Giving the task Designer memory and corpus conditioning yields far more diverse auto-generated environments
natashajaques · x · 2026-08-20
Natasha Jaques presents her team's work on automated curriculum learning, noting these approaches are often bottlenecked by the Designer proposing insufficiently diverse tasks. They overcome this by giving the Designer memory and conditioning it on a sample from a large pretraining corpus each episode. The generated environments are diverse and interesting, and grow increasingly complex throughout training.
Related event: Boosting Agent Training: Memory for Task Designers Yields 10% Gain(2 posts)→
More from Research
- Transformer trained on flying bat recordings yields hippocampus 'digital twin' revealing hidden structure — patrickmineault · 2026-10-16
- One year of Recursive Language Models: GPT-5-mini beats GPT-5 by 110% on 132k-token tasks, cheaper too — a1zhang · 2026-10-16
- MIT PhD Sharut Gupta joins DeepMind; Isola hails her representational alignment defense talk — phillip_isola · 2026-10-16
- 36 primitives, just 10 parameters each: a remarkably precise T-pose body fit — letandrewcook · 2026-10-16
- Interp researcher pushes back: if you can't simulate a neuron, you haven't explained it — nabla_theta · 2026-10-16
- MLP-Splatting (ECCV 2026) code released: MLPs as light-field primitives instead of Gaussians — AjdDavison · 2026-10-16