Mark Ibrahim
@marksibrahim
Followers
388
Following
165
Media
3
Statuses
59
Researching the dark arts of deep learning at Meta's FAIR (Fundamental AI Research) Lab
everywhere
Joined December 2012
Open-weights for our Llip multimodal vision-language model led by @lavoiems are public!. LLIP proposes new pre-training objective to capture the many ways to describe an image leading to strong performance across a suite of 22-zero shot benchmarks.
The code and model weights for this paper are finally open! Despite being a little late for releasing them, I hope you will find them useful!.Code: Models:.- (ViT-G): - (ViT-B):
0
1
13
RT @karen_ullrich: How would you make an LLM "forget" the concept of dog — or any other arbitrary concept? 🐶❓. We introduce SAMD & SAMI — a….
0
12
0
More details:
Excited to release AbstentionBench -- our paper and benchmark on evaluating LLMs’ *abstention*: the skill of knowing when NOT to answer!. Key finding: reasoning LLMs struggle with unanswerable questions and hallucinate!. Details and links to paper & open source code below!.🧵1/9
1
0
4
Check out the thread for details:. Paper: Code: 🧵2/2.
github.com
A holistic benchmark for LLM abstention. Contribute to facebookresearch/AbstentionBench development by creating an account on GitHub.
1
0
8
A good language model should say “I don’t know” by reasoning about the limits of its knowledge. Our new work AbstentionBench carefully measures this overlooked skill in leading models in an open-codebase others can build on!. We find frontier reasoning degrades models’ ability to
3
18
111
RT @polkirichenko: Join us at #CVPR2025 Demographic Diversity in Computer Vision workshop tomorrow!.📅 Wednesday, June 11, 9am-6pm.📍 room 21….
0
21
0
Join us as a PhD research intern at FAIR w/.@polkirichenko .@kamalikac .to start this summer or fall with a focus on open science into multimodal models, agents and beyond! Email polkirichenko@meta.com with the title [Prospective Intern 2025] and attach your CV if interested!.
1
0
20
RT @garridoq_: The last paper of my PhD is finally out ! Introducing."Intuitive physics understanding emerges from self-supervised pretrain….
0
165
0
RT @vlad_is_ai: 𝕏-CLR got accepted to ICLR 2025 @iclr_conf! See you in Singapore!.It was also recently mentioned in The Batch by @DeepLearn….
0
10
0
RT @haidertahan: 🚀 Excited to share our work at #NeurIPS2024! We show how billion parameter VLMs lose to a two-layer MLP on MNIST. Come by….
0
30
0
RT @hall__melissa: Work done w/ amazing collaborators @oscmansan, @ReyhaneAskari, @marksibrahim, @candacerossio, @Piovrasca, Tariq Berrada,….
0
2
0
We found MLM-U training can even outperform transformers trained with additional supervision from A* search traces, showing the promise of alternative learning objectives. Learn more on our site and code at
facebookresearch.github.io
0
0
1
Recently, we also applied the same MLM-U objective to maze navigation. We find when training parameter-matched transformers on identical data, MLM-U without any tweaks outperforms standard next token training across all maze grid sizes (up to 30x30).
1
0
0
We find MLM-U improves knowledge retrieval on Wikipedia-based questions and even outperforms a pretrained 7B Mistral model with a much smaller 100M parameter transformer trained from scratch! Come by our NeurIPS poster Exhibit Halls A-C #3204 11am PST on Thursday to learn more!.
1
0
0
We show training with a factorization agnostic objective, MLM-U (a variable ratio BERT-style loss with links to discrete diffusion), that predicts multiple tokens ahead and back can significantly mitigate the reversal curse!.
1
0
0
Language models struggle with the “reversal curse:” an inability to answer reformulations of a question. We show this stems from the standard next token learning objective in what we call “the factorization curse.”.
1
0
0
Can we boost transformers’ ability to retrieve knowledge and plan in maze navigation by only tweaking the learning objective? We emphatically say YES in our NeurIPS 2024 study! 🧵. w/ @WKitouni, Niklas Nolte, Mike Rabbat, @D_Bouchacourt , @adinamwilliams
1
4
14
RT @AIatMeta: New research from Meta FAIR: UniBench is a unified implementation of 50+ VLM benchmarks spanning a comprehensive range of car….
0
81
0
RT @_akhaliq: Meta announces UniBench. Visual Reasoning Requires Rethinking Vision-Language Beyond Scaling. discuss: .
0
35
0
RT @ylecun: A soft similarity graph improves contrastive learning for image recognition. By @vlad_is_ai and a cast of characters from Meta….
0
13
0