matusp

21 ms·

331.
by matusp 3y ago
In short term yeah, it will probably be too muddy. But it will dry over time and in the long term this unlocked a completely new axis of attack for Ukrainians. Similarly in the downstream regions. The water will go away eventually and it wi
332.
by matusp 3y ago
It will also make it easier for Ukraine to attack upstream when the reservoir is empty. In places where there are virtually no Russian defenses. So that's a win for Ukraine.
333.
by matusp 4y ago
Is Troika! popular? I had a 1-on-1 session of Troika! and I really enjoyed the vibe, but I found it hard to find people interested in this particular style of play.
334.
by matusp 4y ago
Shameless self-promotion, I have recently written a blog about this. ChatGPT actually is usually a little bit worse than older models for these classical NLP tasks. Of course the older models are not zero-shot. https://www.opensa
335.
ChatGPT Survey: Performance on NLP Datasets(opensamizdat.com)
3 points by matusp 4y ago|1 comments
336.
by matusp 4y ago
Abstract: The popularity of ChatGPT and its various impressive capabilities lead some people to believe that it is a significant step forward for linguistic capabilities over existing systems, that the field of NLP will soon be consumed by
337.
by matusp 4y ago
> Computer proliferation is not an existential risk. AI is created and run on computers so AGI risk must by definition be a subset of computer proliferation risk
338.
by matusp 4y ago
Maybe I am missing something, but I don't entirely understand what exactly is the novelty of RLHF in this case. RL is a long-standing field and it is being trained based on human inputs for decades. Even the first paper you mention as
339.
by matusp 4y ago
> The alignment branch (responsible for developing RLHF->ChatGPT) RLHF was not conceived by AI Alignment people. Using RL to train generative models was a thing even ten years ago. Now they finally made it work on scale. This has noth
340.
by matusp 4y ago
I don't think they are taken that seriously. They are writing their op-eds and fiery papers, but the field marches forward nevertheless. But I agree that the whole fields seems like ivory-tower types trying to hijack the latest hottest
341.
by matusp 4y ago
In what regards has it deteriorated so much?
342.
by matusp 4y ago
Is there a strategy to nullify competition in the mails? They simply stated that they are worried about the coming competition, which is completely normal for any business.
343.
by matusp 4y ago
And what is the true legal limit?
344.
by matusp 4y ago
You can argue that there might be some similarities or analogies to be made. But that's it. The book is actually very irrelevant and it had literally no impact on how these new systems were created and conceptualized.
345.
by matusp 4y ago
I think it's quite common for competitive programmers and code golfers to have their own language, often inspired by APL. They often do not require any ecosystem as they are supposed to only solve puzzles, so it's really easy to d
346.
by matusp 4y ago
Yeah, this was my first reaction as well. It's incredibly hard to tell how much signal is in a metric like this. I can imagine two papers or two grant proposal with significantly different degrees of novelty that have the exact same re
347.
Images and Words: AI in 2026(nintil.com)
2 points by matusp 4y ago|1 comments
348.
by matusp 4y ago
> This whole saga has done irreparable damage to crypto. There were so many similar sagas in crypto, that this is not surprising at all by now.
349.
by matusp 4y ago
Did eliminating the accessibility team made the twitter worse for blind people somehow? Was twitter getting better for blind people over the years?
350.
by matusp 4y ago
I am genuinely curious. Does it have anything comparable to the following? - pytorch or tensorflow - pandas - numpy - scipy - matplotlib - huggingface transformers - jupyter
351.
by matusp 4y ago
The "Code leakage" problem can easily be solved by using views. Or am I missing something?
352.
by matusp 4y ago
Yep, the data requirements are especially condemning. We have recently scaled from using Wikipedia as a training corpus to basically the entire Internet. And it helped a lot. But where are we going to scale next? There is simply not enough
353.
by matusp 4y ago
On the other hand, male experts are victims of mangettingcalledretarded.
354.
by matusp 5y ago
It is obviously a matter of context. Russian army in Kyiv is indeed able to regroup and attack again. Iraqi army was by all means a defeated force that was running for its life into Iraq where civil war just started. The risk of it attackin
355.
by matusp 5y ago
What about Oroville dam repairs? It was a pretty successful and fast engineering project. I think that America can still build quickly if she wants to. But most of the projects are not that important.
356.
by matusp 5y ago
Do you really need a reader that is too lazy to click on your article? And from the reader point of view, I assume that whatever is written in twitter threads is not really that important, since the authors could not be bothered with writin