Welcome to be-astute.in — thoughts on technology, entrepreneurship, and building things that matter.
Latest
Adaptive Token Pruning in Mixture-of-Experts [Paper]
A mechanism that dynamically removes non-essential tokens during the forward pass, reducing the computational burden on the MoE layers.
Read
Direct Alignment via Contrastive Preference Optimization [Paper]
A direct alignment method that eliminates the need for a reward model and RL-based optimization.
Read
Neuromorphic State Space Models for Edge Computing [Paper]
A Neuromorphic SSM that integrates the efficiency of spiking neural networks with the structural advantages of SSMs.
Read
Streaming Communication in Multi-Agent Reasoning [Paper]
One-pager summary for Streaming Communication in Multi-Agent Reasoning
Read
Streaming Communication in Multi-Agent Reasoning [Paper]
One-pager summary for Streaming Communication in Multi-Agent Reasoning
Read