DaisyRose-27B (radixark.com)
136 points by de_sty 6 hours ago
woah 5 hours ago
The heavy reliance on eval datasets showcased here and also evidenced by the frontier lab's high payments for such data cast doubt on the claims that AI will become self-improving soon
anon373839 2 hours ago
The frontier labs want you to believe that their models are magically gaining intelligence that generalizes to new areas with each new release. But it’s an illusion: the models are post-trained on a HUGE array of hand-curated datasets for specific tasks, with good enough coverage that it feels general.
harry8 2 hours ago
The frontier labs are using all their customers' logs to train. It's the best training set there is.
boardendgrain an hour ago
0xWTF 2 hours ago
How is that different from you?
StilesCrisis an hour ago
its_down_again 3 hours ago
Curious, where did you see the numbers on how much frontier labs are paying for eval data?
trescenzi 30 minutes ago
I’ve gotten recruiting emails to be a human grader of Claude coded projects. 6 month contract at $150/hour 40hrs/week. Not insanely expansive but also not cheap.
dimes 3 hours ago
charcircuit 3 hours ago
>doubt on the claims that AI will become self-improving soon
It already happened. https://github.com/karpathy/autoresearch
jacquesm 4 hours ago
I always wonder what you lose when you start maxing out some specific benchmark. It is really hard to do this in a balanced way and I would love if everybody that did these kind of efforts would take the time to also list the before/afters of a number of other benchmarks that they did not use in their training.
nblintao 4 hours ago
Good point! First of all, none of the benchmark tasks were used in training, though the training data was designed to improve capabilities on tasks like TB4.1. If you consider TB4.1 the main target, the blog actually shows before/after results on several other benchmarks, with no regressions and even some improvements.
jacquesm 4 hours ago
Yes, I've seen that, but I wonder what is missing because I find it hard to believe that you can just do some RL and then end up with a model that is better across the board. I would find it far more believable if you outperformed on the benchmark that was your goal (assuming you selected training data to give you an edge there) and underperformed on things that you were not optimizing for. But since it all moves in one direction (better) I feel there must be some kind of unknown at play here. Even so, task specific AI is definitely the future, you don't hire a plumber to do your bookkeeping either...
mik1998 2 hours ago
reissbaker 4 hours ago
gardnr 4 hours ago
Check the article. It’s the first graphic that has other benchmarks.
jacquesm 4 hours ago
Yes, they all go up. I am wondering what was lost.
pkoiralap 3 hours ago
Sabi-noema also looks like a close competitor in the same space. Built from qwen3.8 and made better.
https://huggingface.co/sabicap/sabi-noema
Also HF model here: https://huggingface.co/RadixArk/DaisyRose-27B
xlayn 5 hours ago
Snap, no gguf to download, will wait for one to appear to test, any improvement over an already good model is good news.
"18 days on 144 H200 GPUs" insane.. didn't thought so much hardware would be needed to train Qwen.
drdaeman 4 hours ago
Rather than waiting, ask your assistant to quantize for you. It could be not the best quantization possible but it’s relatively quick, and I had an overall positive experience.
smallerize 4 hours ago
It is already quantized. You need to convert to GGUF format though. Did your assistant explain this wrong?
drdaeman 4 hours ago
cookiengineer 36 minutes ago
It isn't if you train the quantized model with a quantized K/V and cache. That's my approach and usually takes around a day for 100k reasoning traces (including abliteration).
Also this is on consumer hardware, meaning 96GB CPU RAM and 48GB GPU VRAM, using an optimized Vulkan concurrent shader pipeline. So costs for the machine are around 3-5k depending on RAM crisis.
mips_avatar 3 hours ago
How did you guys set up the RL topology? Were you guys doing individual RL nodes (4xh200 rollout/4xh200 train) or did you have dedicated trainer and rollout nodes?
nblintao 17 minutes ago
Dedicated rollout and trainer nodes, with RDMA transfers between them.
0x69420 26 minutes ago
3.8 27B is an interesting choice. on the face of it, doing RL on it feels like refrying twice-cooked pork then stuffing it in the microwave for good measure. the fact that even one benchmark (tb4.1) increased more than marginally is kind of miraculous. would be interesting to see where "miles" would take eg. 3.8 flash next, which is already impressive despite being an under-trained architectural preview.
robertkarl 5 hours ago
I look forward to testing this against my private eval set to determine if it's benchmaxxing or actual improvement based on just 1200 tasks. downloading it now
XCSme 5 hours ago
Is it post-training or bench-maxxing?
octoberfranklin 5 hours ago
FWIW that is about $331,000 on low-onboarding-grief clouds (e.g. vast.ai) right now using the SXM variants they used.
You might be able to shave that down using a different flavor of H200 (like PCIe) but it would take more hours, so benefits are unclear.
mips_avatar 3 hours ago
You'd probably end up spending more on pcie h200s. They're not that much cheaper than 8x sxm nodes and they only have 4 way nvlink. At the scale these guys were training at you'd probably want at a minimum an 8x nvlinked node for the trainer, even though you could technically fit the full training and optimizer states on a 4x h200 node.
lumost 4 hours ago
Savings would likely be larger with alternative accelerators such as trainium, no?
octoberfranklin 4 hours ago
Savings would likely be larger with magic beans such as floobarbaz, no?
But seriously. If there were a cheaper and reliable way, I'm sure the authors would have mentioned it instead of H200 SXM modules. They're trying to brag about cheap training.
Maybe you wanted to use an opportunity to shill tranium.