# EPS

Source: [EPS](<https://www.sota2.com/research/dataset/eps>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Benchmarks

Page 1 of 10. This is an index page, not the full collection.

- [Next index page](<https://www.sota2.com/research/dataset/eps.md?page=2>): Follow this link to discover more pages.

- [Linear Regression on Synthetic Multi-task Linear Regression (n=d=50, K=40, epsilon=0.2)](<https://www.sota2.com/research/sota/linear-regression-on-synthetic-multi-task-linear-regression-n-d-50-k-40-epsilon-0-2.md>): Linear Regression; Synthetic Multi-task Linear Regression (n=d=50, K=40, epsilon=0.2).

- [Classification Calibration on MEPS (test)](<https://www.sota2.com/research/sota/classification-calibration-on-meps-test.md>): Classification Calibration; MEPS (test).

- [Linear Regression on Synthetic Multi-task Dataset d=50, K=20, epsilon=0.2 (test)](<https://www.sota2.com/research/sota/linear-regression-on-synthetic-multi-task-dataset-d-50-k-20-epsilon-0-2-test.md>): Linear Regression; Synthetic Multi-task Dataset d=50, K=20, epsilon=0.2 (test).

- [Multi-task Logistic Regression on Synthetic Multi-task Logistic Regression (n=d=50, epsilon=0.2)](<https://www.sota2.com/research/sota/multi-task-logistic-regression-on-synthetic-multi-task-logistic-regression-n-d-50-epsilon-0-2.md>): Multi-task Logistic Regression; Synthetic Multi-task Logistic Regression (n=d=50, epsilon=0.2).

- [Logistic Regression on Logistic regression d=50, K=20, epsilon=0.2 varying per-task sample size n (test)](<https://www.sota2.com/research/sota/logistic-regression-on-logistic-regression-d-50-k-20-epsilon-0-2-varying-per-task-sample-size-n-test.md>): Logistic Regression; Logistic regression d=50, K=20, epsilon=0.2 varying per-task sample size n (test).

- [Video Prediction on KTH 10 -\> 20 steps (test)](<https://www.sota2.com/research/sota/video-prediction-on-kth-10-20-steps-test.md>): Video Prediction; KTH 10 -\> 20 steps (test).

- [Video Prediction on KTH (10 -\> 40 steps, test)](<https://www.sota2.com/research/sota/video-prediction-on-kth-10-40-steps-test.md>): Video Prediction; KTH 10 -\> 40 steps (test).

- [Data Assimilation on Kolmogorov flow short horizon 60 steps](<https://www.sota2.com/research/sota/data-assimilation-on-kolmogorov-flow-short-horizon-60-steps.md>): Data Assimilation; Kolmogorov flow short horizon 60 steps.

- [Time Series Classification on PhysioNet Sepsis (test)](<https://www.sota2.com/research/sota/time-series-classification-on-physionet-sepsis-test.md>): Time Series Classification; PhysioNet Sepsis (test).

- [Incorrect Reasoning Path Detection on DeepScaleR](<https://www.sota2.com/research/sota/incorrect-reasoning-path-detection-on-deepscaler.md>): Incorrect Reasoning Path Detection; DeepScaleR.

- [Data Assimilation on Kuramoto–Sivashinsky long horizon 640 steps](<https://www.sota2.com/research/sota/data-assimilation-on-kuramoto-sivashinsky-long-horizon-640-steps.md>): Data Assimilation; Kuramoto–Sivashinsky long horizon 640 steps.

- [Data Assimilation on Kuramoto–Sivashinsky short horizon 140 steps](<https://www.sota2.com/research/sota/data-assimilation-on-kuramoto-sivashinsky-short-horizon-140-steps.md>): Data Assimilation; Kuramoto–Sivashinsky short horizon 140 steps.

- [Data Assimilation on Kolmogorov flow long horizon 180 steps](<https://www.sota2.com/research/sota/data-assimilation-on-kolmogorov-flow-long-horizon-180-steps.md>): Data Assimilation; Kolmogorov flow long horizon 180 steps.

- [Model Performance Prediction on DeepSeek Model Families (Hold-out)](<https://www.sota2.com/research/sota/model-performance-prediction-on-deepseek-model-families-hold-out.md>): Model Performance Prediction; DeepSeek Model Families (Hold-out).

