state-of-the-art benchmarks
- GIST: Greedy Independent Set Thresholding for Max-Min Diversification with Submodular Utility
- MESS+: Dynamically Learned Inference-Time LLM Routing in Model Zoos with Service Level Guarantees
- Model-Informed Flows for Bayesian Inference
- NEP: Autoregressive Image Editing via Next Editing Token Prediction
- VITA-1.5: Towards GPT-4o Level Real-Time Vision and Speech Interaction