I'm excited to share that our paper "Bounded and Unbounded Verification of RNN-Based Agents in Non-deterministic Environments" has been accepted in @aamas2023! More details on this soon! #AAMAS

Very excited to announce that our paper Posterior Sampling for Deep Reinforcement Learning (PSDRL) has been accepted at #ICML2023! ⌨️Code: github.com/remosasso/PSDRL 📄 Paper: arxiv.org/abs/2305.00477 🧵Thread 👇

If you think ***in an ideal world where you could wave a magic wand to make it happen*** AI development should not be paused, what odds do you assign to AI x-risk (aka "what is your P(doom)")?

DeepMind is dedicated to bringing Safe, Reliable & Trustworthy AI to the real world. Consider applying and joining our amazing team. boards.greenhouse.io/deepmind/jobs/… boards.greenhouse.io/deepmind/jobs/…

📢Are you interested in safe and trusted #AI and #MultiAgent systems?! We have a #PhD position available on the "Verification of Neuro-Symbolic Multi-Agent Systems" at King's College London as part of the UKRI CDT in Safe & Trusted AI @ King's & Imperial #CDT program.

I don't want to summarize too much, since I think everyone should read this latest piece by Yoshua Bengio (Turing award winner for Deep Learning, along with Geoff Hinton and Yann Lecun): yoshuabengio.org/2023/05/22/how… Suffice to say that this piece engages w/AI x-risk very seriously.

Your text reveals whether you are rich or poor! Paper/slides 👉icastro.info and kudos to Waleed Iqbal who did lots of the heavy lifting

yoshuabengio.org/2023/08/12/per…

🚨 REGISTRATION CLOSING TOMORROW, 12:00 NOON (BTS) 🟣 This is your last chance to register (free) for the iGGi Con 2023 - via the registration link on our website iggi2023.org

This figure nicely summarises our proposed Attention mechanisms! Thank you Thibault Castells for the illustration!

We trained a small language model using Efficient Attention we introduced recently, and the result is interesting! You can check it out on Hugging Face. It's been only trained on 15B tokens and achieves similar loss/training token to models of similar size huggingface.co/PeymanHosseini…

📢 I am excited to share that our paper (with Mehran Hosseini) "GeoPos: A Minimal Positional Encoding for Enhanced Fine-Grained Details in Image Synthesis Using Convolutional Neural Networks" has been accepted to be presented at WACV 2025 at Arizona🌵! Keep an eye out