Marco Ciccone

Postdoctoral Fellow @VectorInst - Collaborative, Decentralized, Modular ML - Competition chair @NeurIPSConf 2021, 2022, 2023 - PhD @polimi ex @NVIDIA @NNAISENSE

[Marco Ciccone @mciccone_AI]

🚨 Life update 🚨 I moved to Toronto 🇨🇦and joined @VectorInst as a Postdoctoral Fellow to work with @colinraffel and his lab on collaborative, decentralized, and modular machine learning to democratize ML model development. Exciting times ahead! 🪿

[Marco Ciccone @mciccone_AI]

If you are interested in elastic deployment, have a look at our work on turning pre-trained models into any size! 🧵from @RickZack96 below

[Riccardo Zaccone @RickZack96]

Are curious about how to resize any pretrained model on demand?

Then you'll love our paper "FlexRank: Nested Low-Rank Knowledge Decomposition for Adaptive Model Deployment​", accepted at ICML 2026 (Spotlight)

Joint work with @sam_hrvth, @stevelaskaridis, @mciccone_AI

🧵1/7

[Gintare Karolina Dziugaite @gkdziugaite]

⏳ Don't miss the chance to share your research on Continual Adaptation @ CATS workshop for #ICML2026.

Working on Scale & Efficiency, Alignment, Multimodality, or Forgetting? Get your 4-page submissions in.

🗓️ Deadline: April 30, 2026 (23:59 AoE)🔗 https://t.co/7aqeP342EN

[Marco Ciccone @mciccone_AI]

@Ar_Douillard is this scary innovative?

[Marco Ciccone @mciccone_AI]

Distributed training maxxing from the diloco team 🌐

[Arthur Douillard @Ar_Douillard]

The DiLoCo team at Google DeepMind and Google Research is proud to release Decoupled DiLoCo, the next frontier for resilient AI pre-training.

Decoupled DiLoCo enables training with datacenters across the world, using heterogeneous hardware, and never halting the system despite hardware failures.

[Marco Ciccone @mciccone_AI]

@samsja19 @Ar_Douillard @eliebakouch I always had a cat in my picture, but my face ruined it

[Marco Ciccone @mciccone_AI]

@eliebakouch @PrimeIntellect 🚀🚀🚀

[Marco Ciccone @mciccone_AI]

@soldni @allen_ai Congratulations on your amazing run, and best of luck for what's ahead!

[Marco Ciccone @mciccone_AI]

@MatharyCharles exactly @MatharyCharles. Just thinking about all the NeurIPS submissions in the pipeline...

[Marco Ciccone @mciccone_AI]

@thegautamkamath One-page rebuttals are sufficient in most situations and easier to verify. If reviewers formulate their concerns precisely (unfortunately harder and harder), authors should be able to address them directly and get straight to the point. No fluff.

[Marco Ciccone @mciccone_AI]

@MatharyCharles 100% agreed. But we all have some toxic traits and obsessive behaviours.

[Marco Ciccone @mciccone_AI]

@m_sirovatka I believe (in) you

[Marco Ciccone @mciccone_AI]

@m_sirovatka lol

[Marco Ciccone @mciccone_AI]

@manthanguptaa nice write-up. We have released a study of different tokenizers' characteristics and a multilingual robustness benchmark if you are interested https://t.co/DI6zsgoNP3

[Marco Ciccone @mciccone_AI]

@m_sirovatka I generally don't do that, but when I do, I feel a certain peace of mind in knowing other people are sleeping

[Marco Ciccone @mciccone_AI]

So Claude implemented a long context compression feature! Wondering what method they use

[Marco Ciccone @mciccone_AI]

@SonglinYang4 @rupspace I choose you

[Marco Ciccone @mciccone_AI]

@kchonyc haha +1 for PAI, Mother's Dumpling is good, and BIWON for Korean food! Amal for fancy Lebanese food. I also like Momo Ghar (more east), and RAIJIN or Ikkousha for ramen

[Marco Ciccone @mciccone_AI]

@kchonyc useful for faculty applications and collaborations!

[Marco Ciccone @mciccone_AI]

@samsja19 @_fracapuano @lexfridman Let rephrase: I demand that shirt

[Marco Ciccone @mciccone_AI]

@samsja19 @_fracapuano @lexfridman I need that shirt @samsja19