HN rank: 1

HN score: 558

comments: 492

original url: https://openai.com/index/sharing-ai-progress-in-mathematics/

评论信号：I was a graph theory junkie long ago and even moved to Budapest for awhile to study among the greats. While I was there I started working on Barnette's Conjecture which came to occupy my thoughts over the next 24 years of my life, on and off as I worked in many different fields. Last summer I even thought for a few days that I had actually solved it.But it's supposedly proven here - problem 180. I don't know what to think exactly. I spent thousands of hours on that problem. I really enjoyed it. Hearing that it is solved somehow makes me sad in a far-off way, like hearing an ex-girlfriend died suddenly in a car crash. I don't know, there's probably a lot of people feeling odd emotions tonight.There's no Lean proof for this one so I'm digesting the paper. On the surface it looks like an approach I considered 24 years ago and abandoned.I revisited the problem this summer, along with my partial solutions, when the previous round of stunning proofs came out. Several hours of work with Fable simply convinced me it wasn't yet solvable and reinforced how hard of a problem it was. reply: I have very little understanding of higher math, so I ask you:...

评论信号：As Kevin Buzzard recently said:> In a 2020 piece in the Notices of the AMS, I asked the following question: “If one human had an understanding of all of modern pure mathematics simultaneously, how much further would they immediately be able to see?” Six years later we are beginning to understand the answer to this question. reply: The other crucial part to this is the ability to actually encode and test the theorem (via Lean). Otherwise, we would be swarmed with a billion lines of theorems that no one will be able to ever understand and verify anyway.

评论信号：As a TCS/scheduling person, this one is definitely of lesser importance than UGC, but it has been an open problem since the book of Garey and Johnson in 1979:A Polynomial-Time Algorithm for Three-Machine Unit-Job Scheduling [1]Since some people talk about small numbers that pop up in integer multiplication results, here a completely different number appears:Theorem 1.1. Let an explicitly listed finite directed acyclic graph specify the precedence constraints on n >= 1 nonpreemptive unit-length jobs on three identical machines. There is a uniform deterministic algorithm that constructs a feasible schedule of minimum makespan. Given also an integer deadline 1 https://github.com/openai/math/blob/main/preprints/A-polynom... reply: The largest I've seen [1] is an exponent of 10^12, which I suppose still counts as polynomial time.I'm sure all of these super small or large constants will improve over time, but it's still amusing. It is entertaining to see the exponents directly rather than have them hidden as n^c or epsilon or O(1).[1]: https://github.com/openai/math/blob/main/preprints/Determini...

HN rank: 2

HN score: 1630

comments: 978

original url: https://mistral.ai/news/mistral-large-4/\

评论信号：Surprisingly it only supports reasoning "none" or reasoning "high".That setting didn't seem to make any real difference - it added a tiny bit of thinking trace and high actually produced less output tokens than none.The high bicycle frame is better then the none one though.Pelicans: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...(Definitely the best I've seen from any Mistral model: https://simonwillison.net/tags/pelican-riding-a-bicycle+mist... ) reply: This is such a pristine pelican. Let me say it here first folks, AGI is here.

评论信号：I have a question, and perhaps some of the AI/ML infrastructure experts here could answer: Mistral says, "ML4 was trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral’s own datacenters in Europe."If a 1T params model trained on ~4k NVIDIA GB GPUs could almost match the performance of Kimi's K3 (which is on par with top closed source models of OpenAI/Anthropic) while beating/exceeding other leading SOTA models from top Chinese labs, what are we (in the US) even building these super massive data centers for? Just to churn through more backpropagation reps more quickly?SpaceXAI's Colossus supercluster in Memphis and Colossus 2 in Memphis/Mississippi (Southaven) are supposed to run into hundreds of thousands to a million GPUs. MSFT's Fairwater GPUs are supposed to have hundreds of thousands as well. So, 3800 GB GPUs are an absolute drop in the bucket. I don't understand the strategy of hyperscalers here, especially with edge inference hardware only getting better from here on (Apple, and all).Distillation explains some of the advances, but doesn't that mean hyperscalers have a ton of deadweight wrt GPUs sitting on their balance sheets?... reply:...

评论信号：Impressive vision benchmarking. If the vision model is truly as good as astra, that would make it best in the world.Also strong on cyber benchmarks (better than all chinese models), so this is a good defender model.Lots of people shitting of Mistral for no reason imo. These are pretty good numbers across the board. Definitely good enough to use as a daily driver over other llms, if you have moral qualms with the others. For certain use cases, like cyber security, this may be the go to model.I like to make fun of europe, but there's lots for mistral to be proud about in this release imo. reply: I'm still embedded in the OpenAI ecosystem, but man, when I try out Mistral, it is snappy! Like basically instant responses that seem faster and better in quality than ChatGPT on instant mode. I'm impressed.

HN rank: 3

HN score: 55

comments: 5

original url: https://strandsagents.com/blog/introducing-strands-decider/

评论信号：At this point I can't wait for a comedian to release a decision model backed by humans.Meet Jerry- it's literally a guy named Jerry answering your questions.

评论信号：Fantastically well written. It's rare for me to be able to understand what the AI gurus are talking about, and this was written by humans for humans.It can technically be used for a lot of use cases, I'd like people to chime in on ideas on this?