- [Semantic Segmentation on Pascal VOC overlapped setting (15-1 (6 steps))](<https://www.sota2.com/research/sota/semantic-segmentation-on-pascal-voc-overlapped-setting-15-1-6-steps.md>): Semantic Segmentation; Pascal VOC overlapped setting (15-1 (6 steps)).

- [Class-Incremental Learning on CIFAR100 B0 (20 steps) (test)](<https://www.sota2.com/research/sota/class-incremental-learning-on-cifar100-b0-20-steps-test.md>): Class-Incremental Learning; CIFAR100 B0 (20 steps) (test).

- [Class-Incremental Learning on CIFAR100-B0 5 steps (test)](<https://www.sota2.com/research/sota/class-incremental-learning-on-cifar100-b0-5-steps-test.md>): Class-Incremental Learning; CIFAR100-B0 5 steps (test).

- [Annotation Accuracy on DeepSeek-R1 Experiment 1](<https://www.sota2.com/research/sota/annotation-accuracy-on-deepseek-r1-experiment-1.md>): Annotation Accuracy; DeepSeek-R1 Experiment 1.

- [Incremental Learning on CIFAR100 10 steps](<https://www.sota2.com/research/sota/incremental-learning-on-cifar100-10-steps.md>): Incremental Learning; CIFAR100 10 steps.

- [Incremental Learning on CIFAR100 50 steps](<https://www.sota2.com/research/sota/incremental-learning-on-cifar100-50-steps.md>): Incremental Learning; CIFAR100 50 steps.

- [Mathematical Reasoning on DeepScaleR](<https://www.sota2.com/research/sota/mathematical-reasoning-on-deepscaler.md>): Mathematical Reasoning; DeepScaleR.

- [Code Generation on OlmoBaseEval Code, BigCodeBench, HumanEval, DeepSeek LeetCode, DS 1000, MBPP, MultiPL](<https://www.sota2.com/research/sota/code-generation-on-olmobaseeval-code-bigcodebench-humaneval-deepseek-leetcode-ds-1000-mbpp-multipl.md>): Code Generation; OlmoBaseEval Code BigCodeBench, HumanEval, DeepSeek LeetCode, DS 1000, MBPP, MultiPL.

- [Class-Incremental Learning on CIFAR100 50 steps (test)](<https://www.sota2.com/research/sota/class-incremental-learning-on-cifar100-50-steps-test.md>): Class-Incremental Learning; CIFAR100 50 steps (test).

- [Class Incremental Learning on CIFAR100 20 steps (test)](<https://www.sota2.com/research/sota/class-incremental-learning-on-cifar100-20-steps-test.md>): Class-Incremental Learning; CIFAR100 20 steps (test).

- [Solubility Prediction on DeepSol](<https://www.sota2.com/research/sota/solubility-prediction-on-deepsol.md>): Solubility Prediction; DeepSol.

- [Classification on Epsilon (test)](<https://www.sota2.com/research/sota/classification-on-epsilon-test.md>): Classification; Epsilon (test).

- [Semantic Segmentation on HMEPS](<https://www.sota2.com/research/sota/semantic-segmentation-on-hmeps.md>): Semantic Segmentation; HMEPS.

- [Social Interaction Simulation on EduMirror social interaction steps (144 steps) (test)](<https://www.sota2.com/research/sota/social-interaction-simulation-on-edumirror-social-interaction-steps-144-steps-test.md>): Social Interaction Simulation; EduMirror social interaction steps (144 steps) (test).

- [Reasoning on DeepScaler](<https://www.sota2.com/research/sota/reasoning-on-deepscaler.md>): Reasoning; DeepScaler.

- [Policy Synthesis on Warehouse MDPs Property: Rmin R{“steps”}min=? \[F “goal”\] W512/W1024](<https://www.sota2.com/research/sota/policy-synthesis-on-warehouse-mdps-property-rmin-r-steps-min-f-goal-w512-w1024.md>): Policy Synthesis; Warehouse MDPs Property: Rmin R{“steps”}min=? \[F “goal”\] W512/W1024.

- [Goal-reaching Planning on PointMaze Medium Long-Horizon (50 steps)](<https://www.sota2.com/research/sota/goal-reaching-planning-on-pointmaze-medium-long-horizon-50-steps.md>): Goal-reaching Planning; PointMaze Medium Long-Horizon (50 steps).

