Sadly i don't think DroPE will work for images / videos.
Both NoPE and DroPE rely on the causal mask to leak absolute PE. The number of tokens in the attention gets leaked because you can encode a bias that grows with the number of tokens.
So not a fix for images yet =(

It was a big pleasure to be in Nicolas's committee. Congratulations to Nicolas for the great work, and congratulations to the advisors too!

Apparently some people reported knowing of the bug before 11th of november so even before the release of the reviews

Congrats Nicolas ! On the PhD and on those beautifully crafted slides 🤩

Yes it's latent space just because i had my setup that way. Might try in pixel space in the future.

Yes it's the raw prediction, we predict the velocity directly

It's also very domain dependent. I know that for example, x-pred works better than epsilon pred for human motion generation.

Epsilon loss was used for a while for image generation since DDPM.
Recently it was more flow matching (or v-loss) that is mostly used since SD3 basically.
From my experience, flow doesn't really improve quality, but sampling in fewer steps works better than epsilon prediction

Thanks for the pointer! We were doing something similar in "Don't drop your samples" (arxiv.org/abs/2405.20324)

MIRO is quite different in the sense we focus on improving pretraining (not finetuning). Also, we explore the advantages of having multiple rewards to push the Pareto frontier.

Yes, thanks for pointing it out, will try to clarify

Work with @lucasdegeorge.bsky.social @arrijitghosh.bsky.social @vickykalogeiton.bsky.social and @davidpicard.bsky.social.

This will be the last work of my PhD as I will be defending the 26th of November!

MIRO demonstrates that aligning T2I models during pretraining is not only viable but superior: it's faster, more compute-efficient, and provides fine-grained, interpretable control.

Project page for all the details: nicolas-dufour.github.io/miro

The explicit reward conditioning allows for flexible trade-offs, like optimizing for GenEval by reducing the aesthetic weight in the prompt. We can also isolate the look of a specific reward or interpolate them via multi-reward classifier-free guidance

MIRO excels on challenging compositional tasks (Geneval here)

The multi-reward conditioning fosters better understanding of complex spatial relationships and object interactions.

Despite being a compact model (0.36B parameters), MIRO achieves state-of-the-art results:

GenEval score of 75, outperforming the 12B FLUX-dev (67) for 370x less inference cost.
Conditioning on rich reward signals is a highly effective way to achieve large model capabilities in a compact form!

MIRO dramatically improves sample efficiency for test-time scaling.

On PickScore, MIRO needs just 4 samples to match the baseline's 128 samples (a 32x efficiency gain).
For ImageReward, it's a 16x efficiency gain

This demonstrates superior inference-time efficiency for high-quality generation.

Traditional single-objective optimization often leads to reward hacking. MIRO's multi-dimensional conditioning naturally prevents this by requiring the model to balance multiple objectives simultaneously. This produces balanced, robust performance across all metrics contrary to single rewards.

The multi-reward conditioning provides a dense supervisory signal, accelerating convergence dramatically. A snapshot of the speed-up:

AestheticScore: 19.1x faster to reach baseline quality.
HPSv2: 6.2x faster.

You can clearly see the improvements visually

This reward vector s becomes an explicit, interpretable control input at inference time. We extend classifier-free guidance to the multi-reward setting, allowing users to steer generation toward jointly high-reward regions by defining positive (s^+) and negative (s^−) targets.