AI Training Datasets

Dataset · Speech recognition

DNU-ViSpeech

Trang Lê

DNU-ViSpeech is a Vietnamese scripted/read-speech corpus for automatic speech recognition (ASR). This release contains 31,070 utterances from 26 anonymized speakers, totaling 75.570 hours. Audio is WAV, 16 kHz, mono, uncompressed PCM 16-bit. The primary protocol is DNU-ASR-v1.1. Its evaluation split combines unseen-speaker and recording-configuration shift; those two factors cannot be isolated with the current data. The corpus supports research on Vietnamese ASR adaptation, especially whether compact models can approach larger zero-shot systems after domain fine-tuning on read speech. The repository follows the Hugging Face AudioFolder layout. Each split has its own metadata.csv, and…

Dataset

rocketleague-analysis

Edmund Miller

Local Rocket League replay analysis using Ballchasing API exports and plain DuckDB. The report is meant to answer one practical question: what should I work on next from my saved replay sample? Start with CONTRIBUTING.md before changing the pipeline. Replay files and player-level exports can contain private identifiers; follow docs/data-and-privacy.md before sharing data or running an upload workflow. - db/replayanalysis.duckdb: local DuckDB database - reports/replayteamanalysis.html: static report The DuckDB file and aggregate or benchmark playerframes.tsv tables are reproducible local artifacts and are intentionally ignored by Git. Approved session-local frame tables may be versioned as…

This project uses a 96 GB Blackwell GPU and image-editing models to preserve a pet's identity across portrait references. This release packages the existing Higgsfield template library; it does not enable bulk generation on the live store. 725 distinct reference concepts, all full-resolution PNGs at1792×2240, plus100 separately labeled vintage poster versions. Every master was decoded and SHA-256 matched against its source manifest. There are725 optimized browsing thumbnails. No customer originals, customer portraits, credentials, provider account URLs or private storage keys are included. - Open index.html locally for searchable previews and sex/category/collection filters. - catalog.json…

Dataset

prop-firm-discount-codes

Prop Firm Discount

As of 2026-10-04, the biggest verified prop firm discount code is PFD from Upcomers — 90% off. Every firm's code is listed below, newest deals first. Public dataset mirror of propfirmdiscount.com — verified prop firm discount codes, funding deals, and Trustpilot ratings. The source of truth is the live site; this repository is a read-only distribution channel, synced hourly. - Every firm with published deals has its own page: md/prop-firm/{firm-slug}.md — one file per firm, headed "{Firm} Discount Code", carrying the verified code, its discount, the current deal, and the deal's FAQ. - Seasonal deals (Black Friday, Christmas, Cyber Monday, …): md/tag/ — one page per season, newest offers…

Training rollouts of one verl run, one parquet per trainer step (data/step-.parquet), rows in the order the trainer consumed them; the rollout.n samples of a prompt are contiguous and share uid. prompttokenids / responsetokenids are the exact token ids (special tokens kept, no padding) and rolloutlogprobs the sampling-time log-prob of every response token. input / output are decoded with skipspecialtokens=True and are for reading only. score is the raw reward.

A continually refreshed registry of GitHub repositories across ML fields. The curated current view applies ml-contribution-v5 and seeks projects that present a distinct contribution to an ML model, method, or technique. Broad Search, topic, census, and paper-link discovery is retained as raw provenance; Search observations take precedence during current-view selection. The selector is a high-precision text heuristic, not verification. Self-description cannot establish actual novelty, correctness, reproducibility, or scientific quality. Its rules screen common forks, owner/profile repositories, coursework, resource collections, tutorial-only projects, and unrelated utilities from the curated…

Dataset · Other

shopee-dataset

Rebrowser.net

prettyname: "Shopee Product Listings Dataset" - shopee - e-commerce - southeast-asia - product-data - marketplace - pricing-data - open-data - data-science - web-scraping - scraper - crawler - data-collection sizecategories: taskcategories: - other - configname: product-details datafiles: Daily snapshots of Shopee product listings across Taiwan and Southeast Asian marketplaces with variant pricing, seller metrics, ratings, and shipping data. This dataset is a preview sample of the Shopee dataset published by Rebrowser. If you're doing academic research, you may be eligible for free access to a much larger slice — see Free Datasets for Research. This dataset contains 1 entity, each in its…

Aligned visualizations of tactile-glove human demonstrations (Noitom capture rig) for the tacWAM project. Each clip = 4 camera panels (head RGB / head depth / left & right wrist), both hands' tactile heatmaps, and a total-pressure timeline. English overlays, 15 fps. Shared QC sheet → https://docs.google.com/spreadsheets/d/1jdWoaNQhLIEqkpm8nlQoOuralDeuo69T5fyzmRIWO44/ As you eyeball the clips, record any issues (empty/absent objects, in-air motion, weak/dead tactile channels, choppy video, mislabeled task, alignment glitches, etc.) in the sheet. Use the sessionid (the.mp4 filename) as the row key so findings are traceable. Play any.mp4 inline by clicking it, or open its direct URL…