Profiles2021-2026
Linqi Song
City University of Hong Kong
20Papers
7Last author
37Unique co-authors
0First author
0Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers20
- Scaling Reasoning Hop Exposes Weaknesses: Demystifying and Improving Hop Generalization in Large Language ModelsAuthor 4 of 6ICLR 2026 · 1 citation
- Activation-Guided Consensus Merging for Large Language ModelsAuthor 9 of 9NeurIPS 2025 · 14 citations
- VOVTrack: Exploring the Potentiality in Raw Videos for Open-Vocabulary Multi-Object TrackingAuthor 4 of 5ICCV 2025 · 3 citations
- Sens-Merging: Sensitivity-Guided Parameter Balancing for Merging Large Language ModelsAuthor 6 of 6ACL Findings 2025 · 1 citation
- Determine-Then-Ensemble: Necessity of Top-k Union for Large Language Model EnsemblingAuthor 8 of 8ICLR 2025
- Unlocking the Power of Function Vectors for Characterizing and Mitigating Catastrophic Forgetting in Continual Instruction TuningAuthor 6 of 8ICLR 2025
- Solving Challenging Math Word Problems Using GPT-4 Code Interpreter with Code-based Self-VerificationAuthor 9 of 11ICLR 2024 · 206 citations
- MUSTARD: Mastering Uniform Synthesis of Theorem and Proof DataAuthor 8 of 9ICLR 2024 · 50 citations
- Understanding and Patching Compositional Reasoning in LLMsAuthor 4 of 6ACL Findings 2024 · 4 citations
- Bi-Chainer: Automated Large Language Models Reasoning with Bidirectional ChainingAuthor 3 of 3ACL Findings 2024 · 1 citation
- Mitigating the Language Mismatch and Repetition Issues in LLM-based Machine Translation via Model EditingAuthor 5 of 6EMNLP 2024 · 1 citation
- Benchmarking and Improving Compositional Generalization of Multi-aspect Controllable Text GenerationAuthor 4 of 7ACL 2024
- LoRAN: Improved Low-Rank Adaptation by a Non-Linear TransformationAuthor 2 of 3EMNLP Findings 2024
- ProxyQA: An Alternative Framework for Evaluating Long-Form Text Generation with Large Language ModelsAuthor 10 of 11ACL 2024
- VCSUM: A Versatile Chinese Meeting Summarization DatasetAuthor 6 of 6ACL Findings 2023 · 11 citations
- Reconstruct Before Summarize: An Efficient Two-Step Framework for Condensing and Summarizing Meeting TranscriptsAuthor 8 of 8EMNLP 2023 · 3 citations
- Learning Locality and Isotropy in Dialogue ModelingAuthor 7 of 7ICLR 2023 · 1 citation
- Hiding Images in Deep Probabilistic ModelsAuthor 2 of 5NeurIPS 2022 · 20 citations
- A Sentence is Worth 128 Pseudo Tokens: A Semantic-Aware Contrastive Learning Framework for Sentence EmbeddingsAuthor 5 of 6ACL Findings 2022 · 16 citations
- Live Gradient Compensation for Evading Stragglers in Distributed LearningAuthor 3 of 4INFOCOM 2021 · 29 citations
