✳flâneur — a map of the web's best reading
The Sensory Neuron as a Transformer: Permutation-Invariant Neural Networks for Reinforcement Learning
attentionneuron.github.io · 10,674 words · saved by 1 readers
Permutation-Invariant Neural Networks for Reinforcement Learning
--> This page requires Javascript. Please enable it for https://attentionagent.github.io/ Examples of permutation-invariant reinforcement learning agents In this work, we investigate the properties of RL agents that treat their observations as an arbitrarily ordered, variable-length list of sensory inputs. Here, we partition the visual input from CarRacing (Left) and Atari Pong (right) into a 2D grid of small patches, and shuffled their ordering. Each sensory neuron in the system receives a stream of visual input at a particular permuted patch location, and through coordination, must complete
Explore this link on the map →related reading
- Aman's AI Journal • Primers • Ilya Sutskever's Top 30aman.ai
- Set Transformer: A Framework for Attention-basedPermutation-Invariant Neural Networksarxiv.org
- The 37 Implementation Details of Proximal Policy Optimization · The ICLR Blog Trackiclr-blog-track.github.io
- Learning Beyond Gradientstrinkle23897.github.io
- When Models Manipulate Manifolds: The Geometry of a Counting Tasktransformer-circuits.pub
- Reinforcement learning with imperceptible rewards — AI Alignment Forumalignmentforum.org
- Ch. 21 - Imitation Learningunderactuated.mit.edu
- Transformers from scratch | peterbloem.nlpeterbloem.nl
- e5b5c402bb7bd5e60bede6961d6fe39e-Paper-Conference.pdfproceedings.iclr.cc
- Verifying your browser | OpenReviewopenreview.net
- Deep Reinforcement Learning: Pong from Pixelskarpathy.github.io
- Tensor-Transformer Variants are Surprisingly Performant — LessWronglesswrong.com