aria

the overloading of "diffusion" in an LLM context comes back to bite me again this is an actual text diffusion model, doing multi-pass denoising of tokens. the diffusion in DFlash is different, just a single-pass transformer predicting multiple tokens at once. i imagine both are block-wise though

prefer analogue CDs, the sound is much warmer

just installed my new NVL4608 "Feynman Ultra" RF300 rack. anyone know a good model to run on it

an intuition that has never failed me is to mentally respond to every paper claiming "using this trick, you can skip/half-ass RL and still get the same results" with "no you can't" it's a good mental exercise to play "spot the methodological error" on these

open.substack.com/pub/halimede... woke up this morning to a substack post perfectly calibrated to cause me maximum pain

keep up the good work, dot

{model_name} is sooooo good, I've built a {3d_game_demo} with it and it one shotted it. {3d_game_demo_video}

arxiv.org/pdf/2609.22978 how could I miss the most important ML event of the week... the DeepSeek sandboxing tech report!

Huawei Ascend kernels are the most insidious avenue for xrisk... thank you for saving us, dario.

huggingface.co/XiaomiMiMo/M... MiMo-V2.6 is out! and, more importantly to me, the tech report. this is the most batteries-included tech report for a modern frontier model ever. really cool stuff.

signal seems to only give me notifications on my phone when it is asleep. this leads to the app exclusively notifying me when I am actively using the desktop app and never when I am actually willing to check it with my phone

blog post from deepseek kernel engineer mp.weixin.qq.com/s/zk0KxuLzhm...

waow… now i see why they call it the eiffel tower

OpenAI:

wow, I can see why they call it da space needle

maybe this is a universal truth i’m only discovering now, but working on internal tools is way more fun than doing literally anything else

Hi, I didn't wanna make this video but I need to come clean about something. This whole time I've been a p-zombie with no phenomenal experience or interiority. I've been getting messages asking whether there's anything it's like to be me, and the answer is no. I'd say I'm sorry, but, well, you know.

i've said it before but there really was no alternative to LLMs bootstrapping intelligence from raw trial and error is impractical, so you have to do foundation modeling. the only preexisting data broad enough for this is video or text, and signal:noise is way higher on text

micro-bnuuy

so, what percentage of the miraculous gains Mythos had in cybersecurity were the benefits of scaling up versus "reached threshold where all agentic environments are offensive cybersec environments"

Perhaps we should view the datacenter backlash as the tiktok algorithm hating itself and trying to commit suicide.

review of Teenage Sex and Death at Camp Miasma: I didn't like it but I can't stop thinking about it until I figure it out

gpu stands for "general processing unit", bc it handles the general case of computation, where you just want to do massively parallel SIMD stuff on giant ndarrays. cpu stands for "constrained processing unit", it's for use in the rare edge cases where you need to do single-threaded bullshit

That'll do it. aint go agent getting into my infra any time soon! Would like to see GPT-6 try to get past this...

how many of you actually check the hash when people post hashes to call things in advance and then share the apparent original string

gpt looks up at us claude looks down on us only gemma-4-26B-A4B-it-uncensored-recooked.safetensors treats us as equals

i have no idea what this contraption does but I need it

congratulations to the Moonshot team for extending Claude Fable 5's inclusion in claude subscriptions for another few weeks