8 papers
Self-OPD: On-Policy Distillation for Flow Matching Models without Teacher
Shiyi Zhang, Mushui Liu, Yunze Tong +8
On-policy distillation (OPD), which leverages a pre-trained, specialized teacher model to provide dense supervisory signals, has achieved significant success in Large Language Mode…
InfinityEdit: Infinite Video Editing with a Lightweight Edit-Ignition Adapter
Yunze Tong, Mushui Liu, Canyu Zhao +9
With large pretrained models, existing methods have effectively improved instruction-based video editing. However, most of them rely on an in-place editing assumption. They align t…
Kimi K3: Open Frontier Intelligence
Kimi Team, Tongtong Bai, Yifan Bai +398
We introduce Kimi K3, a 2.8T parameter Mixture-of-Experts model with 104 billion activated parameters, native vision capabilities, and a 1-million-token context window. Kimi K3 is…
ChatUMM: Robust Context Tracking for Conversational Interleaved Generation
Wenxun Dai, Zhiyuan Zhao, Yule Zhong +12
Unified multimodal models (UMMs) have achieved remarkable progress yet remain constrained by a single-turn interaction paradigm, effectively functioning as solvers for independent…
Alleviating Sparse Rewards by Modeling Step-Wise and Long-Term Sampling Effects in Flow-Based GRPO
Yunze Tong, Mushui Liu, Canyu Zhao +7
Deploying GRPO on Flow Matching models has proven effective for text-to-image generation. However, existing paradigms typically propagate an outcome-based reward to all preceding d…
Re-Align: Structured Reasoning-guided Alignment for In-Context Image Generation and Editing
Runze He, Yiji Cheng, Tiankai Hang +11
In-context image generation and editing (ICGE) enables users to specify visual concepts through interleaved image-text prompts, demanding precise understanding and faithful executi…
TAG-MoE: Task-Aware Gating for Unified Generative Mixture-of-Experts
Yu Xu, Hongbin Yan, Juan Cao +11
Unified image generation and editing models suffer from severe task interference in dense diffusion transformers architectures, where a shared parameter space must compromise betwe…
JarvisEvo: Towards a Self-Evolving Photo Editing Agent with Synergistic Editor-Evaluator Optimization
Yunlong Lin, Linqing Wang, Kunjie Lin +11
Agent-based editing models have substantially advanced interactive experiences, processing quality, and creative flexibility. However, two critical challenges persist: (1) instruct…