Jean-Philippe Monteuuis

Don't hesitate to apply to this opening in one of the best labs and most fun places to do research in France.

Love the bike

Well deserved rest!

Thank you @javirandor.com for your nice talk about your latest research on “Gradient-based Jailbreak for Multimodal Fusion Models” during our Qualcomm Academic Lecture Series. It was great to have you!

Sadly, such a behavior is not (enough?) punished. I wish there was a way to reset the citations count of all authors involved in the plagiarism (day dreaming) :).

2 things:

First one is getting used to not do (de)activate to switch between venv.

Second, there is a small learning curve to grasp the full capabilities of the venv setup file. For instance, you can specify dependencies for each system requirements (e.g., macOS, Linux, cudq version) in 1 file.

Here is an image illustrating how fast it is. @berton-gabri.bsky.social : I know you like advertising tips for python development. Have you tried this one?

Lately, I have been experimenting with pixi instead of conda/mamba for managing python packages. One great feature is pixi’s ability to prevent conflicting dependencies between pypi and conda.

Great project + it is blazing fast (Rust based) 🔥!!!

Congratulations!

The best paper awardee from NeuRIPS 2024 has been apparently accused of misconduct by his ByteDance peers. This raises many questions certainly:

var-integrity-report.github.io

Congratulations!

#CVPR2025 Area Chairs: When entering recommendation scores for reviewers, remember that 10 represents the **strongest** preference and 1 represents the **weakest**!

I agree. Cisco courses and packet tracer helped me a lot back in the days :). The ability to experiment network configs on packet tracer without going to the lab was nice.

Wow - it is a positive change compared to 15 years ago (feeling old). Is it a public high school?

I feel a lot of researchers from Cybersecurity are very active in Twitter and only a few of them moved to🦋.

I always feel acknowledging a reviewer’s concern can be a risky gamble. For instance, if the reviewer is dismissive then I feel it gives him an extra boost to shutdown the paper. If you had counter-examples to share that would be great :)