与「Other_way」相关的搜索结果

Okay, the @VulcanBench results for Qwen3.8-Max are in, and it is not what I expected. First, for anyone new to VulcanBench, here's a quick TL;DR on the eval suite: 23 frontier-hard software engineering tasks taken from real merged OSS PRs, run in a Docker sandbox, 3 runs per task across all three of its effort levels. No puzzles, no random abstract stuff, all real things engineering teams would do with these models. It looks like Qwen3.8-Max has a major overthinking problem, it uses a LOT of tokens and is very slow, period, no other way to see it. My cost to run this benchmark was $126.25, to run the exact same eval suite with DeepSeek V4-Flash was only $13.60. This makes Qwen3.8-Max an insanely expensive model. The tasks Qwen genuinely can't solve fail at every effort level, extra reasoning didn't help. The regression is almost all in work it already handles: six tasks that low solves every single time account for 83% of the 26-point drop, three of them collapsing to zero. It's not losing the hard problems. It's losing the ones it already knows how to do. Since Qwen3.8-Max hit a lot of wall clock budget caps, I thought I'd share more about this. - VulcanBench caps both steps (50–200) and wall clock (5–60 min), each scaled by repo size. - This is aligned with how comparable harnesses bound agents, DeepSWE caps rollouts at 100 environment steps, sitting right inside my step range; Terminal-Bench enforces a per-task wall clock; SWE-bench Verified scaffolds typically allow 20–60 min per instance with 250–350 step limits. - Every model on my chart gets the identical budget, and Qwen is the slowest model I've tested at 20–25 min/task. Soooo... Alibaba positions Qwen3.8-Max as trailing only Claude Fable 5. But on the kind of real coding work engineering teams would actually throw at it, under a fixed budget, its best setting lands mid-pack and its default lands last, so common. If you want to optimize for accuracy, Grok 4.5 is the move. If you want accuracy per dollar, DeepSeek V4-Flash is hard to beat, heck it's 10× cheaper than Qwen and you get higher accuracy. Qwen just isn't in the game at this point, this is not a model I could see engineering teams using for daily coding work.

Geoffrey Hinton says a big language model runs on about 1% of your brain's connections and still ends up knowing more than you: "So in your brain, you have a hundred trillion connections, roughly speaking. Okay. That's a lot. And you only live for about two billion seconds. That's not much." "If you compare how many seconds you live for, with how many connections you've got, you have a whole lot more connections than experiences." "Now with these neural nets, it's sort of the other way round. They only have of the order of a trillion connections. So like 1% of your connections, even in a big language model, many of them fewer, but they get thousands of times more experience than you." "So the big language models are solving the problem with not many connections, only a trillion. How do I make use of a huge amount of experience?" "And back propagation is really, really good at packing huge amounts of knowledge into not many connections." "But that's not the problem we're solving. We've got huge numbers of connections, not much experience. We need to sort of extract the most we can from each experience." Two to three billion seconds is the whole budget. Everything you know, you learned inside it. So evolution built you to squeeze a lot out of very little. Hinton's point is that a language model has the opposite problem and the opposite fix, and backprop turned out to be extremely good at that fix. Worth noticing what this predicts about failure. A system running on 1% of your wiring and thousands of times your experience is not going to fail the way you do. You fail from having seen too few examples. It fails from compressing too many into too little, and the compression is where the errors get made. That is a strange thing to be deploying into hospitals and courts with no way to inspect it. We test these systems by asking them questions, which tells you what came out. Nobody can yet look at a trillion connections and say what got packed in. - Geoffrey Hinton, Nobel laureate and Turing Award winner, on StarTalk (@StarTalkRadio) with Neil deGrasse Tyson.

RESEARCH: The June-July window repriced across crypto as stablecoin supply, spot BTC ETF flows and aggregate exchange balances all contracted since 1 June. Binance moved the other way, posting positive flows in early July while the market saw net outflows

Time to be straight with you about hoodl. The idea was good. The implementation wasn't 100%. Two main things I got wrong: 1. The game punished trading. Sell taxes, clocks resetting, every mechanic told you to never touch the token. But trading is what fills the pot. I built a prize pool that needed volume inside a design that suppressed it. 2. The custom hook. It made $hoodl unswappable on normal Uniswap and aggregators, which pushed people into scam pools. Neither is fixable in place. The hook is immutable. I can't patch v1 into what it should have been. So here's what I learned, partly from watching what actually works on this chain (credit where due, stonkbrokers showed what good plumbing looks like): Value has to flow into the liquid token, and the NFT gets priced in the token, never the other way around. Fees have to come from activity the game encourages, not activity it punishes. And the token has to trade clean, anywhere, no tricks in the swap path. And here's where I've landed: I don't want to leave it how it is. The idea deserves better than the version I shipped, and this community deserves the best version possible. So that's what I'm building. hoodl v2 is coming. v1 holders, you won't be left behind. There will be a path in. And v1 isn’t shutting down: it keeps running exactly as it does today, the game, the pot, the fees, all of it, for anyone who decides to stay. The Sheriff got away with one. He won't get away with two. Details soon.

