RL & Agents Reading Group @ University of Edinburgh We discuss recent papers in RL, MARL & related Sign up & join: https://teapawt.net/t.co/i14FJcTX4D
Online
Joined July 2025
- Tweets28
- Following15
- Followers100
- Likes8
Pinned Tweet
Hello world! This is the RL & Agents Reading Group
We organise regular meetings to discuss recent papers in Reinforcement Learning (RL), Multi-Agent RL and related areas (open-ended learning, LLM agents, robotics, etc).
Meetings take place online and are open to everyone 😊
The RL & Agents Reading Group continues today at 3 pm UK time with the final talk in our RLC series!
@Theo_Vincent_ will present "A Path Towards Autonomous Learning Agents", exploring adaptive RL and temporal-difference learning.
Join us!
We are back with the RL & Agents Reading Group tomorrow (Thursday) at 4pm UK!
@mcbeukman is presenting his work on preventing PPO stagnation by scaling to 1M+ parallel environments.
Interested in PPO, scaling or open-ended RL? Join us!
arxiv.org/abs/2603.06009
We are back with the RL & Agents Reading Group tomorrow (Thursday) 3pm, kicking off a three-talk RLC series!
@LHinckeldey is presenting Assistax, a benchmark for multi-agent assistive robotics. Interested in MARL or ad hoc teamwork? Join us!
Paper: arxiv.org/abs/2507.21638
We are back with another RL & Agents Reading Group this Thursday!
@SamuelGarcin is presenting PERSIST; a new world model architecture with a persistent 3D state. If you are interested in world models this is a talk you won't want to miss.
Open to everyone. Details in reply⬇️
Meeting Link: shorturl.at/47vjF
📅Thursday July 23rd
🕒 15:00 UK
Project Page: francelico.github.io/persist…
Paper: arxiv.org/abs/2603.03482
Code: github.com/francelico/PERSIS…
RL & Agents Reading Group retweeted
We’re open-sourcing the models, weights and training code for PERSIST, our 1.7B parameter 3D world model!
By modelling a dynamic 3D world state instead of relying on pixel histories, PERSIST generates experiences that remain spatially and temporally coherent over thousands of steps.
Tom, @kaixin20578389 and I will present PERSIST at @icmlconf in Seoul next month. See you there!
Links in thread⬇️
#WorldModels #ICML2026
RL & Agents Reading Group retweeted
As we build more cooperative multi-agent envs, we should ask: are we really testing the properties that make Dec-POMDPs hard, partial observability and decentralised coordination, or can agents succeed via shortcuts? 🤔
Our #AAMAS2026 Oral probes this question 🔎🧵
RL & Agents Reading Group Tomorrow 3pm GMT (UK) time 📖
@steleac will be joining us to present his ICLR paper on a new approach to discovering coordinated behaviours in multi-agent systems 🤖
Paper: arxiv.org/abs/2512.24827
RL & Agents Reading Group retweeted
PERSIST is a world model that ditches pixel-based histories for a 3D world state. Instead of searching through an ever-growing sequence of past pixel observations, PERSIST retrieves spatial information from a dynamically evolving 3D representation.
This change improves the spatial memory, 3D consistency, and long-horizon stability of the model, enabling interactive experiences within coherent and evolving 3D worlds.
#MachineLearning #WorldModels #GenerativeAI #3DComputerVision #ComputerVision #Genie3 #AI
RL & Agents Reading Group TODAY 3pm GMT (UK) time 📖
Lukas Schäfer (Microsoft Research) will be joining us to present predictive inverse dynamics models (PIDMs) as an alternative to behaviour cloning in offline imitation learning💡
Paper: arxiv.org/abs/2601.21718
Reading group TODAY at 3pm UK time (GMT)!
Ruaridh Mon-Williams will be joining us to present some interesting recent work on emergent partner modelling
Paper: arxiv.org/abs/2505.17323
RL & Agents Reading Group retweeted
Running a short human eval study on interactive world models — would really appreciate any help!
You’ll score short videos based on spatial & temporal consistency. It’s lightweight (and maybe even fun 🙂).
Full study ~25 min, but partial responses are totally fine. Works on mobile too!
Link in thread ⬇️
Reading group TOMORROW 3-4pm UK!
Joe Marino @jl_marino (Google DeepMind) will present SIMA 2, a generalist embodied agent designed to operate across a wide range of 3D virtual worlds 🌎
Join here: edinburgh-rl.github.io/readi…
Reading group TODAY at 2pm BST!
"Despite years of research in offline reinforcement learning (RL), the field has failed to deliver major breakthroughs..."
@JacksonMattT and @JarekLiesen (Oxford) will present Unifloral - unified implementations and evaluations for offline RL.
Next reading group session today at 15:00 GMT! @trevormcinroe (University of Edinburgh) will be presenting Terra Nova, a new comprehensive challenge environment for RL research inspired by Civilization V.
Reading group today at 12:00 BST!
We are continuing our NeurIPS series. Felix Chalumeau (@ChalumeauFelix) from InstaDeep will present his oral paper on how inference strategies can boost performance in MARL. 🚀
Papers:
arxiv.org/abs/2505.21236
Meeting:
bit.ly/488sD6n
Reading group today at 2pm BST!
We are starting our NeurIPS series with Sable and Oryx, sequence models for scalable multi-agent coordination from the RL Research Team at @instadeepai . 🚀
Papers:
bit.ly/3Lme7jH
bit.ly/47GJb4T
Meeting:
bit.ly/3JoEbtU
RL reading group in less than 2 hours at 15:00 BST 🔥🕺
Speaker: Sebastian Griesbach
Title: Learning to Explore in Diverse Reward Settings via Temporal-Difference-Error Maximization 🗺️ (RLJ)
Paper: rlj.cs.umass.edu/2025/papers…
Meeting link: teams.microsoft.com/l/meetup…