TeLAPA preserves behaviorally diverse policy neighborhoods in a shared latent space, improving MiniGrid continual RL transfer, revisit recovery, and retention over single-model preservation.
Clare Lyle, Mark Rowland, and Will Dabney
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
citation-role summary
background 1
method 1
citation-polarity summary
fields
cs.LG 7representative citing papers
Dynamical isometry (Jacobian singular values near 1) preserves plasticity in continual learning; an isometry-promoting regularizer and decoupled AdamO optimizer match or beat prior methods on supervised and RL benchmarks.
RPSFT improves the in-domain versus out-of-domain performance trade-off during LLM supervised fine-tuning by penalizing rotations in pretrained singular subspaces as a proxy for loss-sensitive directions.
This paper defines loss of plasticity via stable manifolds in parameter space and identifies frozen units and cloned-unit manifolds as the main mechanisms that trap gradient trajectories in non-stationary settings.
SFT depth increases pre-RL pass@1 but can cause entropy collapse that inverts GRPO outcomes on Qwen models via reduced group advantage variance.
Growth during training inserts new units into a specialized trajectory, making them forward-active but backward-starved with weaker gradients than existing units.
Smooth-Leaky and Randomized Smooth-Leaky activations mitigate loss of plasticity in continual learning by targeting negative-branch shape and saturation behavior.
citing papers explorer
-
Beyond Single-Model Optimization: Preserving Plasticity in Continual Reinforcement Learning
TeLAPA preserves behaviorally diverse policy neighborhoods in a shared latent space, improving MiniGrid continual RL transfer, revisit recovery, and retention over single-model preservation.
-
Preserving Plasticity in Continual Learning via Dynamical Isometry
Dynamical isometry (Jacobian singular values near 1) preserves plasticity in continual learning; an isometry-promoting regularizer and decoupled AdamO optimizer match or beat prior methods on supervised and RL benchmarks.
-
Rotation-Preserving Supervised Fine-Tuning
RPSFT improves the in-domain versus out-of-domain performance trade-off during LLM supervised fine-tuning by penalizing rotations in pretrained singular subspaces as a proxy for loss-sensitive directions.
-
Barriers for Learning in an Evolving World: Mathematical Understanding of Loss of Plasticity
This paper defines loss of plasticity via stable manifolds in parameter space and identifies frozen units and cloned-unit manifolds as the main mechanisms that trap gradient trajectories in non-stationary settings.
-
SFT Overtraining Predicts Rank Inversion via Entropy Collapse Under RLVR
SFT depth increases pre-RL pass@1 but can cause entropy collapse that inverts GRPO outcomes on Qwen models via reduced group advantage variance.
-
On the Stability of Growth in Structural Plasticity
Growth during training inserts new units into a specialized trajectory, making them forward-active but backward-starved with weaker gradients than existing units.
-
Activation Function Design Sustains Plasticity in Continual Learning
Smooth-Leaky and Randomized Smooth-Leaky activations mitigate loss of plasticity in continual learning by targeting negative-branch shape and saturation behavior.