Nastya Filippova🇺🇦

@NasFilippova

ML research at @apple, previously @EPFL, WorldQuant, MIPT

Paris

nastya236.github.io

Replies Reposts Media

Nastya Filippova🇺🇦
@NasFilippova
Sep 2
Mostik=bridge in Russian! Very cool to see this direction around latent-space reasoning. Curious about what it could unlock for local AI.
Sasha Malysheva
@aimalysheva
Sep 2
meet @mostik_ai! what happens when you put 12 PhDs in one room for four months? first place on the ARC-AGI leaderboard, which I can't say much about while the competition is still running. and this, which I can. everyone's arguing about whether open models will catch up to
2
Nastya Filippova🇺🇦
@NasFilippova
Aug 10
I was working on some optimization for mlx and noticed that our rmsnorm backward was hitting significantly lower bandwidth than forward. I wrote a small note on why that’s the case and how it could be fixed. nastya236.github.io/blog/rmsnorm-b…
1
Nastya Filippova🇺🇦
@NasFilippova
Feb 28
Working with Awni and MLX team has been the best professional experience so far. I’m still very junior and don’t have many years behind me yet, but it already feels like it’ll be hard to beat.
Awni Hannun
@awnihannun
Feb 27
Today is my last day at Apple. Building MLX with our amazing team and community has been an absolute pleasure. It's still early days for AI on Apple silicon. Apple makes the best consumer hardware on the planet. There's so much potential for it to be the leading platform for
3
Nastya Filippova🇺🇦
@NasFilippova
Feb 11, 2025
Update: Our paper has been accepted as a Spotlight at ICLR 2025🎉 A huge thank you to my amazing co-authors @angeloskath , @GrangierDavid , and Ronan Collobert, as well as the entire team @awnihannun, @DiganiJagrit , and @alex_barron1!
Nastya Filippova🇺🇦
@NasFilippova
Jan 22, 2025
Thrilled to share that our work No Need to Talk: Asynchronous Mixture of Language Models [arxiv.org/abs/2410.03529] has been accepted to #ICLR2025! In this paper, we explore strategies to mitigate the communication cost of large language models, both at training and inference,
1
Nastya Filippova🇺🇦
@NasFilippova
Jan 22, 2025
Thrilled to share that our work No Need to Talk: Asynchronous Mixture of Language Models [arxiv.org/abs/2410.03529] has been accepted to #ICLR2025! In this paper, we explore strategies to mitigate the communication cost of large language models, both at training and inference,
Nastya Filippova🇺🇦
@NasFilippova
Oct 11, 2024
Don’t have fast interconnect? No problem! No need for nodes to talk! We introduce SMALLTALK LM, an innovative method for training a mixture of language models (almost) asynchronously. SMALLTALK LM achieves better perplexity, and better accuracy on a majority of downstream tasks,