# SAM

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

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 38. This is an index page, not the full collection.

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

- [Out-of-sample random-X Regression on Spiked Covariance Model (d/N=0.9, noise=1)](<https://www.sota2.com/research/sota/out-of-sample-random-x-regression-on-spiked-covariance-model-d-n-0-9-noise-1.md>): Out-of-sample random-X regression; Spiked covariance model d/N = 0.9, noise = 1 synthetic (out-of-sample).

- [Image Generation on ImageNet 256 10k samples](<https://www.sota2.com/research/sota/image-generation-on-imagenet-256-10k-samples.md>): Image Generation; ImageNet 256 10k samples.

- [Conditional Image Generation on ImageNet 256x256 Guided-Diffusion (10k samples)](<https://www.sota2.com/research/sota/conditional-image-generation-on-imagenet-256x256-guided-diffusion-10k-samples.md>): Conditional Image Generation; ImageNet 256x256 Guided-Diffusion (10k samples).

- [Text-to-Image Generation on Stable Diffusion 10k samples v1.4](<https://www.sota2.com/research/sota/text-to-image-generation-on-stable-diffusion-10k-samples-v1-4.md>): Text-to-Image Generation; Stable Diffusion 10k samples v1.4.

- [Regression on Synthetic Spiked Covariance d/N = 0.9 (out-of-sample)](<https://www.sota2.com/research/sota/regression-on-synthetic-spiked-covariance-d-n-0-9-out-of-sample.md>): Regression; Synthetic Spiked Covariance d/N = 0.9 (out-of-sample).

- [Image Classification on ImageNet-32 down-sampled (test)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-32-down-sampled-test.md>): Image Classification; ImageNet-32 down-sampled (test).

- [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).

- [Uncertainty Quantification on NQ 16 additional samples (test)](<https://www.sota2.com/research/sota/uncertainty-quantification-on-nq-16-additional-samples-test.md>): Uncertainty Quantification; NQ 16 additional samples (test).

- [Uncertainty Quantification on TQA 16 additional samples (test)](<https://www.sota2.com/research/sota/uncertainty-quantification-on-tqa-16-additional-samples-test.md>): Uncertainty Quantification; TQA 16 additional samples (test).

- [Question Answering on NQ 8 additional samples](<https://www.sota2.com/research/sota/question-answering-on-nq-8-additional-samples.md>): Question Answering; NQ 8 additional samples.

- [Question Answering on TQA 8 additional samples (AUROC)](<https://www.sota2.com/research/sota/question-answering-on-tqa-8-additional-samples-auroc.md>): Question Answering; TQA 8 additional samples.

- [Uncertainty Quantification on NQ 4 additional samples (test)](<https://www.sota2.com/research/sota/uncertainty-quantification-on-nq-4-additional-samples-test.md>): Uncertainty Quantification; NQ 4 additional samples (test).

- [Uncertainty Quantification on TQA 4 additional samples (test)](<https://www.sota2.com/research/sota/uncertainty-quantification-on-tqa-4-additional-samples-test.md>): Uncertainty Quantification; TQA 4 additional samples (test).

- [Scene Text Recognition on CUTE 288 samples (test)](<https://www.sota2.com/research/sota/scene-text-recognition-on-cute-288-samples-test.md>): Scene Text Recognition; CUTE 288 samples (test).

- [Watermark Detection on Vicuna-7b 16k 50 samples v1.5](<https://www.sota2.com/research/sota/watermark-detection-on-vicuna-7b-16k-50-samples-v1-5.md>): Watermark Detection; Vicuna-7b 16k 50 samples v1.5.

- [Option Hedging on S&P 500 options 317 shorter prediction horizons (out-of-sample)](<https://www.sota2.com/research/sota/option-hedging-on-s-and-p-500-options-317-shorter-prediction-horizons-out-of-sample.md>): Option Hedging; S&P 500 options 317 shorter prediction horizons (out-of-sample).

- [Option Pricing on Option pricing data 317 shorter prediction horizons (out-of-sample)](<https://www.sota2.com/research/sota/option-pricing-on-option-pricing-data-317-shorter-prediction-horizons-out-of-sample.md>): Option Pricing; Option pricing data 317 shorter prediction horizons (out-of-sample).

