sfa_pool: Fetch or Build the Candidate Pool for a Naming Model
In semanticfa: Semantic Factor Analysis of Language Model Embeddings

sfa_pool

R Documentation

Fetch or Build the Candidate Pool for a Naming Model

Description

Returns the candidate pool used by sfa_name(): the pre-filtered word list (369,703 label-eligible terms with precomputed word-family and dictionary-membership columns) together with its embedding matrix under the given model. Pre-generated pools are downloaded once and cached; for models without a pre-generated pool the word list is embedded locally (slow without a GPU) and cached thereafter.

Usage

sfa_pool(
  model,
  precision = c("int8", "fp16"),
  download = interactive(),
  build = interactive(),
  dir = NULL
)

Arguments

`model`	Embedding model id (as used by the `sbert` backend).
`precision`	`"int8"` (default) or `"fp16"`. int8 pools are half the download and reproduce the fp16 labels on all but 3 of 75 benchmark factors (each a weak factor relabeled with a near-synonym; the diff is listed in the package documentation). Use `"fp16"` for exact parity with the published research pipeline. Locally built pools are always fp16.
`download`	Permission to download missing artifacts. Defaults to `interactive()`; in non-interactive sessions pass `TRUE` explicitly (CRAN policy: no silent large downloads).
`build`	Permission to embed the word list locally when no pre-generated pool exists for `model`. Defaults to `interactive()`.
`dir`	Cache directory override (mainly for tests).

Value

An object of class sfa_pool: a list with words (data.frame: word, family, tier1), emb (matrix-like, one row per word, memory-mapped when read from disk), dim, model, precision.