Search papers, labs, and topics across Lattice.
The paper introduces ThinkRouter, a confidence-aware routing mechanism that dynamically switches between latent and discrete token spaces during inference to improve reasoning efficiency and accuracy. ThinkRouter routes reasoning to the discrete token space when model confidence is low, and to the latent space otherwise, aiming to mitigate noise propagation and avoid overconfident, incorrect reasoning trajectories. Experiments on STEM reasoning and coding benchmarks show that ThinkRouter significantly outperforms baselines like CoT and latent reasoning, achieving an average improvement of 19.70 points in Pass@1 and reducing generation length by up to 15.55%.
Achieve near 20-point accuracy gains in reasoning tasks by dynamically routing between latent and discrete reasoning spaces based on model confidence.
Recent work explores latent reasoning to improve reasoning efficiency by replacing explicit reasoning trajectories with continuous representations in a latent space, yet its effectiveness varies across settings. Analysis of model confidence dynamics under latent reasoning reveals that thinking trajectories ending in incorrect answers contain fewer low-confidence steps than those ending in correct answers. Meanwhile, we suggest that soft embeddings aggregated by multiple low-confidence thinking alternatives may introduce and propagate noise, leading to high confidence in unreliable reasoning trajectories. Motivated by these observations, ThinkRouter, an inference-time confidence-aware routing mechanism is proposed to avoid high confidence and noise for efficient reasoning. ThinkRouter routes thinking to the discrete token space when model confidence is low, and to the latent space otherwise. Extensive experiments on STEM reasoning and coding benchmarks across diverse large reasoning models demonstrate that ThinkRouter outperforms explicit CoT, random routing, and latent reasoning baselines in terms of accuracy, achieving an average improvement of 19.70 points in Pass@1, while reducing generation length by up to 15.55%. Further comprehensive analysis reveals that ThinkRouter can calibrate errors arising from explicit CoT and latent reasoning, and accelerates end-of-thinking token generation by globally lowering model confidence.
Amazon Science