评论信号：This is a great model. I've been running it on device in chrome extension to filter things like email.It is just the right mix of size, capability and speed to make it generally useful for adhoc bulk classification tasks.For those wanting to run it in browser: https://huggingface.co/alxnahas/strands-decider-2B-webgpu reply: Live demo: https://alxnahas.github.io/strands-decider-web/?backend=engi...

HN rank: 4

HN score: 186

comments: 78

original url: https://developers.openai.com/api/docs/guides/decisions

评论信号：curl https://api.openai.com/v1/decisions
-H "Authorization: Bearer $(llm keys get openai)"
-H "Content-Type: application/json"
--data ' { "model": "gpt-6-luna", "input": [{ "role": "user", "content": [ {"type": "input_text", "text": "I am angry about the new product feature"} ] }], "questions": [{ "type": "predicate", "name": "complaint", "instructions": "Is this a complaint?" }, { "type": "predicate", "name": "compliment", "instructions": "Is this a compliment?" }] }'

Returned: { "model": "gpt-6-luna", "answers": [ { "type": "predicate", "name": "complaint", "probability": 0.91 }, { "type": "predicate", "name": "compliment", "probability": 0.06 } ], "usage": { "input_tokens": 310, "input_tokens_details": { "cached_tokens": 0, "cache_write_tokens": 0 }, "output_tokens": 0, "output_tokens_details": { "reasoning_tokens": 0 }, "total_tokens": 310 } }

That https://api.openai.com/v1/decisions endpoint is notable because usually when OpenAI define an endpoint like that it ends up as a defecto standard for other providers.(I turned this all into a new llm plugin: https://github.com/simonw/llm-openai-decisions) reply:...

评论信号：The response to Jev should be the nail in the coffin over whether or not the AI business is a commodity market.Out of no where Jev appeared as the next round of the price wars. Jev showed the value of System One models. A fast yes/no/confidence score not only is cheaper but also often all people want. Open source versions flood hugging face and now the big players are giving up a potentially big driver of output tokens to keep customers and race to the bottom price wise.If I were OpenAI or Anthropic I’d be racing to make their products as sticky as possible bc ppl will flock to what’s cheapest otherwise. reply: But OpenAI and Ant can be in more than just the commodity part of the business. They can both make frontier models and sticky products on top of those that have network or other effects that make competition tough.

评论信号：Ran my decisions evals (still rudimentary, less than 600 calls (UI component selection, chat charting, tag selection, PKM stuff)) on this via OpenRouter against Jev and Mercury Decide. Jev because it has replaced my mt0 efforts by sheer force of affordability (more importantly, the limits running on a MacBook Neo bring even after vocab pruning and quant insanity) and Mercury Decide because I do like dLLM efforts (and I'd like to use fewer model providers if possible).Preliminary of course, but seems to be slower than Jev and similar to Mercury Decides latency, though not in growing linearly with the amount of input (346ms p50 and 860ms p95, (Mercury Decide also had some extremes up to 1,3s that were around 800ms today, likely preview related, it scaled far more consistently with size)), less "confidence" concerning my ambiguous UI component and response shape specific tasks (have very specific use cases for these models which Luna often fails to meet at 0.6 and lower), lead to a few failed calls which neither competitor had (4 vs 0 for both) and measured more expensive than Jev to boot by a factor of 3,1 times on average (Mercury Decide pricing I think is still unknown so no number...

HN rank: 5

HN score: 31

comments: 8

original url: https://www.jeremyreimer.com/rockets-item.lsp?f=true&p=272

评论信号：(2020)I'd hope anyone who saw "Plane Crazy" would know better than to back a Robert X. Cringely Kickstarter, and I'd expect anyone who followed his tech takes for a bit would have acquired some salt to take with any claims he made.There seem(ed) to be a number of folks absolutely obsessed with hounding Cringely (apparently dead, I know) over the Kickstarter in his irregular posts. Given the highest tier was $200 and the whole of the Kickstarter was $35.5k USD, seems like a lot of throwing good time/energy after a bad Kickstarter call. I'd be overjoyed if that was the worst case of someone walking off with my money I'd ever faced.They are, or were I guess, free to do so, but it seems wasteful to spend the amount of time in his comments they did.

评论信号：This has bothered me from the start. I was not one of the buyers, so I have a different take on the issue. From all his writings, you get a sense that Cringely was a big-idea man, but he had very little ability to follow through on his next great idea or venture. It seems like he would lose interest once the hard part of building a business started to appear. In this case, it looks like someone revved him up on the idea of selling a low-cost Mineserver, I think it was his son, but he probably mispriced the true cost and did not calculate the amount of work it would take to get the project finished. I don't think he started out with the idea of not delivering the product. He was just not capable of following through on his grand ideas. Some people are dreamers, and others are doers. He was a dreamer, and you could sense it in all of his writings.That's the thing with Kickstarter you never really know if you will ever get the product. This project just became one of the many Kickstarter projects that failed.

评论信号：>Cringely posted pictures of the boards, the SD cards that would hold the open-source Minecraft server software, and the plastic shells etched by a laser cutter that would serve as the cases.For others who have played plenty of Minecraft but never heard the term "mineserver" used, sounds like he was trying to rewrite the Minecraft game server software.