Jingyu Liu

Switching between autoregressive and diffusion modes allows Nemotron-Labs-Diffusion to achieve unprecedented throughput and efficiency in language modeling.

Evaluator quality for robotic policies hinges more on long-horizon consistency than on short-term visual fidelity, reshaping our approach to world model design.

Task insensitivity in language models leads to significant failures in OOD generalization, but a novel optimization technique can reverse this trend.

CoT training may not boost reasoning capabilities as expected, instead reinforcing reliance on prompts, which could reshape our approach to training language-model agents.

AMM price prediction accuracy jumps 56% by explicitly modeling the uncertainty in block intervals, revealing the critical role of on-chain event timing.

You can train a surprisingly effective medical imaging MLLM using *only* public data by generating knowledge-aware reasoning traces and enforcing self-consistency during reinforcement learning.