74 lo-fi SVD positional embeddings from metal lyrics

A matrix of 74 positional vectors (rows) and 20 dimensions (columns). Positional embeddings derived from truncated SVD on a position-distance-weighted term-co-occurrence matrix (TCM) built from the metal lyrics corpus. Each row represents a sentence-internal position (1 through 74), capturing how co-occurrence patterns vary by word position within a sentence.

Format

A matrix of 74 rows and 20 columns

Source

Derived from the metal lyrics corpus in text2map.corpora

Examples

if (FALSE) { # \dontrun{

## download the model (once per machine)
download_pretrained("vecs_svd20_metal_position")

## load the model each session
wv <- load_pretrained("vecs_svd20_metal_position")

## check dims
dim(wv) == c(74, 20)

} # }