AI Paper Atlas

Research map / Benchmarks / SAMSum

SAMSum leaderboard

10 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on SAMSum. Best ROUGE-L: 52.7 (Adversarial moment-matching, NeurIPS 2024).

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.

SAMSum · ROUGE-L

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

#ROUGE-L ↑MethodPaperTopic
152.7Adversarial moment-matching
T5-Large
Adversarial Moment-Matching Distillation of Large Language Models
NeurIPS 2024 · Table 2
Language, safety and interpretability
249.11DistiLLM
T5-Base
DistiLLM: Towards Streamlined Distillation for Large Language Models
ICML 2024 · Text summarization and machine
Language, safety and interpretability
348.09UAT
T5-large
Beyond Greedy Exits: Improved Early Exit Decisions for Risk Control and Reliability
NeurIPS 2025 · Table 2
RNNs and normalization
444.7PrivateLoRA
Mixtral 8x7B
Open LLMs are Necessary for Current Private Adaptations and Outperform their Closed Alternatives
NeurIPS 2024 · SAMSum results table
Privacy attacks and secure learning
544.4BOMF
LLaMA-2-7B
Model Fusion through Bayesian Optimization in Language Model Fine-Tuning
NeurIPS 2024 · Table 2
Bayesian optimization
644.09MuToR
Gemma 2B
Multi-Token Prediction Needs Registers
NeurIPS 2025 · Summarization table
Instruction and fine-tuning
743.7DoRA + SDT
Mamba-1.4B
Parameter-Efficient Fine-Tuning of State Space Models
ICML 2025 · Table: Performance comparison
Pruning and efficient tuning
843.13LogiGAN largeLogiGAN: Learning Logical Reasoning via Adversarial Pre-training
NeurIPS 2022 · Table 1
Language, safety and interpretability
940.05DA-KD
Qwen2.5-1.5B
DA-KD: Difficulty-Aware Knowledge Distillation for Efficient Large Language Models
ICML 2025 · Table 2
Knowledge distillation
1039.1Hedgehog (LoRA)
Llama-2 7B
The Hedgehog & the Porcupine: Expressive Linear Attentions with Softmax Mimicry
ICLR 2024 · Table: Llama-2 conversion (ROU
Attention and transformers

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