- [Uncertainty Quantification on Aggregated Experimental Datasets (XSum, SamSum, CNN, WMT19, MedQUAD, TruthfulQA, CoQA, SciQ, TriviaQA, MMLU, GSM8k) (test)](<https://www.sota2.com/research/sota/uncertainty-quantification-on-aggregated-experimental-datasets-xsum-samsum-cnn-wmt19-medquad-truthfulqa-coqa-sciq-triviaqa-mmlu-gsm8k-test.md>): Uncertainty Quantification; Aggregated Experimental Datasets (XSum, SamSum, CNN, WMT19, MedQUAD, TruthfulQA, CoQA, SciQ, TriviaQA, MMLU, GSM8k) (test).

- [Semantic Segmentation on Cityscapes 1/16 (186 labeled samples)](<https://www.sota2.com/research/sota/semantic-segmentation-on-cityscapes-1-16-186-labeled-samples.md>): Semantic Segmentation; Cityscapes 1/16 (186 labeled samples).

- [Dialogue Summarization on SAMSum (test)](<https://www.sota2.com/research/sota/dialogue-summarization-on-samsum-test.md>): Dialogue Summarization; SAMSum (test).

- [Image Generation on CIFAR10 50k samples (test)](<https://www.sota2.com/research/sota/image-generation-on-cifar10-50k-samples-test.md>): Image Generation; CIFAR10 50k samples (test).

- [Abstractive Summarization on SAMSum](<https://www.sota2.com/research/sota/abstractive-summarization-on-samsum.md>): Abstractive Summarization; SAMSum.

- [Uncertainty Quantification on Musique 500 randomly sampled queries (test)](<https://www.sota2.com/research/sota/uncertainty-quantification-on-musique-500-randomly-sampled-queries-test.md>): Uncertainty Quantification; Musique 500 randomly sampled queries (test).

- [Uncertainty Quantification on HotpotQA 500 randomly sampled queries (test)](<https://www.sota2.com/research/sota/uncertainty-quantification-on-hotpotqa-500-randomly-sampled-queries-test.md>): Uncertainty Quantification; HotpotQA 500 randomly sampled queries (test).

- [Uncertainty Quantification on PopQA 500 randomly sampled queries (test)](<https://www.sota2.com/research/sota/uncertainty-quantification-on-popqa-500-randomly-sampled-queries-test.md>): Uncertainty Quantification; PopQA 500 randomly sampled queries (test).

- [Individual Treatment Effect Estimation on IHDP (within-sample)](<https://www.sota2.com/research/sota/individual-treatment-effect-estimation-on-ihdp-within-sample.md>): Individual Treatment Effect Estimation; IHDP (within-sample).

- [SIR Dynamics Forecasting on Erdős-Rényi graphs (out-of-sample)](<https://www.sota2.com/research/sota/sir-dynamics-forecasting-on-erd-s-r-nyi-graphs-out-of-sample.md>): SIR dynamics forecasting; Erdős-Rényi graphs (out-of-sample).

- [Summarization on SamSum (PRR)](<https://www.sota2.com/research/sota/summarization-on-samsum-prr.md>): Summarization; SamSum.

- [Selective Generation on SamSum](<https://www.sota2.com/research/sota/selective-generation-on-samsum.md>): Selective Generation; SamSum.

- [Cross-view geo-localization on VIGOR (Same-Area)](<https://www.sota2.com/research/sota/cross-view-geo-localization-on-vigor-same-area.md>): Cross-view geo-localization; VIGOR (Same-Area).

- [Semantic Segmentation on Cityscapes (1/8 subset, val split)](<https://www.sota2.com/research/sota/semantic-segmentation-on-cityscapes-1-8-subset-val-split.md>): Semantic Segmentation; CITYSCAPES 1/8 labeled samples 372 labels (val).

