Maor Ashkenazi

[Maor Ashkenazi @AshMaor]

🎉 Excited to share that our paper "Towards Croppable Implicit Neural Representations" has been accepted to #NeurIPS2024! 🧵

Paper: https://t.co/IT3n3eL3iD
Code: https://t.co/DLCHkCiY79
Additional Results: https://t.co/CLdbJaeWrv

#NeurIPS2024 #NeurIPS

[1/8]

[Ron Vainshtein @ronvain]

🚀Using BFMs to learn a new task with less than 0.1% additional params while still enabling user input??🚀
Task Tokens: A Flexible Approach to Adapting Behavior Foundation Models
📄: https://t.co/Q5sA9eMhUM
💻: https://t.co/N8TDfJDms5
w/@ZoharRimon @shiemannor @ChenTessler
🧵1/10

[Maor Ashkenazi @AshMaor]

@sarahookr https://t.co/Qo5bsvq9tP
🤩

[Maor Ashkenazi @AshMaor]

Excited to present my paper now at NeurIPS 2024! 🥳
Stop by the first poster session to chat about improving Implicit Neural Representations 🤩
Looking forward to sharing insights and meeting fellow AI enthusiasts!

@ East Exhibit Hall, poster 1502

#NeurIPS2024 #NeurIPS #inrs

[Maor Ashkenazi @AshMaor]

🎉 Excited to share that our paper "Towards Croppable Implicit Neural Representations" has been accepted to #NeurIPS2024! 🧵

Paper: https://t.co/IT3n3eL3iD
Code: https://t.co/DLCHkCiY79
Additional Results: https://t.co/CLdbJaeWrv

#NeurIPS2024 #NeurIPS

[1/8]

[Maor Ashkenazi @AshMaor]

🧠We believe Local-Global INRs offer a new direction for INRs that support modifications by-design, laying a foundation for more flexible representations

[8/8]

[Maor Ashkenazi @AshMaor]

🎉 Excited to share that our paper "Towards Croppable Implicit Neural Representations" has been accepted to #NeurIPS2024! 🧵

Paper: https://t.co/IT3n3eL3iD
Code: https://t.co/DLCHkCiY79
Additional Results: https://t.co/CLdbJaeWrv

#NeurIPS2024 #NeurIPS

[1/8]

[Maor Ashkenazi @AshMaor]

✂️After training, Local-Global SIRENs allow cropping without retraining.

Here’s an example with video - cropping both spatial and temporal dimensions

[7/8]

[Yann LeCun @ylecun]

Punchline needs an update:
"Sure, easy image recognition with a ConvNet. Gimme PyTorch and a few hours."

[Hila Chefer @hila_chefer]

Text-to-image models revolutionized computer vision, but what do they learn about the world?🤔

We present Conceptor 🧐 a method to inspect the inner representation of a concept by decomposing it into a set of interpretable tokens, which reveals surprising semantic structures🧵

[Maor Ashkenazi @AshMaor]

Excited to be in Kigali, Rwanda for #ICLR2023!
I'll be presenting our work titled "NeRN - Learning Neural Representations for Neural Networks" at the first oral session (track 6). Right after that I'll be presenting a poster 🤩🥳
#NeRN

[AK @_akhaliq]

Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images

abs: https://t.co/HILwgrg907
project page: https://t.co/SRhgnwsD92
@huggingface dataset: https://t.co/EMSxn1dMFT
@gradio demo: https://t.co/iS03zrnDIn

[AK @_akhaliq]

Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models

abs: https://t.co/tNBuPSi5qM
project page: https://t.co/EEgPxsVuo7
github: https://t.co/G2LNrgYRQb

[Maor Ashkenazi @AshMaor]

Delighted to share that #NeRN was accepted as a notable-top-25% (spotlight) paper to #ICLR2023 #ICLR 🥳🤩

[Maor Ashkenazi @AshMaor]

I’m very excited to share our new work titled “NeRN - Learning Neural Representations for Neural Networks”. Preprint and code are available now😊🧵
@ZoharRimon @ronvain @levishir667 @EladRichardson @PinkyMintz

Paper: https://t.co/TUYgOhVrxk
Code: https://t.co/dYHqyHIY7h

[1/9]

[Maor Ashkenazi @AshMaor]

We hope that our proposed NeRN will give a new perspective on neural networks for future research.

[9/9]

[Maor Ashkenazi @AshMaor]

I’m very excited to share our new work titled “NeRN - Learning Neural Representations for Neural Networks”. Preprint and code are available now😊🧵
@ZoharRimon @ronvain @levishir667 @EladRichardson @PinkyMintz

Paper: https://t.co/TUYgOhVrxk
Code: https://t.co/dYHqyHIY7h

[1/9]

[Maor Ashkenazi @AshMaor]

For example, a limited capacity NeRN will most likely prioritize the weights during training. The reconstruction error can be explored to study the importance of different weights.

[8/9]