datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
PhysicalAI-NuRec-PPISP
PPISP Dataset
Dataset Description:
The PPISP dataset accompanies the work "PPISP: Physically-Plausible Compensation and Control of Photometric Variations in Radiance Field Reconstruction". It contains object-centric scene captures of four outdoor scenes, each captured with three different cameras, for multi-view 3D reconstruction and novel view synthesis. The photos were captured with exposure bracketing of +/-2 EV and re-processed with automatic exposure and color… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-NuRec-PPISP.ppi_SHS148k_bfs_2025ppi_SHS148k_dfs_2025artur_studio_ttsTTS slovenian dataset, contains 40 hours of studio recording of a single speaker.
created from:
Verdonik, Darinka; et al., 2023,
ASR database ARTUR 1.0 (audio), Slovenian language resource repository CLARIN.SI, ISSN 2820-4042,
http://hdl.handle.net/11356/1776.
only studio recordings of speaker G0911
recordings without transcriptions were removed
resampled to 22050Hz 16bit wav
metadata.txt contains the transcriptions in format FILENAME_WITHOUT_EXTENSION|SPEAKER_NAME|TRANSCRIPTION
some… See the full description on the dataset page: https://huggingface.co/datasets/ppisljar/artur_studio_tts.ppi_SHS27k_dfs_2025sloleks-3-sqlSloleks 3.0 database converted to sqlite format for easier consumption
github repo with scripts and more information: https://github.com/ppisljar/sloleks-3-parser
Citations
Čibej, Jaka; et al., 2022,
Morphological lexicon Sloleks 3.0, Slovenian language resource repository CLARIN.SI, ISSN 2820-4042,
http://hdl.handle.net/11356/1745.
ppisp-ablation
PPISP Scene→Sim Ablation — OFF vs ON
Evaluation renders from a single controlled ablation of NVIDIA's PPISP inside
Forenly AI's Scene→Sim (real-scene → simulation) pipeline.
What this is
The same real 3-camera capture reconstructed with the same 30,000 training
iterations, changing exactly one variable — PPISP off vs on.
Run
Result
PPISP off
hazy, washed-out blur; geometry and colour collapse
PPISP on
sharp, colour-correct, stable — +9.5 dB PSNR… See the full description on the dataset page: https://huggingface.co/datasets/ForenlyAI/ppisp-ablation.ppi_SHS27k_random_2025ppi_STRING_dfs_2025ppi_SHS148k_random_2025ppi_STRING_bfs_2025ppi_SHS27k_bfs_2025pp-istio15021g2p_si_1Grapheme-to-Phoneme training, validation and test setsppi_STRING_random_2025testppi-sokoban-v1