AI Paper Atlas

Research map / Benchmarks / CoLA

CoLA leaderboard

29 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on CoLA. Best Matthews correlation: 68.5 (SIFT, ICML 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.

CoLA · Matthews correlation

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

#Matthews correlation ↑MethodPaperTopic
168.5SIFT
RoBERTa-Large
Sparse is Enough in Fine-tuning Pre-trained Large Language Models
ICML 2024 · Table 1
Recognition and long-tail
265.53LoRAM
DeBERTa-v3-base
The Primacy of Magnitude in Low-Rank Adaptation
NeurIPS 2025 · GLUE results table
Pruning and efficient tuning
362.3SUMO (SVD, rank=4)
RoBERTa-Base
SUMO: Subspace-Aware Moment-Orthogonalization for Accelerating Memory-Efficient LLM Training
NeurIPS 2025 · GLUE results table
Optimizers
461.59LSTSD
BERT
Long Short-Term Sample Distillation
AAAI 2020 · GLUE results table
Knowledge distillation
561.4AdaRankGrad (initial rank=4)
RoBERTa-Base
AdaRankGrad: Adaptive Gradient Rank and Moments for Memory-Efficient LLMs Training and Fine-Tuning
ICLR 2025 · Table: GLUE benchmark
Pruning and efficient tuning
660.2Greedy soup
T5
Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time
ICML 2022 · Table: GLUE results
Pruning and efficient tuning
757.4PALBERT, full configuration
ALBERT
PALBERT: Teaching ALBERT to Ponder
NeurIPS 2022 · Table 1
RNNs and normalization
855.9TA-VAT
BERT-base
Token-Aware Virtual Adversarial Training in Natural Language Understanding
AAAI 2021 · Table 2
Attacks on text and graphs
953.74MCA-BERT (α=0.2)
BERT_BASE
Fast Monte-Carlo Approximation of the Attention Mechanism
AAAI 2022 · Table 1
Attention and transformers
1052.3PoWER-BERT
BERT_BASE
PoWER-BERT: Accelerating BERT Inference via Progressive Word-vector Elimination
ICML 2020 · Table: Comparison between PoWE
Attention and transformers
6 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

CoLA · accuracy

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

#accuracy ↑MethodPaperTopic
186.77GP-filter
Qwen3-14B target, Qwen3-8B proxy
Advanced Black-Box Tuning of Large Language Models with Limited API Calls
AAAI 2026 · Table 1
Language, safety and interpretability
285.8StructBERT Base
BERT Base
StructBERT: Incorporating Language Structures into Pre-training for Deep Language Understanding
ICLR 2020 · Ablation over pre-training obj
Language, safety and interpretability
383.8LayerIF Top 25%LayerIF: Estimating Layer Quality for Large Language Models using Influence Functions
NeurIPS 2025 · Expert allocation results tabl
Language, safety and interpretability
483.61Flat-LoRA
T5-Base
Flat-LoRA: Low-Rank Adaptation over a Flat Loss Landscape
ICML 2025 · Table: Comparison with other L
Pruning and efficient tuning
582.04LoRA-One
T5 base
LoRA-One: One-Step Full Gradient Could Suffice for Fine-Tuning Large Language Models, Provably and Efficiently
ICML 2025 · Table: Accuracy comparison on
Optimizers
681CoTACS k=3, 13B, 1000 examples
Llama-2 13B
Learning to Reason via Program Generation, Emulation, and Search
NeurIPS 2024 · main results table
Agents, code and math
773.8LSH-MoE
GPT-MoE (52B)
LSH-MoE: Communication-efficient MoE Training via Locality-Sensitive Hashing
NeurIPS 2024 · GLUE results table
Mixture of experts
872CasMoE (EAP + EAM, eps=0.3)
Mixtral-8x7B (mixed quantization)
CasMoE: A Cascaded Framework for Efficient MoE Inference on Resource-constrained Devices
AAAI 2026 · Table 3
Mixture of experts
972G-fair-Prompting (Greedy)
LLaMA-65B
Fairness-guided Few-shot Prompting for Large Language Models
NeurIPS 2023 · Table 2
Language, safety and interpretability
1071.1Mistral-7B (+ T_pF_o)
Mistral-7B
Language Models Represent Beliefs of Self and Others
ICML 2024 · Table: Model performance compa
Language, safety and interpretability
3 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.