8 papers
MJev-Spatial: System One Spatial Intelligence via Native Attention Matching
Luozheng Qin, Yuhan Zhu, Changhai Zhou +10
Spatial intelligence lets an agent perceive and act in the physical world, and many of its judgments must be made at interaction rates: which spatial relation holds, how far an obj…
LongStraw: Long-Context RL Beyond 2M Tokens under a Fixed GPU Budget
Changhai Zhou, Kieran Liu, Yuhua Zhou +17
Long-context RL post-training is constrained by the lifetime of state and gradients, not attention cost alone. In GRPO, one multi-million-token prompt must serve old-policy and ref…
ProtTeX-CC: Activating In-Context Learning in Protein LLM via Two-Stage Instruction Compression
Chuanliu Fan, Zicheng Ma, Jun Gao +5
Recent advances in protein large language models, such as ProtTeX, represent both side-chain amino acids and backbone structure as discrete token sequences of residue length. While…
The Bidirectional Process Reward Model
Lingyin Zhang, Jun Gao, Xiaoxue Ren +1
Process Reward Models (PRMs), which assign fine-grained scores to intermediate reasoning steps within a solution trajectory, have emerged as a promising approach to enhance the rea…
Interleaved-Modal Chain-of-Thought
Jun Gao, Yongqi Li, Ziqiang Cao +1
Chain-of-Thought (CoT) prompting elicits large language models (LLMs) to produce a series of intermediate reasoning steps before arriving at the final answer. However, when transit…
DNTextSpotter: Arbitrary-Shaped Scene Text Spotting via Improved Denoising Training
Yu Xie, Qian Qiao, Jun Gao +5
More and more end-to-end text spotting methods based on Transformer architecture have demonstrated superior performance. These methods utilize a bipartite graph matching algorithm…
AIM: Let Any Multi-modal Large Language Models Embrace Efficient In-Context Learning
Jun Gao, Qian Qiao, Ziqiang Cao +2
In-context learning (ICL) facilitates Large Language Models (LLMs) exhibiting emergent ability on downstream tasks without updating billions of parameters. However, in the area of…
Guiding ChatGPT to Generate Salient Domain Summaries
Jun Gao, Ziqiang Cao, Shaoyao Huang +2
ChatGPT is instruct-tuned to generate general and human-expected content to align with human preference through Reinforcement Learning from Human Feedback (RLHF), meanwhile resulti…