AI Paper Atlas

Research map / Benchmarks / XSum

XSum leaderboard

36 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on XSum. Best ROUGE-1: 49.77 (SLiC, ICLR 2023).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

XSum · ROUGE-1

27 papers, each with its best reported number for its own method; higher is better.

#ROUGE-1 ↑MethodPaperTopic
149.77SLiC
PEGASUS-XXL
Calibrating Sequence likelihood Improves Conditional Language Generation
ICLR 2023 · Final Results table
Language, safety and interpretability
247.77FES
PEGASUS
Towards Improving Faithfulness in Abstractive Summarization
NeurIPS 2022 · Main results table
Visual question answering
347.76PEGA-CoNT
PEGASUS-large
CoNT: Contrastive Neural Text Generation
NeurIPS 2022 · Table 2
Text-to-image diffusion
447.47LSR-MWF
BART-large
Distilling Structured Rationale from Large Language Models to Small Language Models for Abstractive Summarization
AAAI 2025 · Table 2
Language, safety and interpretability
545.65SeqCo (λ_x-y)
BART Large
Sequence Level Contrastive Learning for Text Summarization
AAAI 2022 · Table 4
Recognition and long-tail
645.2Switchable Decision
BART-large
Switchable Decision: Dynamic Neural Generation Networks
ICML 2024 · Table: Comparison to models on
RNNs and normalization
744UniLMv2 baseUniLMv2: Pseudo-Masked Language Models for Unified Language Model Pre-Training
ICML 2020 · Table: Abstractive summarizati
Language, safety and interpretability
843.93SalientLoRA
BART-large
Unveiling LoRA Intrinsic Ranks via Salience Analysis
NeurIPS 2024 · Table 2
Pruning and efficient tuning
943.83SW (Soft)
BART-large
Watermarking Conditional Text Generation for AI Detection: Unveiling Challenges and a Semantic-Aware Watermark Remedy
AAAI 2024 · Main results table
Language, safety and interpretability
1042.91B-only LoRA (r=16)
BART-large
Asymmetry in Low-Rank Adapters of Foundation Models
ICML 2024 · Table 3
Pruning and efficient tuning
17 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

XSum · ROUGE-L

20 papers, each with its best reported number for its own method; higher is better.

10 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

XSum · ROUGE-2

15 papers, each with its best reported number for its own method; higher is better.

5 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.