datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
msc-cifar100
MSC — Minimum Sufficient Compute
Artifacts for Is Compute Difficulty Architecture-Agnostic? Measuring and
Distilling Per-Sample Minimum Sufficient Computation.
Generated 2026-08-06T03:56:37Z by msc_lib v1.0.0.
Repositories
Shanmuk4622/msc-cifar100 — everything, one folder per run
What MSC is
The smallest cost-normalised configuration at which a network's decision has
stably settled to its full-compute decision, defined uniformly over depth… See the full description on the dataset page: https://huggingface.co/datasets/Shanmuk4622/msc-cifar100.semasia-cifar100
Latents for cifar100 (timm)
This repository hosts precomputed latent representations (embeddings) extracted from timm image-classification backbones on cifar100, released as part of SEMASIA — a large-scale resource for studying semantic communication, cross-model latent space alignment, and explainability.
Each config corresponds to a single model;
only that model's Parquet files are read on load_dataset.
Usage
Load with datasets and… See the full description on the dataset page: https://huggingface.co/datasets/spaicom-lab/semasia-cifar100.cifar100-224-adv-selfgenmmmi-dag1-2modalities-cifar10E2AM_ConvNeXtV2_CIFAR10spotlight-cifar100-enrichment
Dataset Card for "spotlight-cifar100-enrichment"
More Information needed
CIFAR-10fsgld-cifar100n-reprosvae-freckles-4096-cifar10
SVAE Freckles 4096 — CIFAR-10 Omega Tokens
Precomputed spectral decomposition of CIFAR-10 through Freckles v41 (256×256), a frozen Spectral Variational Autoencoder trained exclusively on synthetic noise.
Each CIFAR-10 image is resized to 256×256, decomposed into 4096 patches (4×4 each), and passed through Freckles' encoder → SVD bottleneck. The 4 singular values per patch are stored as a (4, 64, 64) omega map — a 4-channel spatial representation of spectral energy.… See the full description on the dataset page: https://huggingface.co/datasets/AbstractPhil/svae-freckles-4096-cifar10.cifar-10E2AM_EffNetV2_CIFAR10cifar10-stats
CIFAR-10 CNN Layerwise Training Statistics
Dataset Description
This dataset contains layer-wise training statistics for a convolutional network trained on CIFAR-10, together with the corresponding test/acc.
Each row is one point at loss landscape. The features are computed on the last training batch of 1024 samples before the end of an epoch, and test/acc is measured immediately after that epoch.
The dataset includes statistics for several convolutional layers and the… See the full description on the dataset page: https://huggingface.co/datasets/Fullfix/cifar10-stats.CIFAR-100cifar10_TRADES_pgd_ema_ccm_False_ccr_False_detailed_resultsE2AM_MobileViTv2_CIFAR10E2AM_MobileViTv2_CIFAR100cifar10-speedrunE2AM_EffNetV2_CIFAR100cifar10_TRADES_pgd_ema_ccm_False_ccr_False_resultscifar10_vanilla_pgd_ema_ccm_False_ccr_False_detailed_resultscifar10_vanilla_pgd_ema_ccm_False_ccr_False_resultsE2AM_ConvNeXtV2_CIFAR100noisy-cifar10-reconstructionscifar10_tilescifar10 tiles - 4 parts
cifar10_3tilescifar100-captioned-train-balanced-probscifar100-captioned-train-mild_kl0216-probscifar100_florence_captionscifar10_florence_captionscifar10_4tiles