I'll try to interpret this sentence. We all have some m
20.
by markisus 7mo ago
Of course a contractor could not decide to unilaterally shut off their missile system, because that would be a contract violation. A contractor may try to negotiate that unilateral shut off ability with the government, and the government sh
21.
by markisus 7mo ago
I felt the same. > This isn’t a sign of incompetence. It’s a sign that you’re actually doing real work. > Your value as an engineer isn’t in executing the same playbook forever. It’s in expanding the domain of what’s possible. > Th
22.
by markisus 7mo ago
This really reminded me of the first part Flowers for Algernon. The main character undergoes a treatment which improves is intelligence and the story is narrated via a series of diary entries which become successively more fluent and sophis
23.
by markisus 9mo ago
I’ve confirmed this on my iphone as well. Using swipe, no space bar after kill: Kill maps Jill myself Jill myself Using swipe, manually pressing space bar after kill: Kill mussels Kill mussels Kill mussels
24.
by markisus 9mo ago
On the contrary, I think it demonstrates an inherent limit to the kind of tasks / datasets that human beings care about. It's known that large neural networks can even memorize random data. The number of random datasets is unfatho
25.
by markisus 9mo ago
Each fine tune drags the model weights away from the base model in a certain direction. Given 500 fine tune datasets, we could expect the 500 drag directions to span a 500 dimensional space. After all, 500 random vectors in a high dimension
26.
by markisus 9mo ago
Where do lock free algorithms fall in this analysis?
27.
by markisus 9mo ago
Their VGGT, Dinov3, and segment anything models are pretty impressive.
28.
by markisus 10mo ago
The problem becomes complicated once the large discrete objects are not actuated. Even worse if the large discrete objects are not consistently observable because of occlusions or other sensor limitations. And almost impossible if the large
29.
by markisus 11mo ago
The authors somewhat address your questions in the accompanying paper https://arxiv.org/abs/2410.24206 > We emphasize that the central flow is a theoretical tool for understanding optimizer behavior, not a practical
30.
by markisus 11mo ago
First I thought this would be just another gradient descent tutorial for beginners. But the article goes quite deep into gradient descent dynamics, looking into third order approximations of the loss function and eventually motivating a con

More ›