RT shadcnfew weeks ago, Fable 5 was so advanced it needed the gov and had to be taken offline.today, we have access to arguably better models…
RT 熊师傅 weight decay 了吗Fun fact: I'm responsible of editing the limitations part, and this final line was written by Zhilin himself.Lisan al Gaib: from Moonshot's…
Did you think I'm just shitposting, anon?Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞): sorry rationalists, but the People's AGI God will be opensourced
Xi and Makima make a good pairTeortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞): Against UtopiasOne of my pet peeves is when people equate optimism about the future…
How about understanding this bro: 全人类共同的智慧结晶和宝贵财富Tenobrus: registering confusion: i don't really understand why Xi is still allowing Kimi to release such powerful open models. this is…
RT Republicans against TrumpDocument Declassified by Trump Reveals Russia Tried To Help Him Win the 2020 Electionhttps://www.mediaite.com/politics/trump/document-declassified-by-trump-reveals-russia-tried-to-help-him-win-the-2020-election/
The sad thing about GDM is that they aren't really aware of how far behind they are. Internal reports go like "on our internal highly rigorous…
Important to read. China is committed to continuing its open, global ai approach.Vincent Chow: Key takeaways from President Xi's speech in his first ever appearance at…
Zhipu grew their ARR by 4x in ≈3-4 months$16B by Q1 2027 could be doable at this rate… but for the inability to provide so much…
really interesting experimentnewer LLMs are growing Transmetropolitan-pilled"eh. stuff happens. who cares about another cyberattack"Ziqian Zhong: @a_karvonen I feel the timeline we're living in is just weird…
one thing DeepSeek should lean into is multi-agentthey are uniquely unconstrained on inference nowtheir models are *cheap* and *fast* compared to all Chinese competitionV4-Pro@16 costs ≈the…
Google should take 3.5 Pro (as well as all 3rd gen models) out back and deal with them the way Meta did away with Behemoth. That…
This is true. They were talking of stealth research at the time. They chose to build a "fat V3" instead, to quickly catch up to the…
Kimi K3 is a very good model, but people are overindexing on an Arena score again (remember Llama 4?)ELO scores as judged by Arena users are…
Really different calibers of leaders, I'm sorry brosJMB 🧙♂️: the "urgent" trump speech tonight is wild.he's trying to sell us on the idea that "china rigged…
really coolFatescript: I’m a huge Hollow Knight fan working at Kimi. Before K3 was released, I used Hollow Knight assets to build a game where the…
Internalize one thingthe defining characteristic of Xi Jinping is that he's very persistenthe's not like Trump. if he's doing something, almost certainly he knows what he's…
American century of humiliation will continue until revenge fantasies improveI want to see the version where Neuralink-upgraded burgers crush Asians at mafsuntil then, just more sad…
(it got outclassed in a few hours)Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞): Western open source frontier is back.This is on par with the largest of known…
well, with all respect to Thinky, there's a damn good reasonTrinky getting up to speed on training a V3+ tier model is a great win for…
beyond mHC: scaling the residual stream by effectively 16xXiangdong Zhang: Hyper-Connections never fade. They evolve.To push model intelligence further, we need to scale residual streams wider…
RT noahRe @saneord also 29 days ago.
A GPT-4 powered assistant for Pakistani judges increased the amount of cases they saw by 6% with no impact on quality.Elliott Ash: What happens when you…
RT Sasha RushI circled all the Transformer parts of k3. (I'm a sore loser).Sasha Rush: Time for everyone’s favorite game! Is it a Transformer?