AI Paper Atlas

Research map / Benchmarks / CNN/DailyMail

CNN/DailyMail leaderboard

35 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on CNN/DailyMail. Best ROUGE-1: 48.54 (LSR-MWF, AAAI 2025).

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.

CNN/DailyMail · ROUGE-1

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

#ROUGE-1 ↑MethodPaperTopic
148.54LSR-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
247.97SLiC
PEGASUS-XXL
Calibrating Sequence likelihood Improves Conditional Language Generation
ICLR 2023 · Final Results table
Language, safety and interpretability
345.6GENIEText Generation with Diffusion Language Models: A Pre-training Approach with Continuous Paragraph Denoise
ICML 2023 · Table 2
Autoregressive and latent generation
444.94SalientLoRA
BART-large
Unveiling LoRA Intrinsic Ranks via Salience Analysis
NeurIPS 2024 · Table 2
Pruning and efficient tuning
544.51BART + RD
BART
R-Drop: Regularized Dropout for Neural Networks
NeurIPS 2021 · Table 4
Attention and transformers
644.41BART-large w.t. DITTO
BART-large
Learning to Break the Loop: Analyzing and Mitigating Repetitions for Neural Text Generation
NeurIPS 2022 · Table 4
Language, safety and interpretability
744.38BART+R3F
BART
Towards Inferential Reproducibility of Machine Learning Research
ICLR 2023 · Table: reported text summariza
Molecules and proteins
844.38BART+R3F
BART
Better Fine-Tuning by Reducing Representational Collapse
ICLR 2021 · Summarization results table
Pruning and efficient tuning
944.31Switchable Decision
BART-large
Switchable Decision: Dynamic Neural Generation Networks
ICML 2024 · Table: Comparison to models on
RNNs and normalization
1044.26Multi-agent discussion mechanismMulti-Agent Discussion Mechanism for Natural Language Generation
AAAI 2019 · Table 4
LLM agents
20 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

CNN/DailyMail · ROUGE-L

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

#ROUGE-L ↑MethodPaperTopic
145.1LSR-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
241.78SalientLoRA
BART-large
Unveiling LoRA Intrinsic Ranks via Salience Analysis
NeurIPS 2024 · Table 2
Pruning and efficient tuning
341.52BiLD aligned, minimal degradation
T5-small + T5-large
Speculative Decoding with Big Little Decoder
NeurIPS 2023 · Table 1
Reasoning and chain of thought
441.17BART+R3F
BART
Towards Inferential Reproducibility of Machine Learning Research
ICLR 2023 · Table: reported text summariza
Molecules and proteins
541.17BART+R3F
BART
Better Fine-Tuning by Reducing Representational Collapse
ICLR 2021 · Summarization results table
Pruning and efficient tuning
641.16BART-large w.t. DITTO
BART-large
Learning to Break the Loop: Analyzing and Mitigating Repetitions for Neural Text Generation
NeurIPS 2022 · Table 4
Language, safety and interpretability
741.01Switchable Decision
BART-large
Switchable Decision: Dynamic Neural Generation Networks
ICML 2024 · Table: Comparison to models on
RNNs and normalization
840.72B-only LoRA (r=16)
BART-large
Asymmetry in Low-Rank Adapters of Foundation Models
ICML 2024 · Table 3
Pruning and efficient tuning
940.56BERT-Ext+ContextRewriter
BERT-base
Contextualized Rewriting for Text Summarization
AAAI 2021 · Results table
Language, safety and interpretability
1040.51UniLM
BERT_LARGE
Unified Language Model Pre-training for Natural Language Understanding and Generation
NeurIPS 2019 · Table: CNN/DailyMail summariza
Language, safety and interpretability
16 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

CNN/DailyMail · ROUGE-2

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

#ROUGE-2 ↑MethodPaperTopic
123.91LSR-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
221.86SalientLoRA
BART-large
Unveiling LoRA Intrinsic Ranks via Salience Analysis
NeurIPS 2024 · Table 2
Pruning and efficient tuning
321.53BART+R3F
BART
Towards Inferential Reproducibility of Machine Learning Research
ICLR 2023 · Table: reported text summariza
Molecules and proteins
421.53BART+R3F
BART
Better Fine-Tuning by Reducing Representational Collapse
ICLR 2021 · Summarization results table
Pruning and efficient tuning
521.45BART-large w.t. DITTO
BART-large
Learning to Break the Loop: Analyzing and Mitigating Repetitions for Neural Text Generation
NeurIPS 2022 · Table 4
Language, safety and interpretability
620.62B-only LoRA (r=16)
BART-large
Asymmetry in Low-Rank Adapters of Foundation Models
ICML 2024 · Table 3
Pruning and efficient tuning
720.57BERT-Ext+ContextRewriter
BERT-base
Contextualized Rewriting for Text Summarization
AAAI 2021 · Results table
Language, safety and interpretability
819.43Copy/Rewrite + History + HRL + BERTCopy or Rewrite: Hybrid Summarization with Hierarchical Reinforcement Learning
AAAI 2020 · Table 1
Visual and representation RL
919.17HATSExploring Human-Like Reading Strategy for Abstractive Text Summarization
AAAI 2019 · Table 1
Language, safety and interpretability
1018.97TaLK Convolution (Deep)Time-aware Large Kernel Convolutions
ICML 2020 · CNN-DailyMail results table
Attention and transformers
11 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.