News for hackers. Stuff that matters.
Re:GPT-6 Astra, Looped Transformers, and Hidden Reasoning by boredatoms on Wednesday September 09, 2026 @01:50PM (#49630447)
For sure they quickly move to q8, the output quality difference to bf16 is small compared to the speed/capacity gain
Re:GPT-6 Astra, Looped Transformers, and Hidden Reasoning by torginus on Wednesday September 09, 2026 @04:02PM (#49633351)
Some people here have remarked previously that while reduced precision doesn't show up in quick prompts, it does severely impact these models' ability to perform long running tasks - to the point that running these big models with severe quantization might be counterproductive as smaller but less quantized ones perform better.
Re:GPT-6 Astra, Looped Transformers, and Hidden Reasoning by NineStarPoint on Wednesday September 09, 2026 @02:23PM (#49631165)
Yeah q8 made so littler difference back when I was testing such things I'd be surprised if people could quickly notice that as a change. It's got to be either further quantized or some other type of optimization that kicks in when people notice the drop.
Re:GPT-6 Astra, Looped Transformers, and Hidden Reasoning by selectodude on Wednesday September 09, 2026 @03:11PM (#49632434)
NVFP4 would buy them a huge increase in capacity but I think it would be noticeable.
Re:GPT-6 Astra, Looped Transformers, and Hidden Reasoning by Caracas288 on Wednesday September 09, 2026 @03:55PM (#49633225)
Why doesn't someone just try to measure this next time!?
Re:GPT-6 Astra, Looped Transformers, and Hidden Reasoning by embedding-shape on Wednesday September 09, 2026 @06:33PM (#49635527)
Can't really measure without being sure you aren't being messed around with, when it's a remote platform. Stupidly easy to detect when people run such benchmarks/tests against you as well.
Re:GPT-6 Astra, Looped Transformers, and Hidden Reasoning by nonethewiser on Wednesday September 09, 2026 @03:33PM (#49632850)
Could this explain Opus?