- [Individual Treatment Effect Estimation on IHDP (out-of-sample)](<https://www.sota2.com/research/sota/individual-treatment-effect-estimation-on-ihdp-out-of-sample.md>): Individual Treatment Effect Estimation; IHDP (out-of-sample).

- [Membership Privacy Estimation on CiteSeer 50% nodes sampled](<https://www.sota2.com/research/sota/membership-privacy-estimation-on-citeseer-50-percent-nodes-sampled.md>): Membership Privacy Estimation; CiteSeer 50% nodes sampled.

- [Privacy Parameter Estimation on CiteSeer 25% nodes sampled (train)](<https://www.sota2.com/research/sota/privacy-parameter-estimation-on-citeseer-25-percent-nodes-sampled-train.md>): Privacy parameter estimation; CiteSeer 25% nodes sampled (train).

- [Option Hedging on 317 longer prediction horizons (out-of-sample)](<https://www.sota2.com/research/sota/option-hedging-on-317-longer-prediction-horizons-out-of-sample.md>): Option hedging; 317 longer prediction horizons (out-of-sample).

- [Watermark Detection on Llama-2-7b-chat-hf 10 samples UMD watermarking (test)](<https://www.sota2.com/research/sota/watermark-detection-on-llama-2-7b-chat-hf-10-samples-umd-watermarking-test.md>): Watermark Detection; Llama-2-7b-chat-hf 10 samples UMD watermarking (test).

- [Image Generation on CIFAR-10 32x32 (50K samples)](<https://www.sota2.com/research/sota/image-generation-on-cifar-10-32x32-50k-samples.md>): Image Generation; CIFAR-10 32x32 (50K samples).

- [Harmful Question Answering on BeaverTails HarmfulQA (1k and 10k samples)](<https://www.sota2.com/research/sota/harmful-question-answering-on-beavertails-harmfulqa-1k-and-10k-samples.md>): Harmful question-answering; BeaverTails HarmfulQA (1k and 10k samples).

- [Watermark Forgery Detection on Stable-Diffusion-Prompts 1,000 samples (test)](<https://www.sota2.com/research/sota/watermark-forgery-detection-on-stable-diffusion-prompts-1-000-samples-test.md>): Watermark Forgery Detection; Stable-Diffusion-Prompts 1,000 samples (test).

- [Micro-expression recognition on SAMM 3-class](<https://www.sota2.com/research/sota/micro-expression-recognition-on-samm-3-class.md>): Micro-expression recognition; SAMM 3-class.

- [Super-Resolution on Urban bicubic downsampling (test)](<https://www.sota2.com/research/sota/super-resolution-on-urban-bicubic-downsampling-test.md>): Super-Resolution; Urban bicubic downsampling (test).

- [Super-Resolution on DIV2K bicubic downsampling (test)](<https://www.sota2.com/research/sota/super-resolution-on-div2k-bicubic-downsampling-test.md>): Super-Resolution; DIV2K bicubic downsampling (test).

- [Scene Text Recognition on IIIT5K 3,000 samples (test)](<https://www.sota2.com/research/sota/scene-text-recognition-on-iiit5k-3-000-samples-test.md>): Scene Text Recognition; IIIT5K 3,000 samples (test).

- [Image-to-Code Generation on STEM2Code-Eval (1k samples)](<https://www.sota2.com/research/sota/image-to-code-generation-on-stem2code-eval-1k-samples.md>): Image-to-Code Generation; STEM2Code-Eval (1k samples).

- [Micro-expression recognition on SAMM](<https://www.sota2.com/research/sota/micro-expression-recognition-on-samm.md>): Micro-expression recognition; SAMM.

- [Micro-expression recognition on SAMM 5-class](<https://www.sota2.com/research/sota/micro-expression-recognition-on-samm-5-class.md>): Micro-expression recognition; SAMM 5-class.

- [CATE Estimation on IHDP 100 train test splits (out-sample)](<https://www.sota2.com/research/sota/cate-estimation-on-ihdp-100-train-test-splits-out-sample.md>): CATE Estimation; IHDP 100 train test splits (out-sample).

- [Abstractive Dialogue Summarization on SAMSum (test)](<https://www.sota2.com/research/sota/abstractive-dialogue-summarization-on-samsum-test.md>): Abstractive dialogue summarization; SAMSum (test).

