yorwba

18 ms·

421.
by yorwba 7mo ago
The report doesn't even count downloads from ModelScope.cn, the Chinese HuggingFace competitor.
422.
by yorwba 7mo ago
Yeah, the LLM judge is a bit too gullible. GLM 5.1 here https://ndaybench.winfunc.com/traces/trace_585887808ff443cca... claims that onnx/checker.cc doesn't reject hardlinks, even though it does (and the model
423.
by yorwba 7mo ago
Yes, metamath uses a large collection of specialized but reusable building blocks, so it doesn't blow up exponentially. However, if you want to "just do gradient descent" on general trees built from a single universal primiti
424.
by yorwba 7mo ago
Yeah, of those 6 tasks, only "halluc-doc-http-handler" isn't within 1% of the previous result. 86.6% is 13/15 rounded down, so if they sampled 15 attempts for that task, the probability of getting 100% when the true succ
425.
by yorwba 7mo ago
I think you underestimate the amount of knowledge needed to deal with the complexities of language in general as opposed to specific applications. We had algorithms to do complex mathematical reasoning before we had LLMs, the drawback being
426.
by yorwba 7mo ago
OpenRouter has a list of providers: https://openrouter.ai/minimax/minimax-m2.5
427.
by yorwba 7mo ago
There are US-based companies offering inference for MiniMax models charging slightly less than what MiniMax charges. MiniMax themselves claim to be using data centers in the US. US companies training their own closed-weight models charge so
428.
by yorwba 7mo ago
When people criticize Aisle's methodology, they aren't "defending Mythos," they're bashing Aisle for their disingenuous claims.
429.
by yorwba 7mo ago
The first time was under the name Adriana Shayota: https://www.justice.gov/usao-sdca/pr/federal-jury-convicts-s...
430.
by yorwba 7mo ago
We don't even need to hypothesize that much on the irrelevant nonsense, since they helpfully provide data with the detected vulnerability patched: https://aisle.com/blog/ai-cybersecurity-after-mythos-the-jag... an
431.
Creating new knobs of control in biology(owlposting.com)
2 points by yorwba 7mo ago|0 comments
432.
by yorwba 7mo ago
Previous discussion: https://news.ycombinator.com/item?id=47716043 (764 points 5 hours ago, 384 comments)
433.
by yorwba 7mo ago
I guess it's based on Wikidata's "coordinate location" property. You can find the Wikidata item corresponding to a Wikipedia article using the "Tools" dropdown in the desktop view. (Or just search on Wikidata d
434.
by yorwba 7mo ago
Mozilla Corporation may have enough money, but they don't develop Thunderbird. If you used the donation form on this page, you didn't donate to Mozilla Corporation, but to the company developing Thunderbird. So all is fine.
435.
by yorwba 7mo ago
If Issue 1 is out, then where can it be read?
436.
by yorwba 7mo ago
This article seems to completely ignore the fact that languages change over time on their own and attributes all differences between Literary Chinese and Modern Standard Chinese to contact with European languages, which is rather excessive.
437.
by yorwba 7mo ago
Because the constitution only allows drafting men: https://www.gesetze-im-internet.de/englisch_gg/englisch_gg.h... The intersection of parties wanting to reinstate compulsory military service and parties supporting gen
438.
by yorwba 7mo ago
Western countries did not raise the standard of living using cash transfers, so of course looking at what they did won't tell you whether cash transfers are effective or not.
439.
by yorwba 7mo ago
It's a point update to the closed-weight Qwen3.5-Plus. Of course there are no weights. Alibaba has consistently not released weights for their best models.
440.
by yorwba 7mo ago
It's not a big problem in practice since you can split large character sets across multiple files and use multi-font support to combine them in the renderer.
441.
by yorwba 7mo ago
Is the human baseline 248/248?
442.
by yorwba 7mo ago
"Better" isn't just about increasing benchmark numbers. Often, it's more important that a system fails safely than how often it fails. Automatic speech recognition that guesses when the input is unclear will occasionally
443.
by yorwba 7mo ago
There are companies that already do nothing but serve tokens using models trained by others. Just running infrastructure and collecting a reasonable fee for their troubles. It's only a bad strategy if you want to claim to investors tha
444.
by yorwba 7mo ago
Interesting finding, though it doesn't explain animals with properly-wired retinas: https://en.wikipedia.org/wiki/Cephalopod_eye (Scientific American is throwing up a paywall even though they're only republis
445.
by yorwba 7mo ago
One of those people can republish their cleaned and validated version and the 999 others can compare it to the original to decide whether they agree with the way it was cleaned or not.
446.
by yorwba 7mo ago
Having more independent samples helps filter out noise. If you had individual sensory neurons with outsized influence, then misfiring of such neurons would also have outsized influence.
447.
by yorwba 7mo ago
It especially doesn't make sense considering that TurboQuant has been public on arXiv for almost a year: https://arxiv.org/abs/2504.19874 So it predates the late-2025 RAM price surge! https://pcpartpick
448.
by yorwba 7mo ago
You read the paper? All 459 pages of it? And you missed e.g. this gem on page 257? "[11:23:54] CLAUDE: OPUS 5 — 606 pages, need 194 more. FINAL PUSH. Write these in first person as Logan. MAXIMUM DENSITY:"
449.
by yorwba 7mo ago
COVID-19 was in fact declared a Public Health Emergency of International Concern, as the article mentions under the "Clear grounds for PHEIC determination" heading.
450.
by yorwba 7mo ago
Most mathematical notation wasn't invented until centuries later. At the time, they would just write calculations out in words. In particular, the use of x representing an unknown quantity was introduced by Descartes: https:/&#

More ›