1 citations · 1 across the 10 of their papers we have counts for
13 papers
Does On-Policy Distillation Need to Complete Every Rollout?
Zhengxin Zhang, Maosen Tang, Ning Wang +6
On-policy distillation (OPD) transfers teacher knowledge to a student through feedback on student-generated rollouts, but generating complete rollouts and computing per-token teach…
Matryoshka Agent: Unfolding Sub-Agents for Long-Horizon Machine Learning Engineering
Rushi Qiang, Changhao Li, Haotian Sun +3
Machine learning engineering (MLE) tasks require long-horizon decision making over iterative solution debugging and refinement, under expensive and feedback-driven environment inte…
IMUG-Bench: Benchmarking Unified Multimodal Models on Interleaved Understanding and Generation
Lingyi Meng, Zecong Tang, Haoran Li +12
In recent years, unified multimodal models (UMMs) have emerged to support both understanding and generation within a single framework. Mastering dynamic, multi-turn interleaved ima…
Forward-Free Diffusion Language Models with BPTT-Free Looped Refinement
Haotian Sun, Rushi Qiang, Yuqian Zheng +1
Diffusion language models generate text through iterative denoising, offering a powerful alternative to autoregressive generation. However, discrete language spaces lack a natural…
Towards Better Instruction Following Retrieval Models
Yuchen Zhuang, Aaron Trinh, Rushi Qiang +4
Modern information retrieval (IR) models, trained exclusively on standard <query, passage> pairs, struggle to effectively interpret and follow explicit user instructions. We introd…
MLE-Dojo: Interactive Environments for Empowering LLM Agents in Machine Learning Engineering
Rushi Qiang, Yuchen Zhuang, Yinghao Li +8
We introduce MLE-Dojo, a Gym-style framework for systematically reinforcement learning, evaluating, and improving autonomous large language model (LLM) agents in iterative machine…
AmorLIP: Efficient Language-Image Pretraining via Amortization
Haotian Sun, Yitong Li, Yuchen Zhuang +3
Contrastive Language-Image Pretraining (CLIP) has demonstrated strong zero-shot performance across diverse downstream text-image tasks. Existing CLIP methods typically optimize a c…
Matryoshka Pilot: Learning to Drive Black-Box LLMs with LLMs
Changhao Li, Yuchen Zhuang, Rushi Qiang +4
Despite the impressive generative abilities of black-box large language models (LLMs), their inherent opacity hinders further advancements in capabilities such as reasoning, planni…
EC-DIT: Scaling Diffusion Transformers with Adaptive Expert-Choice Routing
Haotian Sun, Tao Lei, Bowen Zhang +5
Diffusion transformers have been widely adopted for text-to-image synthesis. While scaling these models up to billions of parameters shows promise, the effectiveness of scaling bey…
Spectral Representation for Causal Estimation with Hidden Confounders
Haotian Sun, Antoine Moulin, Tongzheng Ren +2
We address the problem of causal effect estimation where hidden confounders are present, with a focus on two settings: instrumental variable regression with additional observed con…
HYDRA: Model Factorization Framework for Black-Box LLM Personalization
Yuchen Zhuang, Haotian Sun, Yue Yu +4
Personalization has emerged as a critical research area in modern intelligent systems, focusing on mining users' behavioral history and adapting to their preferences for delivering…
BBox-Adapter: Lightweight Adapting for Black-Box Large Language Models
Haotian Sun, Yuchen Zhuang, Wei Wei +2
Adapting state-of-the-art Large Language Models (LLMs) like GPT-4 and Gemini for specific tasks is challenging. Due to the opacity in their parameters, embeddings, and even output…
On Task-personalized Multimodal Few-shot Learning for Visually-rich Document Entity Retrieval
Jiayi Chen, Hanjun Dai, Bo Dai +2
Visually-rich document entity retrieval (VDER), which extracts key information (e.g. date, address) from document images like invoices and receipts, has become an important topic i…