이재길Jae-Gil Lee

20268편

• CVPRT1

  Accelerating Diffusion via Hybrid Data-Pipeline Parallelism Based on Conditional Guidance Scheduling

  Jung, E., Kim, B., Kim, H., Cho, S., and Lee, J.

• ICMLT1

  Breaking the Reference Bottleneck via Learning to Rewrite Conversational Queries without Gold Reference Passages

  Kim, D., Lee, Y., Kim, J., Bang, J., Song, H., Yoon, S., and Lee, J.

• EMNLPT1

  Clustering-Based Balanced Sampling and Allocation with Data Parallelism for High-Performance Fine-Tuning

  Kim, H., Nam, Y., Han, J., Choi, W., and Lee, J.

• ICLRT1

  Completing Missing Annotation: Multi-Agent Debate for Accurate and Scalable Relevant Assessment for IR Benchmarks

  Ban, M., Choi, J., Min, H., Kim, N. H., Kim, M., Lee, J., and Song, H.

• CIKMT1

  Compositional Spectral Prompts for LLM-based Online Time Series Forecasting

  Choi, S., Kim, H., Lee, J., and Park, C.

• ECCVT1

  CURE: Cumulative Knowledge Reuse for Efficient Device-Server Hybrid Inference in Vision-Language Models

  Lee, Y., Kim, D., Kang, J., Song, H., and Lee, J.

• ACLT1

  QuDAR: Query-Wise Dual-Perspective Adaptive Retrieval

  Kim, J., Yoon, S., Le, X., Nam, Y., Kim, D., Song, H., and Lee, J.

• CVPRT1

  See and Fix the Flaws: Enabling VLMs and Diffusion Models to Comprehend Visual Artifacts via Agentic Data Synthesis

  Park, J., Ahn, M., Kim, M., Lee, J., Lee, J., and Park, D.