Long-T5 Chapterization

An attempted implementation of the method used in podtail based on prechaptered youtube episodes. This model is fine-tuned from google/long-t5-tglobal-base on creator provided YouTube chapter annotation pairs.

Usage

This model is designed to segment long-form audio transcripts into semantically coherent chapters.

Methodology

Based on the approach discussed by the PodTile paper published by Spotify.

Downloads last month: 41

Safetensors

Model size

0.3B params

Tensor type

F32

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for milz/long-t5-chapterization

Base model

google/long-t5-tglobal-base

Finetuned

(37)

this model