DigThatData's profile picture. Generative AI MLE, FOSS toolmaker, innovation catalyst @CoreWeave + @AiEleuther. https://bsky.app/profile/digthatdata.bsky.social

David Marx (@digthatdata.bsky.social)

@DigThatData

Generative AI MLE, FOSS toolmaker, innovation catalyst @CoreWeave + @AiEleuther. https://bsky.app/profile/digthatdata.bsky.social

Remember: by participating on twitter, you are adding value to it, thereby incentivizing others to join/stay, and amplifying the voice/reach of the people who set its algorithm's biases. Is this the ideology you want to amplify? Because if you are here, you are amplifying this.

DigThatData's tweet image. Remember: by participating on twitter, you are adding value to it, thereby incentivizing others to join/stay, and amplifying the voice/reach of the people who set its algorithm's biases.

Is this the ideology you want to amplify? Because if you are here, you are amplifying this.

TAXATION WITHOUT REPRESENTATION.


David Marx (@digthatdata.bsky.social) đã đăng lại

'we're in this bizarre world where the best way to learn about llms... is to read papers by chinese companies. i do not think this is a good state of the world' - us labs keeping their architectures and algorithms secret is ultimately hurting ai development in the us.


> Republicans: "We love America! It has the greatest system of governance. Look, I even carry the constitution next to my heart like a little bible :*) " > Also republicans: "DISMANTLE THE GOVERNMENT! FUCK THE SEPARATION OF POWERS! GOD KING PRESIDENT CULT OF PERSONALITY!"


David Marx (@digthatdata.bsky.social) đã đăng lại

Happy Christmas friends


David Marx (@digthatdata.bsky.social) đã đăng lại

🔥 Are you ever dissatisfied with the imprecise names in vision-language datasets? 🚀 At #NeurIPS2024, we introduce 𝐑𝐄𝐍𝐎𝐕𝐀𝐓𝐄, showing how better segmentation dataset names lead to 𝐛𝐞𝐭𝐭𝐞𝐫 𝐭𝐫𝐚𝐢𝐧𝐢𝐧𝐠 & 𝐞𝐯𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧. Let’s dive in! 🧵👇

HaiwenHuang_'s tweet image. 🔥 Are you ever dissatisfied with the imprecise names in vision-language datasets?

🚀 At #NeurIPS2024, we introduce 𝐑𝐄𝐍𝐎𝐕𝐀𝐓𝐄, showing how better segmentation dataset names lead to 𝐛𝐞𝐭𝐭𝐞𝐫 𝐭𝐫𝐚𝐢𝐧𝐢𝐧𝐠 & 𝐞𝐯𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧.

Let’s dive in! 🧵👇

Yo this paper is wild.

🌟Announcing NeurIPS spotlight paper on the transition from lazy to rich🔦 We reveal through exact gradient flow dynamics how unbalanced initializations promote rapid feature learning co-led @AllanRaventos and @ClementineDomi6 @FCHEN_AI @klindt_david @SaxeLab @SuryaGanguli



David Marx (@digthatdata.bsky.social) đã đăng lại

New NanoGPT training speed record: 3.28 FineWeb val loss in 4.66 minutes Previous record: 5.03 minutes Changelog: - FlexAttention blocksize warmup - hyperparameter tweaks

hi_tysam's tweet image. New NanoGPT training speed record: 3.28 FineWeb val loss in 4.66 minutes

Previous record: 5.03 minutes
Changelog: 
- FlexAttention blocksize warmup
- hyperparameter tweaks

David Marx (@digthatdata.bsky.social) đã đăng lại

This is officially the new record! Congrats @hi_tysam (who is also an OG of CIFAR-10 speedrunning) x.com/hi_tysam/statu…

New NanoGPT training speed record: 3.28 FineWeb val loss in 4.66 minutes Previous record: 5.03 minutes Changelog: - FlexAttention blocksize warmup - hyperparameter tweaks

hi_tysam's tweet image. New NanoGPT training speed record: 3.28 FineWeb val loss in 4.66 minutes

Previous record: 5.03 minutes
Changelog: 
- FlexAttention blocksize warmup
- hyperparameter tweaks


David Marx (@digthatdata.bsky.social) đã đăng lại

Is KL-regularization the right tool for language model alignment? The χPO algorithm: We show that a one-line change to DPO—moving from KL to chi-squared regularization—is sufficient to achieve state-of-the-art theoretical guarantees, provably alleviating over-optimization.

canondetortugas's tweet image. Is KL-regularization the right tool for language model alignment? 

