Lune

Profiles2021-2026

Yiwei Li

Beijing Institute of Technology

25Papers
0Last author
19Unique co-authors
6First author
0Solo

First and last author come from byline position, which some venues assign alphabetically.

Recent papers25

  1. LLM-Powered Benchmark Factory: Reliable, Generic, and EfficientAuthor 3 of 10ACL 2026 · 8 citations
  2. Breaking the Self-Confirming Loop: Diagnosing and Mitigating Systemic Reward Bias in Self-Rewarding RLAuthor 4 of 11ICML 2026 · 1 citation
  3. Learning More from Less: Unlocking Internal Representations for Benchmark CompressionAuthor 6 of 12ICML 2026
  4. PatternKV: Flattening KV Representation Expands Quantization HeadroomAuthor 2 of 11ICML 2026
  5. Every Rollout Counts: Optimal Resource Allocation for Efficient Test-Time ScalingAuthor 2 of 10NeurIPS 2025 · 16 citations
  6. Beyond One-Size-Fits-All: Tailored Benchmarks for Efficient EvaluationAuthor 4 of 10ACL 2025 · 6 citations
  7. Silencer: From Discovery to Mitigation of Self-Bias in LLM-as-Benchmark-GeneratorAuthor 2 of 10NeurIPS 2025 · 4 citations
  8. From Sub-Ability Diagnosis to Human-Aligned Generation: Bridging the Gap for Text Length Control via MarkerGenAuthor 4 of 10ACL 2025 · 2 citations
  9. Speculative Decoding for Multi-Sample InferenceAuthor 1 of 11EMNLP Findings 2025 · 1 citation
  10. InsBank: Evolving Instruction Subset for Ongoing AlignmentAuthor 2 of 11EMNLP Findings 2025
  11. Mind the Quote: Enabling Quotation-Aware Dialogue in LLMs via Plug-and-Play ModulesAuthor 3 of 10NeurIPS 2025
  12. Revisiting Self-Consistency from Dynamic Distributional Alignment Perspective on Answer AggregationAuthor 1 of 11ACL Findings 2025
  13. UniCBE: An Uniformity-driven Comparing Based Evaluation Framework with Unified Multi-Objective OptimizationAuthor 3 of 10ICLR 2025
  14. Escape Sky-high Cost: Early-stopping Self-Consistency for Multi-step ReasoningAuthor 1 of 7ICLR 2024 · 101 citations
  15. Turning Dust into Gold: Distilling Complex Reasoning Capabilities from LLMs by Leveraging Negative DataAuthor 1 of 7AAAI 2024 · 31 citations
  16. Dynamic Stochastic Decoding Strategy for Open-Domain Dialogue GenerationAuthor 1 of 6ACL Findings 2024 · 2 citations
  17. BatchEval: Towards Human-like Text EvaluationAuthor 3 of 7ACL 2024
  18. Focused Large Language Models are Stable Many-Shot LearnersAuthor 3 of 10EMNLP 2024
  19. Integrate the Essence and Eliminate the Dross: Fine-Grained Self-Consistency for Free-Form Language GenerationAuthor 2 of 8ACL 2024
  20. Poor-Supervised Evaluation for SuperLLM via Mutual ConsistencyAuthor 3 of 8ACL Findings 2024
  21. Towards Diverse, Relevant and Coherent Open-Domain Dialogue Generation via Hybrid Latent VariablesAuthor 5 of 6AAAI 2023 · 8 citations
  22. Better Correlation and Robustness: A Distribution-Balanced Self-Supervised Learning Framework for Automatic Dialogue EvaluationAuthor 4 of 5NeurIPS 2023 · 5 citations
  23. Heterogeneous-Branch Collaborative Learning for Dialogue GenerationAuthor 1 of 4AAAI 2023 · 4 citations
  24. Modeling Complex Dialogue Mappings via Sentence Semantic Segmentation Guided Conditional Variational Auto-EncoderAuthor 3 of 7EMNLP Findings 2022 · 1 citation
  25. Generating Relevant and Coherent Dialogue Responses using Self-Separated Conditional Variational AutoEncodersAuthor 3 of 5ACL 2021
Dusk over the sea between two cliffs drawn in fine vertical lines