Daniel Scalena

Intern of TS @cohere | PhDing @unimib 🇮🇹 & @GroNlp 🇳🇱, interpretability et similia

[Daniel Scalena @daniel_sc4]

You can easily save up to 65% of compute while improving performance on reasoning tasks 🤯 👀

Meet EAGer: We show that monitoring token-level uncertainty lets LLMs allocate compute dynamically - spending MORE on hard problems, LESS on easy ones.
🧵👇

[Daniel Scalena @daniel_sc4]

personal update: I'll be starting an internship at @cohere, working on code agents, one of the most interesting things happening in AI right now. feeling really grateful for this one and genuinely excited to see where it goes. here we go! 👀

[Daniel Scalena @daniel_sc4]

Really glad to see this work accepted at ICML 2026🎉
Can’t wait to present this work in Seoul!

[Daniel Scalena @daniel_sc4]

You can easily save up to 65% of compute while improving performance on reasoning tasks 🤯 👀

Meet EAGer: We show that monitoring token-level uncertainty lets LLMs allocate compute dynamically - spending MORE on hard problems, LESS on easy ones.
🧵👇

[Daniel Scalena @daniel_sc4]

@glnmario My year has 11 months because one month’s salary goes to ACL

[Daniel Scalena @daniel_sc4]

Presenting this tomorrow at EACL2026, poster session at 9am

If you’re around come say hi, happy to chat about the work and ideas

More details in the thread 👇

[Daniel Scalena @daniel_sc4]

📢 New paper: Applied interpretability 🤝 MT personalization!

We steer LLM generations to mimic human translator styles on literary novels in 7 languages. 📚

SAE steering can beat few-shot prompting, leading to better personalization while maintaining quality.

🧵1/

[Daniel Scalena @daniel_sc4]

@GoodfireAI Nice work! I wonder, probe trained on answer choices needs known options. What if you probe model confidence and early exit there regardless of the answer it's thinking? I feel like after some t the model already knows and the rest is just overthinking

[Daniel Scalena @daniel_sc4]

@paradigmainc Ok I was trying to cook something to improve model’s scientific creativity, throwing the repo into flywheel feels like the next logical step

[Daniel Scalena @daniel_sc4]

@gsarti_ Gemini listened to its AGI intrusive thoughts on this

[Daniel Scalena @daniel_sc4]

@thelokasiffers They were pioneers vibecoding it with gpt2 back then

[Gabriele Sarti @gsarti_]

Happy to announce I will be mentoring a SPAR project this Spring! ✨Check out the programme and apply by Jan 14th to work with me on understanding and mitigating implicit personalization in LLMs, i.e. how models form hidden beliefs about users that shape their responses.

[Gabriele Sarti @gsarti_]

Now accepted at EACL main! Check it out! ⬇️

[Daniel Scalena @daniel_sc4]

Want models to translate in the style you actually like?

Our paper just got accepted at EACL Main 🚀, check out our work on using interpretability for MT personalization!

And, see you in Morocco! 🇲🇦

[Daniel Scalena @daniel_sc4]

📢 New paper: Applied interpretability 🤝 MT personalization!

We steer LLM generations to mimic human translator styles on literary novels in 7 languages. 📚

SAE steering can beat few-shot prompting, leading to better personalization while maintaining quality.

🧵1/

[Daniel Scalena @daniel_sc4]

@thelokasiffers Woo big congrats on the launch and shoutout for starting in Rome. Can’t wait to hear more!

[Daniel Scalena @daniel_sc4]

@Turn_Trout @GladiaLab SVs are approximate directions in the latent space. They look for exact matches in the latent space. This could make things harder, but I’m still curious to know!

[Daniel Scalena @daniel_sc4]

@andy_peng05 @Cohere_Labs @UvA_Amsterdam Hi, thank you very much! Good catch on the p@1, it was meant to be p@k (pass@k). We’ll fix it asap in the preprint!

[Daniel Scalena @daniel_sc4]

@ZotosLeo @FersiniE @MalvinaNissim @ahmetustun89 Also cc @jiawzhao and @FuYichao123 — I guess your great work on DeepConf is highly relevant to this project!

[Daniel Scalena @daniel_sc4]

You can easily save up to 65% of compute while improving performance on reasoning tasks 🤯 👀

Meet EAGer: We show that monitoring token-level uncertainty lets LLMs allocate compute dynamically - spending MORE on hard problems, LESS on easy ones.
🧵👇

[Daniel Scalena @daniel_sc4]

Takeaway: EAGer shows we can be MORE efficient & MORE effective by letting models focus compute where it matters most.
📄Paper: https://t.co/JCHBWnKhUX
💻Code: https://t.co/Cp5nr38DLk
✨Huge thanks to my mentors and collaborators @ZotosLeo @FersiniE @MalvinaNissim @ahmetustun89

[Daniel Scalena @daniel_sc4]

Had so much fun working on this! Thank you @Cohere_Labs for sharing our work!

[Cohere Labs @Cohere_Labs]

How can we make reasoning models more efficient without sacrificing performance? 

Introducing EAGER, our new entropy-aware generation method, saving compute by up to 65% while lifting Pass@k by up to 37% on benchmarks like AIME.