ElephantBench is a benchmark of 1,094 questions testing whether LLMs can recall multiple divergent accounts of long-tail facts, built via…
EASEL is a benchmark evaluating multimodal agents’ ability to perform dexterous visual tool use through reference-guided visual reconstruction tasks, accompanied…
StepGuard is a guard model that audits LLM-agent trajectories and monitors tool actions before execution, paired with StepGen, an automatic…
Ring Forcing addresses long-term memory limitations in autoregressive video diffusion using a ring-structured training strategy combined with compression and timestep…
GeoNeXt repurposes pretrained video generative models for geometry-estimation tasks such as depth and surface-normal prediction by formulating geometry estimation as…
PonderPounce is a robot-control system using a pretrained multimodal LLM as episodic memory: “Ponder,” a reasoning-focused MLLM, accumulates observations and…
Rubric-to-Code Credit Assignment (RCCA) is a reinforcement-learning framework that converts functional feedback from rubrics into localized optimization signals for generated…
Cross-lingual Ranking Preference Optimization (CRPO) transfers preference knowledge from English to improve LLM alignment in other languages, using a hierarchical…
“Acquire, Repair, Preserve” is a three-stage post-training recipe for small language models playing interactive dialogue games: supervised fine-tuning for broad…
GGSS (Geodesic-Gated Spherical Steering) is an inference-time debiasing method for generative vision-language models that discovers a counterfactual bias subspace on…
Reward-Optimized Probe-and-Respond (RO-PnR) is a framework that decides when conversational agents should ask clarifying questions versus give direct corrections during…
This paper introduces a Generation-Fact Graph (GFG) framework unifying training, learning, and inference dynamics in neural systems, developing a predictive…
