OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

BYT5 Geotagging

BYT5 Geotagging is an open-source Python repository containing a customizable ByT5-based model that predicts geographic coordinates from text, training code and datasets for geolocation-detection scenarios.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

BYT5 Geotagging — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you need a modifiable text-to-coordinate model and can prepare Python training and test inputs; its repository documentation focuses on model building and training rather than a hosted prediction service.

Who it is for

  • Developers building or customizing text-to-coordinate geotagging models.
  • Teams exploring geographic metadata for text search or content organization.
  • Researchers working with text, location and post-time relationships.

Documented uses

  • Estimate latitude and longitude, with confidence, from text.
  • Train and test customized geolocation-detection models.
  • Use geographic predictions to add location metadata to text collections.
  • Explore the supplied Regions and Seasons datasets for training or validation.
Before you choose

Requirements & limits

The GitHub repository is public and identifies the project as MIT-licensed.

  • The documented stack specifies Transformers 4.29.1, tqdm 4.63.2, pandas 1.4.4 and PyTorch 1.7.1.
  • The README reports 30 km median error on Haversine distance for the top 10% most relevant texts; performance for other data is not established by that reported figure.
  • The Regions dataset is described as 500,000 annotated texts covering 123 regions; the Seasons dataset is intended to examine relationships among post time, content and location.
  • The documented training command requires separate training and test input files; consult train_model.py for additional parameters.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Train a text-to-location model

Run the repository’s documented training and testing command with your own input files.

Before you start

  • Python environment with Transformers 4.29.1, tqdm 4.63.2, pandas 1.4.4 and PyTorch 1.7.1 installed.
  • Training and test input files.
  • The repository, including train_model.py and clustering resources.

Steps

  1. Set up the listed dependencies in your Python environment.

  2. Prepare training and test input files for the model.

  3. From the repository, run: python train_model.py --train_input_file --test_input_file --do_train true --do_test true --load_clustering .

  4. Consult train_model.py for additional available parameters.

Expected outcome

The command is configured to train and test a ByT5-encoder-based geotagging model.

Guide limitations

  • The README does not specify the input-file schema in the provided documentation.
  • The reported 30 km median Haversine error applies to the top 10% most relevant texts and is not a performance guarantee for custom data.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
Yachay-AI/byt5-geotagging · Official homepage title · retained in its source language
How we review sourcesSuggest a correction