Research2026
Latent Thinking Optimization
Code for supervising and improving latent reasoning by treating hidden-state correctness signals as a latent reward model.
PythonLatent ReasoningReward Modeling
Projects
Reproducible benchmarks, research systems, and open-source implementations connected to our publications.
14 projects
Research2026
LLMPopcorn
An LLM-assisted pipeline for generating and evaluating titles, cover prompts, and short-video prompts designed for audience appeal.
PythonRAGVideo Generation
Open source2026
MetaSR
The official implementation of a meta-learning framework that bridges next-item prediction and masked-language modeling for recommendation.
PythonMeta-LearningSequential Recommendation
Benchmarks2026
MMPCBench
A benchmark for measuring how multimodal language models reconstruct missing product text or imagery and support recommendation.
PythonMultimodal LLMsRecommendation
Benchmarks2026
PhysicsMind
A simulation-and-real-world benchmark for testing physical reasoning and prediction in vision-language and world models.
PythonVLM EvaluationVideo Generation
Benchmarks2026
SkillAudit
Skill-centered assessment for agent skills across utility, efficiency and cost, and safety, backed by sandboxed execution evidence.
PythonDockerBrowser Extension
Open source2025
AMBER
An adaptive meta-balancing framework for integrating heterogeneous graph signals in knowledge tracing.
PythonGraph LearningKnowledge Tracing
Open source2025
FDRec
Frequency-decoupled knowledge distillation for reducing multimodal recommendation cost while preserving useful cross-modal signals.
PythonKnowledge DistillationRecommendation
Open source2025
Guider
The official implementation of a guided-calibration framework for denoising multimodal recommender systems.
PythonPyTorchMultimodal Recommendation
Open source2025
IISAN-Versa
A decoupled parameter-efficient adaptation framework for symmetric and asymmetric multimodal foundation models in recommendation.
PythonPEFTSequential Recommendation
Research2025
SAPIENT
A conversational recommendation system that combines a learned agent with Monte Carlo tree search for multi-turn planning.
PythonMCTSConversational AI
Research2025
SOLAR
An open implementation for aligning language-model recommenders with both relevance and serendipity objectives.
PythonLLM AlignmentRecommendation
Open source2025
TARec
A teacher-assisted Wasserstein distillation pipeline for compressing multimodal recommenders into efficient ID-based student models.
PythonPyTorchKnowledge Distillation
Open source2025
UKT
Uncertainty-aware knowledge tracing that represents student states as distributions instead of fixed deterministic embeddings.
PythonKnowledge TracingUncertainty Modeling