@DejounteMurray Hey man would love to dm u, I dmed on instagram too 🙏🏽
English
ryan mathieu
258 posts
ryan mathieu
@gapDEEPry
kernel guy. || Fast vs Slow Thinking
Inside Kernels Katılım Ocak 2018
526 Takip Edilen148 Takipçiler
I Feel Like Giving A Couple Of My FOLLOWERS That FW Me $5,000 Each!!! 🖤 I Really Love The REAL AND GENUINE ONES!!!!!!!
English
@kalomaze @teortaxesTex Nvm seem to have found your thought here
English
@teortaxesTex i will eat crow the second reliable estimates of in-context learning scaling well with diffusion lms crops up here
English
> MDM-Prime-v2 is 21.8× more compute-efficient than autoregressive models
I may be humiliated extremely hard with my diffusionLM skepticism.
You Jiacheng@YouJiacheng
HUGE if true. If true, this is probably a larger efficiency gain than ALL publicly available techniques since DeepSeekMoE(Jan 2024) COMBINED. And it can just win modded-nanogpt speedrun. (1e18 is 250s@50%MFU, but the loss is significantly lower than 3.28) cc @classiclarryd
English
English
HUGE if true.
If true, this is probably a larger efficiency gain than ALL publicly available techniques since DeepSeekMoE(Jan 2024) COMBINED.
And it can just win modded-nanogpt speedrun.
(1e18 is 250s@50%MFU, but the loss is significantly lower than 3.28)
cc @classiclarryd
Chen-Hao (Lance) Chao@chenhao_chao
(2/7) 💵 With training costs exceeding $100M for GPT-4, efficient alternatives matter. We show that diffusion LMs unlock a new paradigm for compute-optimal language pre-training.
English
@xeophon Idk if it’s the prime effect but you are either tweeting way more or you are now all over my feed (no complaints)
English
You guys think Honey Bunnie actually cares about distribution of training data of svg in frontier LLMs?
English
I'm so excited to introduce this!
We've worked on a million different moving parts to produce this. I'm fairly confident it's the best multimodal model that exists, period -- and it's not too shabby at pushing back the LIMITs of retrieval either...
Mixedbread@mixedbreadai
Introducing Mixedbread Wholembed v3, our new SOTA retrieval model across all modalities and 100+ languages. Wholembed v3 brings best-in-class search to text, audio, images, PDFs, videos... You can now get the best retrieval performance on your data, no matter its format.
English
@fkasummer pricing power on things like compute / hiring / acquisitions. investors largely don’t care about realized profit at this stage, they care about theoretical margins + growth. if you’re profitable as a pre-IPO AI SaaS company in 2026, you’re probably not spending enough on R&D
English
Why would you dilute your equity if you have healthy margins?
Techmeme@Techmeme
Sources: Cursor is in early talks to raise billions in a new funding round at a post-money valuation of up to $60B; one source puts the round size at $5B (@cityofthetown / Newcomer) newcomer.co/p/cursor-weigh… #a260311p57" target="_blank" rel="nofollow noopener">techmeme.com/260311/p57#a26…
📥 Send tips! techmeme.com/contact English
you can create complex agentic environments and launch RL training runs with a single prompt.
deploy trained inference endpoints with a single click.
no GPUs, no SSH, no vLLM. just `prime`.
guide: docs.primeintellect.ai/guides/rl-trai…
English
How do you talk to Claude? Say, when using Claude Code, at the end of a session you asked it to write its learnings in a text file. In a new session:
A: "a previous claude wrote this summary, read it"
B: "you previously wrote this summary, read it"
English
@btgille @Steele_Jake @andyguoa @ninerpride_sol @mikeydsoftware @hinzan @5cottFive Yeah I experienced the saw the thing lol, but it does seem they want it as exclusive arm64 as you said it was not hard to make work on Intel (at all). I just was looking at threads on X and seemed enough people were complaining ab it and seemed easy enough to do
English
@gapDEEPry @Steele_Jake @andyguoa @ninerpride_sol @mikeydsoftware @hinzan @5cottFive they REALLY wanted that to be a mac arm64 only thing. the port was simple to all operating systems. toward the end they even used some js to try and force people not on official to update... it was funny and trivial to bypass with codex.
English
#issuecomment-4015274777" target="_blank" rel="nofollow noopener">github.com/openai/codex/i…
This comment should be a nice little work around as I had the same issue would love feedback if possible
Alex Strick van Linschoten@strickvl
@thsottiaux Intel Mac support for the codex app…
English
@btgille @Steele_Jake @andyguoa @ninerpride_sol @mikeydsoftware @hinzan @5cottFive Oh went through the work myself for no reason I guess and wasn’t able to find that repo … more people should just use this rather then my repackaged fix
English
@gapDEEPry @Steele_Jake @andyguoa @ninerpride_sol @mikeydsoftware @hinzan @5cottFive you mean this? my installer is out of date, but i got it working a couple weeks ago. github.com/helix4u/ported…
English