6 ms·
DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning
- victorbuilds 11mo agoNotable: they open-sourced the weights under Apache 2.0, unlike OpenAI and DeepMind whose IMO gold models are still proprietary.
- SilverElfin 11mo agoIf they open source just weights and not the training code and data, then it’s still proprietary.
- mips_avatar 11mo agoYeah but you can distill
- amelius 11mo agoIs that the equivalent of decompile?
- c0balt 11mo agoNo, that is the equivalent of lossy compression.
- littlestymaar 11mo agoYou can distill closed weights models as well. (Just not logit-distillation)
- mips_avatar 11mo agoThough it violates their terms of service
- falcor84 11mo agoIsn't that a bit like saying that if I open source a tool, but not a full compendium of all the code that I had read, which led me to develop it, then it's not really open source?
- fragmede 11mo agoNo. In that case, you're providing two things, a binary version of your tool, and the tool's source. That tool's source is available to inspect and build their own copy. However, given just the weights, we don't have the source, and can't inspect what alignment went into it. In the case of DeepSeek, we know they had to purposefully cause their model to consider Tiananmen Square something it shouldn't discuss. But without the source used to create the model, we don't know what else is lurking around inside the model.
- NitpickLawyer 11mo ago> However, given just the weights, we don't have the source This is incorrect, given the definitions in the license. > (Apache 2.0) "Source" form shall mean the preferred form for making modifications, including but not limited to software source code, documentation source, and configuration files. (emphasis mine) In LLMs, the weights are the preferred form of making modifications. Weights are not compiled from something else. You start with the weights (randomly initialised) and at every step of training you adjust the weights. That is not akin to compilation, for many reasons (both theoretical and practical). In general licenses do not give you rights over the "know-how" or "processes" in which the licensed parts were created. What you get is the ability to inspect, modify, redistribute the work as you see fit. And most importantly, you modify the work just like the creators modify the work (hence the preferred form). Just not with the same data (i.e. you can modify the source of chrome all you want, just not with the "know-how and knowledge" of a google engineer - the license can not offer that). This is also covered in the EU AI act btw. > General-purpose AI models released under free and open-source licences should be considered to ensure high levels of transparency and openness if their parameters, including the weights, the information on the model architecture, and the information on model usage are made publicly available. The licence should be considered to be free and open-source also when it allows users to run, copy, distribute, study, change and improve software and data, including models under the condition that the original provider of the model is credited, the identical or comparable terms of distribution are respected.
- amelius 11mo agoTrue. But the headline says open weights.
- very_illiterate 11mo ago[flagged]
- ekianjo 11mo agoIt's just open weights, the source has no place in this expression
- jimmydoe 11mo agoyou are absolutely right. I'd rather use true closed models, not fake open source ones from China.
- PunchyHamster 11mo agoI think we should treat copyright for the weights the same way the AI companies treat source material ;)
- littlestymaar 11mo agoWe don't even have to do that: weights being entirely machine generated without human intervention, they are likely not copyrightable in the first place. In fact, we should collectively refuse to abide to these fantasy license before weight copyrightability gets created out of thin air because it's been commonplace for long enough.
- mitthrowaway2 11mo agoThere's an argument by which machine-learned neural network weights are a lossy compression of (as well as a smooth interpolator over) the training set. An mp3 file is also a machine-generated lossy compression of a cd-quality .wav file, but it's clearly copyrightable. To that extent, the main difference between a neural network and an .mp3 is that the mp3 compression cannot be used to interpolate between two copyrighted works to output something in the middle. This is, on the other hand, perhaps the most common use case for genAI, and it's actually tricky to get it to not output something "in the middle" (but also not impossible). I think the copyright argument could really go either way here.
- littlestymaar 11mo ago> An mp3 file is also a machine-generated lossy compression of a cd-quality .wav file, but it's clearly copyrightable. Not the .mp3 itself, the creative piece of art that it encode. You can't record Taylor Swift at a concert and claim copyright on that. Nor can you claim copyright on mp3 re-encoded old audio footage that belong to the public domain. Whether LLMs are in the first category (copyright infringement of copyright holders of the training data) or in the second (public domain or fair use) is an open question that jurisprudence is slowly resolving depending on the jurisdiction, but that doesn't address the question of the weight themselves.
- ilmj8426 11mo agoIt's impressive to see how fast open-weights models are catching up in specialized domains like math and reasoning. I'm curious if anyone has tested this model for complex logic tasks in coding? Sometimes strong math performance correlates well with debugging or algorithm generation.
- stingraycharles 11mo agokimi-k2 is pretty decent at coding but it’s nowhere near the SOTA models of Anthropic/OpenAI/Google.
- tripplyons 11mo agoAre you referring to the new reasoning version of Kimi K2?