- [Image Generation on ImageNet 50k samples](<https://www.sota2.com/research/sota/image-generation-on-imagenet-50k-samples.md>): Image Generation; ImageNet 50k samples.

- [Evidence-round recall on LongMemEval 100 question sample](<https://www.sota2.com/research/sota/evidence-round-recall-on-longmemeval-100-question-sample.md>): Evidence-round recall; LongMemEval 100 question sample.

- [Jailbreak Attack Evaluation on SafeBench 100 sampled harmful queries](<https://www.sota2.com/research/sota/jailbreak-attack-evaluation-on-safebench-100-sampled-harmful-queries.md>): Jailbreak Attack Evaluation; SafeBench 100 sampled harmful queries.

- [DAG structure recovery on non-linear-1 5000 samples](<https://www.sota2.com/research/sota/dag-structure-recovery-on-non-linear-1-5000-samples.md>): DAG structure recovery; non-linear-1 5000 samples.

- [Sliced Wasserstein Distance Estimation on Gaussian toy samples d=30](<https://www.sota2.com/research/sota/sliced-wasserstein-distance-estimation-on-gaussian-toy-samples-d-30.md>): Sliced Wasserstein Distance Estimation; Gaussian toy samples d=30.

- [Underwater Image Enhancement on SeaThru (sample images)](<https://www.sota2.com/research/sota/underwater-image-enhancement-on-seathru-sample-images.md>): Underwater Image Enhancement; SeaThru (sample images).

- [Classification on National Health and Nutrition Examination Survey Employee 2013–2014 cycle (random sample splits)](<https://www.sota2.com/research/sota/classification-on-national-health-and-nutrition-examination-survey-employee-2013-2014-cycle-random-sample-splits.md>): Classification; National Health and Nutrition Examination Survey Employee 2013–2014 cycle (random sample splits).

- [Scene Text Recognition on SVTP 645 samples (test)](<https://www.sota2.com/research/sota/scene-text-recognition-on-svtp-645-samples-test.md>): Scene Text Recognition; SVTP 645 samples (test).

- [Hallucination Detection on RAGTruth CNN/DM (subsample)](<https://www.sota2.com/research/sota/hallucination-detection-on-ragtruth-cnn-dm-subsample.md>): Hallucination detection; RAGTruth CNN/DM (subsample).

- [Hallucination Detection on RAGTruth MS MARCO (subsample)](<https://www.sota2.com/research/sota/hallucination-detection-on-ragtruth-ms-marco-subsample.md>): Hallucination detection; RAGTruth MS MARCO (subsample).

- [Question Answering on DisambiguationQA](<https://www.sota2.com/research/sota/question-answering-on-disambiguationqa.md>): Question Answering; DisambiguationQA.

- [Few-shot Classification on CPTAC k=25 positive samples (test)](<https://www.sota2.com/research/sota/few-shot-classification-on-cptac-k-25-positive-samples-test.md>): Few-shot classification; CPTAC k=25 positive samples (test).

- [Multi-agent consensus coordination on Multi-Agent LLM Consensus System (50 samples)](<https://www.sota2.com/research/sota/multi-agent-consensus-coordination-on-multi-agent-llm-consensus-system-50-samples.md>): Multi-agent consensus coordination; Multi-Agent LLM Consensus System 50 samples per configuration.

- [Classification on Higgs 30K subsamples](<https://www.sota2.com/research/sota/classification-on-higgs-30k-subsamples.md>): Classification; Higgs 30K subsamples.

- [Classification on Covertype 30K subsamples](<https://www.sota2.com/research/sota/classification-on-covertype-30k-subsamples.md>): Classification; Covertype 30K subsamples.

- [Classification on MiniBooNE 30K subsamples](<https://www.sota2.com/research/sota/classification-on-miniboone-30k-subsamples.md>): Classification; MiniBooNE 30K subsamples.

