10 papers
AsyncLane: Decoupling Refinement from Advancement in Diffusion Language Model Decoding
Yingxuan Ren, Yuxuan Lou, Yong Liu +4
Block-wise semi-autoregressive decoding is the standard inference paradigm for diffusion large language models (DLMs), but it imposes a strict dependency between blocks: the next b…
On-Policy Adversarial Flow Distillation for Autoregressive Video Generation
Yang Luo, Shengju Qian, Xiaohang Tang +4
Autoregressive video generators are attractive for streaming, long-horizon, and interactive applications, but distilling strong black-box teachers into causal students remains diff…
CAMEL: Confidence-Gated Reflection for Reward Modeling
Zirui Zhu, Hailun Xu, Yang Luo +4
Reward models play a fundamental role in aligning large language models with human preferences. Existing methods predominantly follow two paradigms: scalar discriminative preferenc…
FOCUS: Efficient Keyframe Selection for Long Video Understanding
Zirui Zhu, Hailun Xu, Yang Luo +4
Multimodal large language models (MLLMs) represent images and video frames as visual tokens. Scaling from single images to hour-long videos, however, inflates the token budget far…
Unbiased Gradient Low-Rank Projection
Rui Pan, Yang Luo, Yuxing Liu +2
Memory-efficient optimization is critical for training increasingly large language models (LLMs). A popular strategy involves gradient low-rank projection, storing only the project…
POME: Post Optimization Model Edit via Muon-style Projection
Yong Liu, Di Fu, Yang Luo +4
We introduce Post-Optimization Model Edit (POME), a new algorithm that enhances the performance of fine-tuned large language models using only their pretrained and fine-tuned check…
MERIT: Maximum-normalized Element-wise Ratio for Language Model Large-batch Training
Yang Luo, Zangwei Zheng, Ziheng Qin +3
Large-batch training has become a cornerstone in accelerating the training of deep neural networks, yet it poses challenges in optimization and generalization. Existing optimizers…
How Does the Textual Information Affect the Retrieval of Multimodal In-Context Learning?
Yang Luo, Zangwei Zheng, Zirui Zhu +1
The increase in parameter size of multimodal large language models (MLLMs) introduces significant capabilities, particularly in-context learning, where MLLMs enhance task performan…
Helen: Optimizing CTR Prediction Models with Frequency-wise Hessian Eigenvalue Regularization
Zirui Zhu, Yong Liu, Zangwei Zheng +2
Click-Through Rate (CTR) prediction holds paramount significance in online advertising and recommendation scenarios. Despite the proliferation of recent CTR prediction models, the…
Sparse MeZO: Less Parameters for Better Performance in Zeroth-Order LLM Fine-Tuning
Yong Liu, Zirui Zhu, Chaoyu Gong +3
While fine-tuning large language models (LLMs) for specific tasks often yields impressive results, it comes at the cost of memory inefficiency due to back-propagation in gradient-b…