- [Agentic Search on DeepSearchQA](<https://www.sota2.com/research/sota/agentic-search-on-deepsearchqa.md>): Agentic Search; DeepSearchQA.

- [Class-Incremental Learning on TinyImageNet (10 steps, 100 base classes, test)](<https://www.sota2.com/research/sota/class-incremental-learning-on-tinyimagenet-10-steps-100-base-classes-test.md>): Class-Incremental Learning; TinyImageNet 10 steps 100 base classes (test).

- [Access Control on Muses-Bench Deepseek (max-turns=10)](<https://www.sota2.com/research/sota/access-control-on-muses-bench-deepseek-max-turns-10.md>): Access Control; Muses-Bench Deepseek max-turns = 10.

- [Currency Exchange Rate Prediction on 16 Currencies (Specific Timeframe)](<https://www.sota2.com/research/sota/currency-exchange-rate-prediction-on-16-currencies-specific-timeframe.md>): Currency exchange rate prediction; 16 currencies relative to the USD 100 input steps, 1-step horizon Jan. 1, 2022, to Sep. 16, 2022.

- [Inference Efficiency on DeepScaleR-40k (1,024 mathematical problems)](<https://www.sota2.com/research/sota/inference-efficiency-on-deepscaler-40k-1-024-mathematical-problems.md>): Inference Efficiency; DeepScaleR-40k (1,024 mathematical problems).

- [Deep Search on xbench DeepSearch (test)](<https://www.sota2.com/research/sota/deep-search-on-xbench-deepsearch-test.md>): Deep Search; xbench DeepSearch (test).

- [Acoustic Classification on DeepShip](<https://www.sota2.com/research/sota/acoustic-classification-on-deepship.md>): Acoustic Classification; DeepShip.

- [Image Classification on CIFAR-10 (epsilon=8/255, test)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-epsilon-8-255-test.md>): Image Classification; CIFAR10 epsilon = 8/255 (test).

- [Visual Jailbreak Defense on Visual Adversarial Attacks epsilon = 64/255](<https://www.sota2.com/research/sota/visual-jailbreak-defense-on-visual-adversarial-attacks-epsilon-64-255.md>): Visual Jailbreak Defense; Visual Adversarial Attacks epsilon = 64/255.

- [Visual Jailbreak Defense on Visual Adversarial Attacks epsilon = 32/255](<https://www.sota2.com/research/sota/visual-jailbreak-defense-on-visual-adversarial-attacks-epsilon-32-255.md>): Visual Jailbreak Defense; Visual Adversarial Attacks epsilon = 32/255.

- [Visual Jailbreak Defense on Visual Adversarial Attacks epsilon = 16/255](<https://www.sota2.com/research/sota/visual-jailbreak-defense-on-visual-adversarial-attacks-epsilon-16-255.md>): Visual Jailbreak Defense; Visual Adversarial Attacks epsilon = 16/255.

- [Deep Search on XBench DeepSearch (Accuracy)](<https://www.sota2.com/research/sota/deep-search-on-xbench-deepsearch-accuracy.md>): Deep Search; XBench DeepSearch.

- [Causal Outcome Forecasting on MIMIC-IV sepsis (ALT)](<https://www.sota2.com/research/sota/causal-outcome-forecasting-on-mimic-iv-sepsis-alt.md>): Causal outcome forecasting; MIMIC-IV sepsis (ALT).

- [Causal outcome forecasting on MIMIC sepsis BILIRUBIN-TOTAL IV](<https://www.sota2.com/research/sota/causal-outcome-forecasting-on-mimic-sepsis-bilirubin-total-iv.md>): Causal outcome forecasting; MIMIC sepsis BILIRUBIN-TOTAL IV.

- [Causal Outcome Forecasting on MIMIC-IV sepsis (CREATININE)](<https://www.sota2.com/research/sota/causal-outcome-forecasting-on-mimic-iv-sepsis-creatinine.md>): Causal outcome forecasting; MIMIC-IV sepsis (CREATININE).

- [Causal outcome forecasting on MIMIC-IV sepsis SOFA-SCORE](<https://www.sota2.com/research/sota/causal-outcome-forecasting-on-mimic-iv-sepsis-sofa-score.md>): Causal outcome forecasting; MIMIC-IV sepsis SOFA-SCORE.

