datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
hq-20severyayah_curated_1s_20s_balancedeveryayah_curated_1s_20severyayah_curated_1s_20s_balanced_largeyodas-granary-it-neucodec-10s-20s
Granary Italian VoxPopuli NeuCodec
NeuCodec-tokenized rows for NeuTTS fine-tuning. The source rows were streamed from espnet/yodas-granary / Italian and uploaded as resumable Parquet shards.
{
"source_dataset": "espnet/yodas-granary",
"source_config": "Italian",
"source_splits": [
"ast"
],
"validation_source_split": null,
"codec_checkpoint": "neuphonic/neucodec",
"columns": [
"text",
"codes",
"duration",
"source_dataset",
"source_config"… See the full description on the dataset page: https://huggingface.co/datasets/Steveeeeeeen/yodas-granary-it-neucodec-10s-20s.everyayah_curated_1s_20s_balanced_mediumeveryayah_curated_1s_20s_balanced_tinysong_dataset_training_20s_cleanedsong_dataset_training_20salex-20sAlex-20s: contains ~1.7M general inorganic materials curated from the Alexandria database, with energy above the convex hull less than 0.1 eV/atom and no more than 20 Wyckoff sites in the conventional cell.
The Alex-20s dataset differs from Alex-20 in that it restricts structures to those containing no more than 20 Wyckoff sites in the conventional cell, while Alex-20 limits the total number of atoms to 20 in the unit cell.
memoni_20sec_segments
Memoni Audio Dataset - 20 Second Segments
Dataset Description
This dataset contains 3,058 audio segments derived from the original Aqiba/memoni_clean_audio dataset. Each audio segment is exactly 20 seconds long, making it compatible with the Sarvam AI REST API (which has a 30-second limit per request).
Dataset Summary
Source: YouTube videos (Memoni/Gujarati/Urdu content)
Original files: 371 audio files (avg. 82 seconds each)
Segments created: 3,058… See the full description on the dataset page: https://huggingface.co/datasets/Aqiba/memoni_20sec_segments.episode-0003-20sec
Episode_0003__20_seconds_with_movement_
LeRobot v2.1 format dataset for robot manipulation.
Dataset Structure
Episodes: 1 episodes of robot manipulation
Total Frames: 620 frames
Cameras: 3 camera views per episode
observation.images.base_camera_sensor_image_raw
observation.images.arm1_camera_sensor_image_raw
observation.images.arm2_camera_sensor_image_raw
Robot: Bimanual manipulator with 34 joints
Format: LeRobot v2.1
Usage with LeRobot
from… See the full description on the dataset page: https://huggingface.co/datasets/Sraghvi/episode-0003-20sec.Dataset10-batch2_seg_20s_doneiqra_curated_normalised_1s_20s_finalDataset5-batch1_part1_subpart1_seg_20sDataset13-batch1_seg_20s_donebench-traces-480B_g16_pp8_tp1_r2_ep8_c64-20sectimeoutterminal_bench_2_Qwen3_32B_SweSmith_20step_20260317_215519terminal_bench_2_Qwen3_32B_SweSmith_20step_20260317_223743sonnet_combined_all_traj_20stepterminal_bench_2_Qwen3_32B_SweSmith_20step_20260318_064735terminal_bench_2_Qwen3_32B_SweSmith_20step_20260318_023437terminal_bench_2_Qwen3_32B_SweSmith_20step_20260501_180019msmarco-item-id-hardneg-filter-20shot-v4bench-traces-480B_g16_pp8_tp1_r2_ep8_c32-20sectimeoutbench-traces-480B_g16_pp8_tp1_r2_ep8_c128-20sectimeoutterminal_bench_2_Qwen3_32B_SweSmith_20step_20260317_223400pubmed-test-memsum-20sentbench-traces-480B_g4_pp4_tp1_r1_ep8_c32-20sectimeoutbench-traces-480B_g8_pp4_tp1_r2_ep8_c32-20sectimeout