https://github.com/diyclassics/latincy-w2v-eval
https://github.com/diyclassics/latincy-w2v-eval
Last synced: 5 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/diyclassics/latincy-w2v-eval
- Owner: diyclassics
- Created: 2025-07-30T15:49:06.000Z (12 months ago)
- Default Branch: main
- Last Pushed: 2025-07-30T21:24:47.000Z (12 months ago)
- Last Synced: 2025-07-31T00:15:06.420Z (12 months ago)
- Size: 22.5 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# LatinCy word2vec Evaluation Datasets
Developed by [Patrick J. Burns](https://diyclassics.github.io/), William Bramwell [last updated 2025-07-31]
## Overview
This repository contains datasets for evaluating Latin word2vec models, and in particular, as a development benchmark for the [LatinCy word2vec](https://huggingface.co/latincy/latincy-w2v) (and related) word embeddings models. The LatinCy word2vec models are published here: https://huggingface.co/latincy/latincy-w2v.
The basic work of identifying categories useful for evaluation as well as bootstrapping the word lists per category was completed done by William Bramwell in summer 2024 as part of an internship in the [Institute for the Study of the Ancient World Library](https://isaw.nyu.edu/library). Patrick J. Burns wrote code to extend the datasets and to perform the evaluations.
The following datasets are available:
- [LatinCy word2vec Analogies Dataset (v0.1)](https://github.com/diyclassics/latincy-w2v-eval/blob/main/data/evaluation-datasets/latin_analogies_generated.txt)
- [LatinCy word2vec Odd-one-out Dataset (v0.1)](https://github.com/diyclassics/latincy-w2v-eval/blob/main/data/evaluation-datasets/latin_oddoneout_generated.tsv)
## Report
A report on the LatinCy word2vec Evaluation Datasets is available [here](https://github.com/diyclassics/latincy-w2v-eval/blob/main/reports/burns-2025-latincy-w2v-evaluation-datasets-report.pdf).