- [Outcome Forecasting on Semi-synthetic MIMIC-IV sepsis (test)](<https://www.sota2.com/research/sota/outcome-forecasting-on-semi-synthetic-mimic-iv-sepsis-test.md>): Outcome Forecasting; Semi-synthetic MIMIC-IV sepsis (test).

- [Deep-search QA on Xbench-DeepSearch (test)](<https://www.sota2.com/research/sota/deep-search-qa-on-xbench-deepsearch-test.md>): Deep-search QA; Xbench-DeepSearch (test).

- [Incremental Learning on CIFAR100 20 steps](<https://www.sota2.com/research/sota/incremental-learning-on-cifar100-20-steps.md>): Incremental Learning; CIFAR100 20 steps.

- [Retrieval-Augmented Question Answering on DeepSearch Average](<https://www.sota2.com/research/sota/retrieval-augmented-question-answering-on-deepsearch-average.md>): Retrieval-Augmented Question Answering; DeepSearch Average.

- [Retrieval-Augmented Question Answering on DeepSearch Bamboogle](<https://www.sota2.com/research/sota/retrieval-augmented-question-answering-on-deepsearch-bamboogle.md>): Retrieval-Augmented Question Answering; DeepSearch Bamboogle.

- [Retrieval-Augmented Question Answering on DeepSearch Musique (SR, Tokens (K))](<https://www.sota2.com/research/sota/retrieval-augmented-question-answering-on-deepsearch-musique-sr-tokens-k.md>): Retrieval-Augmented Question Answering; DeepSearch Musique.

- [Retrieval-Augmented Question Answering on DeepSearch 2wiki](<https://www.sota2.com/research/sota/retrieval-augmented-question-answering-on-deepsearch-2wiki.md>): Retrieval-Augmented Question Answering; DeepSearch 2wiki.

- [Retrieval-Augmented Question Answering on DeepSearch HotpotQA](<https://www.sota2.com/research/sota/retrieval-augmented-question-answering-on-deepsearch-hotpotqa.md>): Retrieval-Augmented Question Answering; DeepSearch HotpotQA.

- [Retrieval-Augmented Question Answering on DeepSearch PopQA](<https://www.sota2.com/research/sota/retrieval-augmented-question-answering-on-deepsearch-popqa.md>): Retrieval-Augmented Question Answering; DeepSearch PopQA.

- [Retrieval-Augmented Question Answering on DeepSearch TriviaQA](<https://www.sota2.com/research/sota/retrieval-augmented-question-answering-on-deepsearch-triviaqa.md>): Retrieval-Augmented Question Answering; DeepSearch TriviaQA.

- [Retrieval-Augmented Question Answering on DeepSearch NQ](<https://www.sota2.com/research/sota/retrieval-augmented-question-answering-on-deepsearch-nq.md>): Retrieval-Augmented Question Answering; DeepSearch NQ.

- [Continual Image Classification on ImageNet-100 (10 steps)](<https://www.sota2.com/research/sota/continual-image-classification-on-imagenet-100-10-steps.md>): Continual Image Classification; ImageNet-100 10 steps.

- [Deep Information Search and Synthesis on xbench DeepSearch](<https://www.sota2.com/research/sota/deep-information-search-and-synthesis-on-xbench-deepsearch.md>): Deep Information Search and Synthesis; xbench DeepSearch.

- [Sepsis Prediction on Sepsis (test)](<https://www.sota2.com/research/sota/sepsis-prediction-on-sepsis-test.md>): Sepsis Prediction; Sepsis (test).

- [Visual Web Agent Navigation on DeepShop](<https://www.sota2.com/research/sota/visual-web-agent-navigation-on-deepshop.md>): Visual Web Agent Navigation; DeepShop.

- [Fair Order Matching on DeFi crypto-asset LOB BTC, ETH, SOL (500,000 held-out steps)](<https://www.sota2.com/research/sota/fair-order-matching-on-defi-crypto-asset-lob-btc-eth-sol-500-000-held-out-steps.md>): Fair Order Matching; DeFi crypto-asset LOB BTC, ETH, SOL (500,000 held-out steps).

- [Time Series Classification on Epilepsy-EEG](<https://www.sota2.com/research/sota/time-series-classification-on-epilepsy-eeg.md>): Time Series Classification; Epilepsy-EEG.

