Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Try Monotonic Attention

See original GitHub issue

I tried Monotonic attention and got better result (alignment is clear at step 10k). eval-125000-0

update to TensorFlow 1.3
modify models/tacotron.py BahdanauAttention(256, encoder_outputs) -> BahdanauMonotonicAttention(256, encoder_outputs)

eval-125k.zip

Issue Analytics

State:
Created 6 years ago
Reactions:5
Comments:16 (1 by maintainers)

Top GitHub Comments

1reaction

lifeitengcommented, Sep 15, 2017

results after 125k (This repo)

step-550k.zip step-883k.zip

multi-speaker results (my implementation)

read.zip

He has read the whole thing.
He reads books.

1reaction

lifeitengcommented, Aug 24, 2017

Yuxuan the first author of Tacotron said that they also use Monotonic Attention in their newest version. They showed paragraph synthesis(more than 400 chars).

Top Results From Across the Web

MONOTONIC CHUNKWISE ATTENTION - Colin Raffel

This mechanism has an efficient training-time algorithm and enjoys online and linear-time decoding at test time. We attempt to quantify the resulting speedup ......

Online and Linear-Time Attention by Enforcing Monotonic ...

we propose an end-to-end differentiable method for learning monotonic alignments which, at test time, enables computing attention online and in linear time.

Monotonic Chunkwise Attention - OpenReview

An online and linear-time attention mechanism that performs soft attention over adaptively-located chunks of the input sequence.

Tutorial on Attention-based Models (Part 2) - Karan Taneja

In this post, I'm going to discuss about various monotonic attention mechanisms. 3.2. Monotonic Alignments. Monotonic alignments are motivated ...

Deep Learning Text Corrector using Monotonic Attention (with ...

Seq2Seq model with monotonic Attention. ... Whereas in the attention mechanism, we try to mimic how humans process sequential information.