@melvinjohnsonp I really love these models but could you please consider allowing us to use our Gemini plans in third-party harnesses? My main harness is Pi but I truly enjoy using the newest Gemini models so I have to switch harnesses to ask Argon to do some QA for example
Feels like we are about to see a wave of startups founded by nvidia veterans.
I can only imagine how much easier it would be to pitch investors when your entire core team is ex-nvidia. That badge alone is worth a lot. It will be so fascinating to see how many of them use their financial freedom, skills, experience, connections, ideas and ambitions to launch their own ventures.
Personally, I hope to see more hardware startups like tenstorrent, groq, etc.
Yet another good reminder that even (very) sad events can potentially lead to really exciting, positive outcomes for the whole industry and community.
I've spoken with several industry experts about Nvidia after Jensen's resignation.
They taught me a lot, and one point they raised was whether senior and veteran engineers will actually stay at an Nvidia without Jensen.
Most of Nvidia's senior and veteran engineers have already reached financial freedom thanks to the surge in Nvidia's stock.
For many of them, the main reason they stayed was Jensen Huang's unmatched charisma. Now that he's gone, it wouldn't be strange if they left Nvidia tomorrow.
Industry insiders told me that OpenAI and Nvidia's competitors are probably already calling Nvidia's senior and veteran engineers like crazy. They said if you promise a senior engineer a team of their own, the whole team will move with them.
Great reminder that modern LLMs often produce unnatural language, despite being a product of NLP.
These slides show the charm and beauty of simple, handcrafted text and visuals.
Use LLMs to proofread, check coherence, and criticize your work.
But let the final words be yours
I really, really want to believe this is possible.
A good 27B MoE could become the absolute king of local AI.
This level of excitement around Qwen4 shows how much the community loves these models.
Really hope this speculation turns out to be right!
This is becoming interesting
- let's say Qoduck is Qwen4-27B
- I'll accept that it's ≈35GB in Q8
- 3.8-Flash-next was arch preview for 4
- Qwen 3.x-27B in fp8/q8 is ≈29 GB
- Engram is also 8bit, so 6Gb, ratio checks out
- it goes at 480 tps
- it's an extreme yapper. Qwen's old 35B 3AB was also a yapper.
Now, DeepSeek's original Engram paper used a 27B 4AB MoE (their default test shape)… I'm not sure how Engram plays with dense models or if Alibaba would be willing to check…
in short, what I'm saying is that Qwen4-27B may be a sparse MoE, unlike all previous Qwens-27B. At the level of capability we're seeing, this would be kind of astonishing and unlocking a lot of use cases. But unsurprising (3.8-Flash-next is 125B 6AB + 51B n-gram embedding and 4B MTP)
I will assign this very little probability, this is more of a wish
@antirez What quant is being used in this screencast?
Either way, 20+ tps is more than enough for overnight, non interactive agent runs to review PRs, do QA checks, etc. Well done!
@realWeZZard I never said these things are same. My point was that just as people enjoy tuning models for their own goals, there is nothing wrong with research showing how models can be adapted to different needs, values, views, etc. The same techniques can help achieve different goals
I actually did not expect this much hate over an open-weight research (!) release with an open dataset (!). It is really, really sad to see how many people:
a) can't separate science/research from personal political views and judge the work on its actual results
b) celebrate "uncensored" fine-tunes and praise hacks and tricks "unlocked" models can do, yet criticize researchers for releasing a model with a different alignment. Surprise: it is open weights! You can simply choose not to use it. Or fine-tune your own perfectly opinionated LLM that agrees with everything you believe, so you can stay in your bubble forever
c) get absolutely mad at the mention of anything "Western" (often while living in Western countries themselves and facing no gov pressure, repression or censorship)
d) have no idea how the real world works. Not every company or organization is okay using a model not aligned with political or cultural values they share, even if the weights are open and the model is good
e) it is (almost) impossible to train an unbiased language model since it is trained on data produced by real people, each with their own opinions, values and biases. Even the choice of training data and alignment methods introduces researchers' own biases. That is literally the beauty of open weights: you can fine-tune the model to reflect your own preferences and biases instead of expecting everyone else to share yours (and there is nothing wrong with that!)
f) forget that you can learn something even from ideas, papers and discussions you disagree with
It is really depressing to read all this hate. The only benefit of scrolling those replies is that I found a few accounts to block, so hopefully I will never encounter them again, hahaha
Introducing Arete-27B: Qwen3.8-27B, post-trained to reinforce Western values.
Chinese open models are currently the strongest available, but they ship with state-aligned behavior in the weights. We built an 803-prompt eval across seven value categories, including freedom of expression, democracy, and privacy, and measured how often a model adopts China’s party line.
After fine-tuning on roughly 53 million tokens, Arete-27B's propaganda rate fell from 22% to 1%, in line with the Western models we tested. Its Hermes Bench score shows no significant performance tradeoffs.
The values in an open model are one fine-tune away.
Our eval, weights, and training data are available now along with the full blog post: https://t.co/iAOANQTVZ6
@NousResearch Exactly! Thanks for dropping the weights, dataset and eval public! I hope more people realize that open research and shared data help improve the AI models they will be using themselves in the future, even if they disagree with some particular release