- stingraycharles 11mo agoThis one => https://openrouter.ai/moonshotai/kimi-k2-thinking https://openrouter.ai/moonshotai/kimi-k2-thinking
- alansaber 11mo agoIt makes complete sense to me: highly-specific models don't have much commercial value, and at-scale llm training favours generalism.
- yorwba 11mo agoPrevious discussion: https://news.ycombinator.com/item?id=46072786 https://news.ycombinator.com/item?id=46072786 218 points 3 days ago, 48 comments
- victorbuilds 11mo agoAh, missed that one. Thanks for the link.
- terespuwash 11mo agoWhy isn’t OpenAI’s gold medal-winning model available to the public yet?
- esafak 11mo ago'coz it was for advertisement. They'll roll their lessons into the next general purpose model.
- H8crilA 11mo agoHow do you run this kind of a model at home? On a CPU on a machine that has about 1TB of RAM?
- pixelpoet 11mo agoWow, it's 690GB of downloaded data, so yeah, 1TB sounds about right. Not even my two Strix Halo machines paired can do this, damn.
- Gracana 11mo agoYou can do it slowly with ik_llama.cpp, lots of RAM, and one good GPU. Also regular llama.cpp, but the ik fork has some enhancements that make this sort of thing more tolerable.
- bertili 11mo agoTwo 512GB Mac Studios connected with thunderbolt 5.
- sschueller 11mo agoHow is OpenAI going to be able to serve ads in chatgpt without everyone immediately jumping ship to another model?
- miroljub 11mo agoI don't care about OpenAI even if they don't serve ads. I can't trust any of their output until they become honest enough to change their name to CloseAI.
- Coffeewine 11mo agoI suppose the hope is that they don’t, and we wind up with commodity frontier models from multiple providers at market rates.
- KeplerBoy 11mo agoGoogle served ads for decades and no one ever jumped ship to another search engine.
- sschueller 11mo agoBecause Google gave the best results for a long time.
- PunchyHamster 11mo agoand now, when they are not, everyone else's results are also pretty terrible...
- bootsmann 11mo agoThey pay $30bn (more than OpenAIs lifetime revenue) each year to make sure noone does.
- KeplerBoy 11mo agoWhat are you referring to?
- simianwords 11mo agoA bit important that this model is not general purpose whereas the ones Google and OpenAI used were general purpose.
- mangolie 11mo agohttps://x.com/deepseek_ai/status/1995452646459858977 https://x.com/deepseek_ai/status/1995452646459858977 Boom
- simianwords 11mo agoOh you may be correct. Are these models general purpose or fine tuned for mathematics?
- yorwba 11mo agoThat's a different model: https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Speciale https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Speciale
- andy12_ 11mo agoDo note that that is a different model. The one we are talking about here, DeepSeekMath-V2, is indeed overcooked with math RL. It's so eager to solve math problems, that it even comes up with random ones if you prompt it with "Hello". https://x.com/AlpinDale/status/1994324943559852326?s=20 https://x.com/AlpinDale/status/1994324943559852326?s=20
- yorwba 11mo agoBoth OpenAI and Google used models made specifically for the task, not their general-purpose products. OpenAI: https://xcancel.com/alexwei_/status/1946477756738629827#m https://xcancel.com/alexwei_/status/1946477756738629827#m "we are releasing GPT-5 soon, and we’re excited for you to try it. But just to be clear: the IMO gold LLM is an experimental research model. We don’t plan to release anything with this level of math capability for several months." DeepMind: https://deepmind.google/blog/advanced-version-of-gemini-with-deep-think-officially-achieves-gold-medal-standard-at-the-international-mathematical-olympiad/ https://deepmind.google/blog/advanced-version-of-gemini-with... "we additionally trained this version of Gemini on novel reinforcement learning techniques that can leverage more multi-step reasoning, problem-solving and theorem-proving data. We also provided Gemini with access to a curated corpus of high-quality solutions to mathematics problems, and added some general hints and tips on how to approach IMO problems to its instructions."
- letmetweakit 11mo agoDoes anyone know if this will become available on OpenRouter?
- WhitneyLand 11mo agoShouldn’t there be a lot of skepticism here? All the problems they claim to have solved are on are the Internet and they explicitly say they crawled them. They do not mention doing any benchmark decontamination or excluding 2024/2025 competition problems from training. IIRC correctly OpenAI/Google did not have access to the 2025 problems before testing their experimental math models.
- YouAreWRONGtoo 11mo ago[dead]
- OBELISK_ASI 11mo ago[dead]
- LZ_Khan 11mo agoDon't they distill directly off OpenAI/Google outputs?
- Jeff-Collins 11mo ago[dead]
- Scott-David 11mo ago[dead]