budzianowski/PyTorch-Beam-Search-Decoding

PyTorch implementation of beam search decoding for seq2seq models

47
/ 100
Emerging

This tool helps machine learning engineers and researchers generate high-quality text sequences from sequence-to-sequence models. You input a trained model and an initial sequence, and it outputs a more refined, contextually appropriate sequence of text. This is designed for those working with natural language generation tasks.

338 stars. No commits in the last 6 months.

Use this if you need to improve the output quality of your PyTorch sequence-to-sequence models by exploring multiple candidate sequences.

Not ideal if you are looking for a general-purpose natural language processing library rather than a specific decoding strategy.

natural-language-generation sequence-to-sequence machine-translation text-summarization AI-model-decoding
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 21 / 25

How are scores calculated?

Stars

338

Forks

63

Language

Python

License

Apache-2.0

Last pushed

Jan 20, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/nlp/budzianowski/PyTorch-Beam-Search-Decoding"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.