Light

jreremy / conformer Goto Github PK

View Code? Open in Web Editor NEW

21.0 1.0 4.0 24 KB

Pytorch implementation of conformer with with training script for end-to-end speech recognition on the LibriSpeech dataset.

License: Apache License 2.0

Python 100.00%

conformer pytorch librispeech librispeech-dataset machine-learning speech-recognition asr

conformer's Introduction

Pytorch Conformer

Pytorch implementation of conformer model with training script for end-to-end speech recognition on the LibriSpeech dataset.

Usage

Train model from scratch:

python train.py --data_dir=./data --train_set=train-clean-100 --test_set=test_clean --checkpoint_path=model_best.pt

Resume training from checkpoint

python train.py --load_checkpoint --checkpoint_path=model_best.pt

Train with mixed precision:

python train.py --use_amp

For a full list of command line arguments, run python train.py --help. Smart batching is used by default but may need to be disabled for larger datasets. For valid train_set and test_set values, see torchaudio's LibriSpeech dataset. The model parameters default to the Conformer (S) configuration. For the Conformer (M) and Conformer (L) models, refer to the table below:

Other Implementations

TODO:

Language Model (LM) implementation
Multi-GPU support
Support for full LibriSpeech960h train set
Support for other decoders (ie: transformer decoder, etc.)

conformer's People

Stargazers

Watchers

Forkers

eric-debug666 niehen6174 titiaffandi ramsrahim

conformer's Issues

Can you reproduce experiment results?

Hey, I am trying to reproduce conformer from scratch these days. But my model seems to be converged too fast. Maybe you find same problem I got. CTCLoss will converge to about 7.x during first epoch. Can you give me some advice?

How use with Beam search decoder

Hi, I am trying to use Conformer with ctc beam search decoder from https://github.com/parlance/ctcdecode. I have created 1000 subword tokens as classes using sentencepiece. Now whenever I use your code it runs smoothly for some iteration then suddenly gets killed and stops training. I use tiny batch size and model size and continuously monitor the GPU memory usage. Gpu use is always under half of the capacity of the GPU memory (using A5000 with 24 gb memory but training only using 7 gb). what could be the issue?

Thanks in advance.

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.