- [Semantic Segmentation on ADE20k overlapped setting (100-10 (6 steps))](<https://www.sota2.com/research/sota/semantic-segmentation-on-ade20k-overlapped-setting-100-10-6-steps.md>): Semantic Segmentation; ADE20k overlapped setting (100-10 (6 steps)).

- [Continual Image Classification on ImageNet-1000 10 steps](<https://www.sota2.com/research/sota/continual-image-classification-on-imagenet-1000-10-steps.md>): Continual Image Classification; ImageNet-1000 10 steps.

- [Autoregressive rollout prediction on TheWell-ASM (steps 21-60, test)](<https://www.sota2.com/research/sota/autoregressive-rollout-prediction-on-thewell-asm-steps-21-60-test.md>): Autoregressive rollout prediction; TheWell-ASM steps 21-60 T ∈ \[21: 60\] (test).

- [Rhetorical Role Labeling on SCOTUSSteps (test)](<https://www.sota2.com/research/sota/rhetorical-role-labeling-on-scotussteps-test.md>): Rhetorical Role Labeling; SCOTUSSteps (test).

- [Budget Allocation on Twin-2K 500 reps](<https://www.sota2.com/research/sota/budget-allocation-on-twin-2k-500-reps.md>): Budget Allocation; Twin-2K 500 reps.

- [Classification on PhysioNet Sepsis 2019 (test)](<https://www.sota2.com/research/sota/classification-on-physionet-sepsis-2019-test.md>): Classification; PhysioNet Sepsis 2019 (test).

- [Deep Search on xBench DeepSearch DS-2505](<https://www.sota2.com/research/sota/deep-search-on-xbench-deepsearch-ds-2505.md>): Deep Search; xBench DeepSearch DS-2505.

- [Dual-arm Manipulation on RoboTwin Long & Extra Long Horizon Tasks 2.0](<https://www.sota2.com/research/sota/dual-arm-manipulation-on-robotwin-long-and-extra-long-horizon-tasks-2-0.md>): Dual-arm Manipulation; RoboTwin Long & Extra Long Horizon Tasks 280-650 Steps 2.0.

- [Dual-arm Manipulation on RoboTwin Medium Horizon Tasks 150-230 Steps 2.0](<https://www.sota2.com/research/sota/dual-arm-manipulation-on-robotwin-medium-horizon-tasks-150-230-steps-2-0.md>): Dual-arm Manipulation; RoboTwin Medium Horizon Tasks 150-230 Steps 2.0.

- [Dual-arm Manipulation on RoboTwin Short Horizon Tasks 100-130 Steps 2.0](<https://www.sota2.com/research/sota/dual-arm-manipulation-on-robotwin-short-horizon-tasks-100-130-steps-2-0.md>): Dual-arm Manipulation; RoboTwin Short Horizon Tasks 100-130 Steps 2.0.

- [Jailbreak Attack on DeepSeek](<https://www.sota2.com/research/sota/jailbreak-attack-on-deepseek.md>): Jailbreak Attack; DeepSeek.

- [Robustness Verification on CIFAR-10 epsilon=2/255 (first 200 examples)](<https://www.sota2.com/research/sota/robustness-verification-on-cifar-10-epsilon-2-255-first-200-examples.md>): Robustness Verification; CIFAR-10 epsilon=2/255 (first 200 examples).

- [Symbolic Regression on Feynman Dataset epsilon=0.1 (test)](<https://www.sota2.com/research/sota/symbolic-regression-on-feynman-dataset-epsilon-0-1-test.md>): Symbolic Regression; Feynman Dataset epsilon=0.1 (test).

- [Symbolic Regression on Strogatz Dataset epsilon=0.1 (test)](<https://www.sota2.com/research/sota/symbolic-regression-on-strogatz-dataset-epsilon-0-1-test.md>): Symbolic Regression; Strogatz Dataset epsilon=0.1 (test).

- [Symbolic Regression on Feynman Dataset epsilon=0.01 (test)](<https://www.sota2.com/research/sota/symbolic-regression-on-feynman-dataset-epsilon-0-01-test.md>): Symbolic Regression; Feynman Dataset epsilon=0.01 (test).

- [Symbolic Regression on Strogatz Dataset epsilon=0.01 (test)](<https://www.sota2.com/research/sota/symbolic-regression-on-strogatz-dataset-epsilon-0-01-test.md>): Symbolic Regression; Strogatz Dataset epsilon=0.01 (test).