media Hugging Face Forums · 1h ago · src: 1d ago · open_models

Harmonic Frontier releases Celtic Constellation Reference Sessions dataset preview

from English

Harmonic Frontier has published a preview of the Celtic Constellation Reference Sessions dataset, featuring aligned multitrack recordings of traditional and contemporary Celtic ensemble arrangements. The architecture consists of aligned pairs where each piece includes dry isolated stems, the same performance processed through a production chain, and the finished mix.

The preview includes two full mixed tracks and short paired stem excerpts from Scottish Smallpipes, Highland Bagpipes, Irish Tin Whistle, Irish Low Whistle, Uilleann Pipes, and Bodhrán.
Technical specifications include 96 kHz capture, 24-bit WAV delivery, and articulation-level metadata.
The dataset is rights-cleared with full performer consent, sitting alongside the Human Vocality Primitives series.

The intent is for models to learn the relationship between raw source material and produced output rather than seeing one or the other in isolation.

Importance 1/3 Trust 1/3 Hugging Face Forums Training data

Read original

media Hugging Face Forums · 1d ago · 2 views

vldmrbesk releases 51k-sheet Tsiolkovsky archive with measured recognition accuracy

A complete personal archive of Konstantin Tsiolkovsky, comprising 51,008 sheets and 2,019 archival files, has been released as a CC0 dataset. The collection spans fifty years of work by the self-taught rocket theorist and includes handwritten manuscripts alongside typed copies.

media Hugging Face Forums · 6d ago · 6 views

Discussion on Parsewave and quality over scale in post-training data

The article discusses the diminishing returns of increasing post-training data volume as models improve, suggesting a shift toward higher-quality datasets. It highlights Parsewave, which generates post-training data from real-world engineering problems along with evaluation traces.

media Hugging Face Forums · 8d ago · 2 views

Young25 releases Humor_Vote, a Chinese humor dataset with human vote labels

User Young25 has released the Humor_Vote dataset on Hugging Face, providing a resource for evaluating and training models on Chinese humor. The dataset consists of native Chinese humorous questions paired with four candidate answers and corresponding human vote counts.

media Hugging Face Forums · 8d ago · 3 views

Harmonic Frontier releases Human Vocality Primitives preview set with 17 categories

Harmonic Frontier has published the full preview set for Human Vocality Primitives (HVP), a purpose-recorded dataset of non-lexical human vocal sounds. The project isolates breath, phonation modes, resonance behavior, and vocal gestures as primary subjects rather than incidental byproducts of speech or performance.

media Hugging Face Forums · 12d ago · 7 views

helloadhavan releases CC-FilteredCorpus to extract structured Markdown from Common Crawl

User helloadhavan has created a dataset called CC-FilteredCorpus designed to filter Common Crawl and extract content as clean Markdown. The project aims to preserve useful document structure, including headings, lists, links, code blocks, and tables, rather than flattening everything into plain text.