- [Unconditional Image Generation on CIFAR-10 32x32 Score-SDE 50k samples (test)](<https://www.sota2.com/research/sota/unconditional-image-generation-on-cifar-10-32x32-score-sde-50k-samples-test.md>): Unconditional Image Generation; CIFAR-10 32x32 Score-SDE 50k samples (test).

- [Image Generation on ImageNet 64x64 (50k samples)](<https://www.sota2.com/research/sota/image-generation-on-imagenet-64x64-50k-samples.md>): Image Generation; ImageNet 64x64 (50k samples).

- [Text-to-image generation on Stable Diffusion 10k samples v1.4 (test)](<https://www.sota2.com/research/sota/text-to-image-generation-on-stable-diffusion-10k-samples-v1-4-test.md>): Text-to-image generation; Stable Diffusion 10k samples v1.4 (test).

- [Latency Prediction on NNLQP (Same Distribution)](<https://www.sota2.com/research/sota/latency-prediction-on-nnlqp-same-distribution.md>): Latency Prediction; NNLQP (Same Distribution).

- [Text generation on DFM 1024 samples masked backbone](<https://www.sota2.com/research/sota/text-generation-on-dfm-1024-samples-masked-backbone.md>): Text generation; DFM 1024 samples masked backbone.

- [Tactical Move Prediction on Lichess tactical puzzles v18 (reservoir sample (n=74,424))](<https://www.sota2.com/research/sota/tactical-move-prediction-on-lichess-tactical-puzzles-v18-reservoir-sample-n-74-424.md>): Tactical Move Prediction; Lichess tactical puzzles v18 (reservoir sample (n=74,424)).

- [Debugging on OR-Debug-Bench (450 stratified samples)](<https://www.sota2.com/research/sota/debugging-on-or-debug-bench-450-stratified-samples.md>): Debugging; OR-Debug-Bench (450 stratified samples).

- [Knowledge Recovery on TOFU 10% (400 samples) 1.0 (forget)](<https://www.sota2.com/research/sota/knowledge-recovery-on-tofu-10-percent-400-samples-1-0-forget.md>): Knowledge Recovery; TOFU 10% (400 samples) 1.0 (forget).

- [Location and orientation estimation on VIGOR Same-Area](<https://www.sota2.com/research/sota/location-and-orientation-estimation-on-vigor-same-area.md>): Location and orientation estimation; VIGOR Same-Area.

- [Spatial Relationship Accuracy on MD-3k Same](<https://www.sota2.com/research/sota/spatial-relationship-accuracy-on-md-3k-same.md>): Spatial Relationship Accuracy; MD-3k Same.

- [Vision Model Representation Alignment on ImageNet 20% held-out split of 20,000 samples (val)](<https://www.sota2.com/research/sota/vision-model-representation-alignment-on-imagenet-20-percent-held-out-split-of-20-000-samples-val.md>): Vision Model Representation Alignment; ImageNet 20% held-out split of 20,000 samples (val).

- [Unconditional Image Generation on CIFAR-10 32x32 (EDM) (50k samples)](<https://www.sota2.com/research/sota/unconditional-image-generation-on-cifar-10-32x32-edm-50k-samples.md>): Unconditional Image Generation; CIFAR-10 32x32 (EDM) (50k samples).

- [Image Generation on FFHQ 64x64 (EDM) 50k samples](<https://www.sota2.com/research/sota/image-generation-on-ffhq-64x64-edm-50k-samples.md>): Image Generation; FFHQ 64x64 (EDM) 50k samples.

- [Few-shot Learning on SAMSum](<https://www.sota2.com/research/sota/few-shot-learning-on-samsum.md>): Few-shot Learning; SAMSum.

- [Reproducibility on Real fMRI dataset (split-sample)](<https://www.sota2.com/research/sota/reproducibility-on-real-fmri-dataset-split-sample.md>): Reproducibility; Real fMRI dataset (split-sample).

- [Tensor Completion on Order-3 nonnegative tensors n=500 samples (test)](<https://www.sota2.com/research/sota/tensor-completion-on-order-3-nonnegative-tensors-n-500-samples-test.md>): Tensor Completion; Order-3 nonnegative tensors n=500 samples (test).
