7 citations · 10 across the 14 of their papers we have counts for
17 papers
Preserving Speech-to-Text LLM Capabilities in Speech-to-Speech Generation
Yuxuan Hu, Heng Lu, Ruchao Fan +8
Strong speech-to-text (S2T) LLMs already provide robust speech perception and text reasoning, but adding speech-to-speech (S2S) output is challenging: fine-tuning the backbone can…
Latent Recurrent Transformer: Architecture Exploration, Training Strategies, and Scaling Behavior
Zeyi Huang, Xuehai He, LiLiang Ren +8
We study Latent Recurrent Transformer (LRT), a lightweight augmentation of autoregressive transformers that reuses a high-level source-layer hidden state from the previous token as…
ThetaEvolve: Test-time Learning on Open Problems
Yiping Wang, Shao-Rong Su, Zhiyuan Zeng +13
Recent advances in large language models (LLMs) have enabled breakthroughs in mathematical discovery, exemplified by AlphaEvolve, a closed-source system that evolves programs to im…
Decoder-Hybrid-Decoder Architecture for Efficient Reasoning with Long Generation
Liliang Ren, Congcong Chen, Haoran Xu +11
Recent advances in language modeling have demonstrated the effectiveness of State Space Models (SSMs) for efficient sequence modeling. While hybrid architectures such as Samba and…
Routing Mamba: Scaling State Space Models with Mixture-of-Experts Projection
Zheng Zhan, Liliang Ren, Shuohang Wang +5
Linear State Space Models (SSMs) offer remarkable performance gains in efficient sequence modeling, with constant inference-time computation and memory complexity. Recent advances,…
Soft Thinking: Unlocking the Reasoning Potential of LLMs in Continuous Concept Space
Zhen Zhang, Xuehai He, Weixiang Yan +5
Human cognition typically involves thinking through abstract, fluid concepts rather than strictly using discrete linguistic tokens. Current reasoning models, however, are constrain…
Phi-4-Mini-Reasoning: Exploring the Limits of Small Reasoning Language Models in Math
Haoran Xu, Baolin Peng, Hany Awadalla +11
Chain-of-Thought (CoT) significantly enhances formal reasoning capabilities in Large Language Models (LLMs) by training them to explicitly generate intermediate reasoning steps. Wh…
Reinforcement Learning for Reasoning in Large Language Models with One Training Example
Yiping Wang, Qing Yang, Zhiyuan Zeng +11
We show that reinforcement learning with verifiable reward using one training example (1-shot RLVR) is effective in incentivizing the math reasoning capabilities of large language…
Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs
Microsoft, :, Abdelrahman Abouelenin +73
We introduce Phi-4-Mini and Phi-4-Multimodal, compact yet highly capable language and multimodal models. Phi-4-Mini is a 3.8-billion-parameter language model trained on high-qualit…
Is Your World Simulator a Good Story Presenter? A Consecutive Events-Based Benchmark for Future Long Video Generation
Yiping Wang, Xuehai He, Kuan Wang +5
The current state-of-the-art video generative models can produce commercial-grade videos with highly realistic details. However, they still struggle to coherently present multiple…
Mojito: Motion Trajectory and Intensity Control for Video Generation
Xuehai He, Shuohang Wang, Jianwei Yang +7
Recent advancements in diffusion models have shown great promise in producing high-quality video content. However, efficiently training video diffusion models capable of integratin…
Temperature-Centric Investigation of Speculative Decoding with Knowledge Distillation
Siru Ouyang, Shuohang Wang, Minhao Jiang +4
Speculative decoding stands as a pivotal technique to expedite inference in autoregressive (large) language models. This method employs a smaller draft model to speculate a block o…
LoRC: Low-Rank Compression for LLMs KV Cache with a Progressive Compression Strategy
Rongzhi Zhang, Kuang Wang, Liyuan Liu +4
The Key-Value (KV) cache is a crucial component in serving transformer-based autoregressive large language models (LLMs), enabling faster inference by storing previously computed K…
GRIN: GRadient-INformed MoE
Liyuan Liu, Young Jin Kim, Shuohang Wang +14
Mixture-of-Experts (MoE) models scale more effectively than dense models due to sparse computation through expert routing, selectively activating only a small subset of expert modu…
Cost-Effective Proxy Reward Model Construction with On-Policy and Active Learning
Yifang Chen, Shuohang Wang, Ziyi Yang +6
Reinforcement learning with human feedback (RLHF), as a widely adopted approach in current large language model pipelines, is \textit{bottlenecked by the size of human preference d…
Self-Exploring Language Models: Active Preference Elicitation for Online Alignment
Shenao Zhang, Donghan Yu, Hiteshi Sharma +6
Preference optimization, particularly through Reinforcement Learning from Human Feedback (RLHF), has achieved significant success in aligning Large Language Models (LLMs) to adhere…
Multi-LoRA Composition for Image Generation
Ming Zhong, Yelong Shen, Shuohang Wang +6
Low-Rank Adaptation (LoRA) is extensively utilized in text-to-image models for the accurate rendition of specific elements like distinct characters or unique styles in generated im…