A guy pulled $62,000 on YouTube last month without opening a browser. Claude Code opens it for him. He points Claude Code at his script and lets it walk through the workflow. Open Higgsfield. Generate 38 images matched to the timestamps. Save each one under the exact second it belongs in the video. Grab the transcript from TurboScribe. Drop everything in a folder. Done before he refills his coffee. Most people do YouTube automation backwards. Sign up for a tool. Copy prompts from a template. Wonder why it flopped. He does it the other way around. Write the script first. Let Claude Code drive whichever tool the script needs. Higgsfield for stills. TurboScribe for the transcript. Nano Banana for the covers. Every asset lands pre-labeled and ready to cut. The tools are not the moat. The orchestration is. AI made every content tool cheap. It did not make one workflow smart. That still comes down to knowing which tools stack and in what order. The next wave of YouTubers will not win by trying every new AI toy. They will win by writing the pipeline once and letting it print for a year.

Some observations on Kimi: 1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also seemed very token hungry. It's not obvious to me that this model is actually that cheap to run. 2. I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks. To be clear, I *myself* might be fine with models presenting this level of marginal risk being open weight, but I am surprised that China is fine with it. I suspect the reason they are is 75% explained by strategic blindness/lack of AGI-pilledness (the CCP is very Yann Lecun-y in its views of AI). The other 25% or so is their lack of compute for customer inference (making China's open-weight strategy an unintended byproduct of US export controls) and the normal Chinese strategy of aggressive exports. For the companies, as opposed to the government, the decision to open source is partially ideological and partially because they are behind, and they know that very few people would pay for sub-frontier models from China. 3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex. 4. One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape, but I've never met an open-weight models advocate who doesn't ultimately concede this is where things end. You'd be surprised how many 'accelerationists' lobbied me, while I was in government, to support an eleven or twelve-figure federally funded data center so that startups could train models at a subsidy and then give them away for free. There was no other way for AI to progress, they said. Perhaps this is the logical end state of things. Nonetheless, I find myself surprised to see supposed accelerationists excited about such an outcome. I think many of them just don't know what they're doing. Many accelerationists do not view the creation and serving of frontier models as a legitimate business. 5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this. 6. It's probably true that open-weight models of this capability make the world a bit more dangerous, but not so much more that you'll really notice. At some point the models will be capable enough that you will notice. "A nonliving, invisible, dangerous, and infinitely self-replicating agent escaped from a Chinese lab," you say? Color me shocked.

No words are big enough right now to overcome this empty feeling in the stomach. We were close, really close to another final but it wasn’t enough. We’ve given everything over these last 7 weeks and to fall short is hard to take! I know the expectations are high and rightly so, we’ve been knocking on the door for 8 years now but again are missing that final piece of the jigsaw! That’s where we have to go away, process it and find a way to get better. I’m so proud of the boys and what we have shown throughout this tournament - some tough games and tough environments that we have overcome. Some memories that will stay with us players and I’m sure you fans for a long long time! Going for glory doesn’t always mean you will get it. You have to fight for it, get knocked down, pick yourself up and go again and that’s what we will do, there’s no other way but to keep believing and keeping pushing. Thank you to every single fan that travelled and showed their support in the stadiums. Thank you to every fan back home for believing in us. Thank you to the boys and staff for everything you have given. As always Win or lose, we learn and go again! 🏴󠁧󠁢󠁥󠁮󠁧󠁿🦁

it’s remarkable seeing online dating as a product complete its lifecycle. i have rarely ever witnessed products that just mechanically stop working for most of the customers ever before like this, does anyone have a good comparison? these products worked when the userbase was positively selected (early adopters, normal-ish people, 2012–2018 tinder boom) & the algo was quite dumb & they were not monetized to hell. but now selection runs the other way where staying on the apps is itself a negative signal, which is a death spiral no product fix can ever escape. this is pure adverse selection eats the market, aka lemons but for romance. pretty fun to see.