9 papers
Cache Locally, Recompute Selectively for Streaming Video Understanding
Junhyeok Lee, Sangyoon Lee, Jaeho Lee
Streaming video understanding (SVU) requires video-LLMs to process incoming video continuously while answering user queries with low latency. Recent methods reduce query-time cost…
Motif-Video 2B: Technical Report
Junghwan Lim, Wai Ting Cheung, Minsu Ha +25
Training strong video generation models usually requires massive datasets, large parameter counts, and substantial compute. In this work, we ask whether strong text-to-video qualit…
Limited Stereotype Control Through Routing Reweighting in MoE Language Models
Junhyeok Lee, Han Jang, Kyu Sung Choi
Demographic prompts are routed differently from neutral prompts in Mixture-of-Experts (MoE) language models, motivating tests of routing-level stereotype control. We introduce Fair…
citecheck: An MCP Server for Automated Bibliographic Verification and Repair in Scholarly Manuscripts
Junhyeok Lee
Reference lists in scholarly manuscripts frequently contain errors, including incorrect identifiers, incomplete metadata, misattributed authors, and mismatches between preprint and…
Motif-2-12.7B-Reasoning: A Practitioner's Guide to RL Training Recipes
Junghwan Lim, Sungmin Lee, Dongseok Kim +23
We introduce Motif-2-12.7B-Reasoning, a 12.7B parameter language model designed to bridge the gap between open-weight systems and proprietary frontier models in complex reasoning a…
Motif 2 12.7B technical report
Junghwan Lim, Sungmin Lee, Dongseok Kim +22
We introduce Motif-2-12.7B, a new open-weight foundation model that pushes the efficiency frontier of large language models by combining architectural innovation with system-level…
Grouped Differential Attention
Junghwan Lim, Sungmin Lee, Dongseok Kim +7
The self-attention mechanism, while foundational to modern Transformer architectures, suffers from a critical inefficiency: it frequently allocates substantial attention to redunda…
Expanding Foundational Language Capabilities in Open-Source LLMs through a Korean Case Study
Junghwan Lim, Gangwon Jo, Sungmin Lee +16
We introduce Llama-3-Motif, a language model consisting of 102 billion parameters, specifically designed to enhance Korean capabilities while retaining strong performance in Englis…
Motif 2.6B Technical Report
Junghwan Lim, Sungmin Lee, Dongseok Kim +22
Recent advancements in Large Language Models (LLMs) have revolutionized artificial intelligence, yet developing an effective foundational LLM that balances high performance with co…