The χPO algorithm: We show that a one-line change to DPO—moving from KL to chi-squared regularization—is sufficient to achieve state-of-the-art theoretical guarantees, provably alleviating over-optimization.
canondetortugas's tweet image. Is KL-regularization the right tool for language model alignment? 

The χPO algorithm: We show that a one-line change to DPO—moving from KL to chi-squared regularization—is sufficient to achieve state-of-the-art theoretical guarantees, provably alleviating over-optimization.

DigThatData's tweet image.

And just like that, the AI/ML migration off twitter finally happened. RIP, this shitty pay-to-play platform.

Narrative on X: 🦋 has no AI/ML and just talks about itself My actual feed on 🦋:

jeremyphoward's tweet image. Narrative on X: 🦋 has no AI/ML and just talks about itself
My actual feed on 🦋:
jeremyphoward's tweet image. Narrative on X: 🦋 has no AI/ML and just talks about itself
My actual feed on 🦋:
jeremyphoward's tweet image. Narrative on X: 🦋 has no AI/ML and just talks about itself
My actual feed on 🦋:
jeremyphoward's tweet image. Narrative on X: 🦋 has no AI/ML and just talks about itself
My actual feed on 🦋:


David Marx (@digthatdata.bsky.social) đã đăng lại

Narrative on X: 🦋 has no AI/ML and just talks about itself My actual feed on 🦋:

jeremyphoward's tweet image. Narrative on X: 🦋 has no AI/ML and just talks about itself
My actual feed on 🦋:
jeremyphoward's tweet image. Narrative on X: 🦋 has no AI/ML and just talks about itself
My actual feed on 🦋:
jeremyphoward's tweet image. Narrative on X: 🦋 has no AI/ML and just talks about itself
My actual feed on 🦋:
jeremyphoward's tweet image. Narrative on X: 🦋 has no AI/ML and just talks about itself
My actual feed on 🦋:

David Marx (@digthatdata.bsky.social) đã đăng lại

You can choose what feeds you have on your homepage -- I personally like the "popular with friends" feed (in the above tweet). The default is "following" - which looks like this for me:

jeremyphoward's tweet image. You can choose what feeds you have on your homepage -- I personally like the "popular with friends" feed (in the above tweet). The default is "following" - which looks like this for me:
jeremyphoward's tweet image. You can choose what feeds you have on your homepage -- I personally like the "popular with friends" feed (in the above tweet). The default is "following" - which looks like this for me:
jeremyphoward's tweet image. You can choose what feeds you have on your homepage -- I personally like the "popular with friends" feed (in the above tweet). The default is "following" - which looks like this for me:
jeremyphoward's tweet image. You can choose what feeds you have on your homepage -- I personally like the "popular with friends" feed (in the above tweet). The default is "following" - which looks like this for me:

David Marx (@digthatdata.bsky.social) đã đăng lại

Because X has tended to censor discussion of social networks I won't link directly, but look for this post to get an instant AI/ML feed thanks to @maosbot

jeremyphoward's tweet image. Because X has tended to censor discussion of social networks I won't link directly, but look for this post to get an instant AI/ML feed thanks to @maosbot

David Marx (@digthatdata.bsky.social) đã đăng lại

We announce LAION-DISCO-12M - a collection of 12 million links to publicly available YouTube samples paired with metadata to support basic machine learning research in foundation models for generic audio and music. laion.ai/blog/laion-dis…


BRRRRRRRRRRRRRR

We’re proud to bring up the first @NVIDIA GB200 NVL72 from @Dell with NVIDIA Quantum InfiniBand, setting a new bar for AI infrastructure. This wouldn’t have been possible without the support of our valued partners at @Dell and @Switch.

CoreWeave's tweet image. We’re proud to bring up the first @NVIDIA GB200 NVL72 from @Dell with NVIDIA Quantum InfiniBand, setting a new bar for AI infrastructure. This wouldn’t have been possible without the support of our valued partners at @Dell and @Switch.
CoreWeave's tweet image. We’re proud to bring up the first @NVIDIA GB200 NVL72 from @Dell with NVIDIA Quantum InfiniBand, setting a new bar for AI infrastructure. This wouldn’t have been possible without the support of our valued partners at @Dell and @Switch.
CoreWeave's tweet image. We’re proud to bring up the first @NVIDIA GB200 NVL72 from @Dell with NVIDIA Quantum InfiniBand, setting a new bar for AI infrastructure. This wouldn’t have been possible without the support of our valued partners at @Dell and @Switch.


Loading...

Something went wrong.


Something went wrong.