Tyler LaBonte

MTS @MicrosoftAI working on frontier VLM training and multimodal reinforcement learning. ML Theory PhD @GeorgiaTech, Math BS @USC.

[Tyler LaBonte @tmlabonte]

@JohnCLangford Thanks for sharing, John. It was a great run -- happy to have been a part of it, if only for a little while.

[Tyler LaBonte @tmlabonte]

Exciting 3T native multimodal pretraining happening at @MicrosoftAI! Post-training 🔜

[Mustafa Suleyman @mustafasuleyman]

We now have state-of-the-art pre-trained base models in the 3T-parameter weight class, built entirely in-house at MAI.

Post-training results coming soon -- stay tuned!

Lots of details in our technical blog post.

https://t.co/xcZ5uV4TAp

[Tyler LaBonte @tmlabonte]

@canondetortugas @UCBerkeley Congrats Dylan! Berkeley is lucky to have you!

[Tyler LaBonte @tmlabonte]

This work has been accepted to NeurIPS 2026 - see you in Atlanta!

[Tyler LaBonte @tmlabonte]

New paper — the last and best from my PhD!

We show the first end-to-end feature learning analysis of how neural networks overfit to spurious correlations. Our theory implies phase transitions in SGD learning dynamics that match empirical results.

Paper: https://t.co/V0GALNgLdO

[Tyler LaBonte @tmlabonte]

@sid_devic I somehow missed this when you posted, but very nice and thoughtful writeup echoing my own decision to "leave" theory. It's also interesting to think about the role of the new-age theoretician as more interpretive/expositional as AI proofs need to be communicated and organized.

[Tyler LaBonte @tmlabonte]

I've transitioned to Member of Technical Staff @MicrosoftAI focused on frontier VLM training and multimodal reinforcement learning. Looking forward to an exciting new chapter!

[Tyler LaBonte @tmlabonte]

@GrantStenger Compaction still sucks

[Tyler LaBonte @tmlabonte]

@lmathur_ @GrantStenger I still wonder whether they'll make a comeback. I think they'd make much better skill docs than the typical SKILL.md which just has text. Combining instructions, code with verified outputs, and charts/plots... seems pretty powerful and I haven't seen anyone thinking about it

[Tyler LaBonte @tmlabonte]

@jxbz Congrats really cool work! And great to see a strong new open-weight model family

[Tyler LaBonte @tmlabonte]

@ajonelagadda Thanks! (And that's a nice paper!) Yes, I think there is promise in SDE and mean field-type approaches for spurious feature learning. Of note, our Phase II spurious evolution is closely related to the classical ODE dot(x) = x * e^-(x^2).

[Tyler LaBonte @tmlabonte]

New paper — the last and best from my PhD!

We show the first end-to-end feature learning analysis of how neural networks overfit to spurious correlations. Our theory implies phase transitions in SGD learning dynamics that match empirical results.

Paper: https://t.co/V0GALNgLdO

[Tyler LaBonte @tmlabonte]

@_Suresh2 Good question: this is the online setting (fresh data at every step), so no such thing as full-batch GD. This is to avoid data reuse, mostly for mathematical convenience.

With that said, the code wouldn't be too hard to extend to full-batch GD: https://t.co/853xDoXsuD

[Tyler LaBonte @tmlabonte]

@fabianfranz Our theory requires the % of minority group data (i.e. data with the spurious feature) to go to zero, to guarantee total dependence on the spurious feature.

OTOH, if the % is constant, our experiments show the true signal (blue) can eventually surpass the spurious signal (green)

[Tyler LaBonte @tmlabonte]

Overall we believe this is a step towards a rigorous foundation for popular robustness algorithms like Just Train Twice and Deep Feature Reweighting, and could even inform new robust interventions.

This paper was the result of a wonderful collaboration with @v__muthukumar!

[Tyler LaBonte @tmlabonte]

Even more interesting, if the spurious correlation is sufficiently weak, we show empirically that the XOR signal can eventually be learned. (Though the spurious feature is not forgotten!)

[Tyler LaBonte @tmlabonte]

Very cool work from new colleagues at @ms_aifrontiers: most LLM benchmarks are redundant!

[Microsoft AI Frontiers @ms_aifrontiers]

Most LLM benchmark scores are predictable before you ever run them.
New from the MS AI Frontiers team: BenchPress. The 84-model × 133-benchmark score matrix turns out to be effectively rank-2, so matrix completion fills in the rest. 5 probes recover a model's whole profile.
Paper, code, demo 👇

[Tyler LaBonte @tmlabonte]

@julian_asilis @GeorgiaTech @MSFTResearch @ms_aifrontiers Thanks Julian!!

[Tyler LaBonte @tmlabonte]

Overdue update: I've defended my dissertation and graduated from @GeorgiaTech! Next, I've joined @MSFTResearch @ms_aifrontiers as a Senior Researcher working on future capabilities of multimodal agents 🚀

[Tyler LaBonte @tmlabonte]

@v__muthukumar Thanks Vidya! It was truly a pleasure to collaborate with and learn from you!