Adriana Romero-Soriano
@adri_romsor
Followers
1K
Following
920
Media
32
Statuses
475
Researcher @ FAIR, Meta AI. Adjunct prof @ McGill. Core industry member @ Mila. Opinions are my own. (she/her)
Montreal
Joined February 2014
This work wouldn't have been possible without my incredible internship at Meta AI in Montreal. A massive thanks to the team for their support and for making it such a memorable experience! @AZUREANg @ReyhaneAskari @hall__melissa @JakobVerbeek @ @michal_drozdzal @adri_romsor
1
1
2
"Increasing the Utility of Synthetic Images through Chamfer Guidance" has been accepted at NeurIPS 2025! 🎉 We're happy to share our approach for generating synthetic images that are both diverse and high-quality. #NeurIPS2025 Paper: https://t.co/rITziRNqk5
1
6
14
Generative Point Tracking with Flow Matching My latest project with @AdamWHarley @CSProfKGD @DerekRenderling @chrisjpal Project page: https://t.co/cs4zFEuLYU Paper: https://t.co/sa9NdFlOgP Code: https://t.co/F4Ug3JWkRX
2
25
134
After 7 years at FAIR, I've been affected by the recent AI layoffs. If you are interested in robotics learning, let's chat :)
63
54
1K
🌺 Attending @ICCVConference in Honolulu this week! I'll be presenting our work on multimodal reward-guided decoding. Come check it out on October 21 (morning), poster #122. If you’re around, I’d love to connect and chat about multimodal models and real-time video generation!
I’m happy to share that our paper "Controlling Multimodal LLMs via Reward-guided Decoding" has been accepted to #ICCV2025! 🎉 w/ @proceduralia, @koustuvsinha, @adri_romsor, @michal_drozdzal, and @aagrawalAA 🔗 Read more: https://t.co/wIRL9jsAr1 🧵 Here's what we did:
0
6
20
1/ My model can’t generate “full” refrigerators when I ask for it?? — but it does by default anyway when I don’t specify!? At #ICCV2025, we’re presenting DIMCIM — a new framework to quantify diversity and capacity in T2I models, that enables pinpointing such failure modes
2
5
9
7/ Come by our poster session on Wed 22 Oct Afternoon at #ICCV2025 Exhibit Hall I #1511 to chat! Co-authors: Candace Ross @candacerossio, Karen Ullrich @karen_ullrich, Sumit Chopra, Adriana Romero-Soriano @adri_romsor, Melissa Hall @hall__melissa, Matthew Muckley @mattmucklm
0
1
2
Transfusion combines autoregressive with diffusion to train a single transformer, but what if we combine Flow with Flow? 🤔 🌊OneFlow🌊 the first non-autoregressive model to generate text and images concurrently using a single transformer—unifying Edit Flow (text) with Flow
7
83
408
I’m happy to share that our paper "Controlling Multimodal LLMs via Reward-guided Decoding" has been accepted to #ICCV2025! 🎉 w/ @proceduralia, @koustuvsinha, @adri_romsor, @michal_drozdzal, and @aagrawalAA 🔗 Read more: https://t.co/wIRL9jsAr1 🧵 Here's what we did:
2
33
117
With two guest speakers presenting their work, Aaron and Salma!
0
1
2
Excited for our final Speaker, Olga Russakovsky, presenting on “Revealing Bias in Data and Models”!
1
1
3
Now we’ve got our first contributed talk: “Understanding and Mitigating Toxicity in Image-Text Pretraining Datasets”!
0
1
2
Our second Contributed Talk is focused on “Energy-Conscious Optimization for AI Training”!
0
1
2
Next up we have Chloé Bakalar presenting on responsible generative AI at Meta!
0
1
2
A great opening talk from @adri_romsor and next up is one of our invited speakers Tamar Rott!
0
1
2
Our first invited Speaker is Tamar Rott, presenting on “Scaling AI Auditing with an Automated Interpretability Agent” 📈
0
1
2
Less than one week away from the 2nd Workshop on Responsible Generative AI @CVPR! It’ll be on Thursday afternoon, and we’ve got a great set of presenters and participants! Check out the program & accepted work: https://t.co/r5VPOunpAl See you in Nashville! 🎶 #cvpr2025
sites.google.com
Workshop Schedule
0
3
5
Deliberate practice is accepted to #ICML2025 as a spotlight (top 2.6%!) 🚀
🚀 New Paper Alert! Can we generate informative synthetic data that truly helps a downstream learner? Introducing Deliberate Practice for Synthetic Data (DP)—a dynamic framework that focuses on where the model struggles most to generate useful synthetic training examples. 🔥
1
19
153