Skander Moalla

model compression (1) structured matrix (1) representation learning (1) multi-agent reinforcement learning (1) reinforcement learning (1) neural network training (1) procedural generation (1) partial observability (1) cooperative learning (1) proximal policy optimization (1) cooperative multi-agent (1) representation collapse (1) trust region (1) loss of plasticity (1) capacity loss (1) feedforward network (1) parameter efficiency (1) large language model (1) training regime (1) feature rank (1)