Summary of Two Are Better Than One: Context Window Extension with Multi-grained Self-injection, by Wei Han et al.
Two are better than one: Context window extension with multi-grained self-injectionby Wei Han, Pan Zhou,…
Two are better than one: Context window extension with multi-grained self-injectionby Wei Han, Pan Zhou,…
Bridging the Training-Inference Gap in LLMs by Leveraging Self-Generated Tokensby Zhepeng Cen, Yao Liu, Siliang…
What Do LLMs Need to Understand Graphs: A Survey of Parametric Representation of Graphsby Dongqi…
Enhancing JEPAs with Spatial Conditioning: Robust and Efficient Representation Learningby Etai Littwin, Vimal Thilak, Anand…
Interpolating Video-LLMs: Toward Longer-sequence LMMs in a Training-free Mannerby Yuzhang Shang, Bingxin Xu, Weitai Kang,…
ChatQA 2: Bridging the Gap to Proprietary LLMs in Long Context and RAG Capabilitiesby Peng…
Towards a theory of how the structure of language is acquired by deep neural networksby…
Exploring Context Window of Large Language Models via Decomposed Positional Vectorsby Zican Dong, Junyi Li,…
LongEmbed: Extending Embedding Models for Long Context Retrievalby Dawei Zhu, Liang Wang, Nan Yang, Yifan…
LLoCO: Learning Long Contexts Offlineby Sijun Tan, Xiuyu Li, Shishir Patil, Ziyang Wu, Tianjun Zhang,…