Skip to main content

Multiple CRF implementation for PyTorch

Project description

Masked CRF

NOT Official Pytorch implemented Masked CRF.

Installation

Dependencies

  • Python >= 3.6
    • torch == 1.5.1 (better > 1.0)
    • tqdm == 4.53.0
    • pyyaml == 5.3.1

Download and Install

  1. Editable installation:
pip install -e .
  1. Install from PyPI:
pip install pytorch-mcrf -i https://pypi.org/simple

Quick Start

Settings

All the settings are in config.yaml, you can change model settings from this file.

Run

python run.py

Results

Micro-F1 results

WeiboNER

Batch Size Optimizer Learning Rate Max Seqence Length
8 SGD 0.015 128
Method Dev Test #Illegal Tags
PlainCRF 57.564 51.733 4
MaskedCRF ( decoding only ) 55.662 51.351 0
MaskedCRF 55.380 50.287 0

MSRA

Batch Size Optimizer Learning Rate Max Seqence Length
8 SGD 0.015 250

Model is selected directly from test set since there is no official dev set.

Method Test #Illegal Tags
PlainCRF 86.274 1
MaskedCRF ( decoding only ) 86.285 0
MaskedCRF 84.714 0

Acknowledgements

UPDATES

  • v0.0.2: fix setuptools packages finding issue

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pytorch-mcrf-0.0.3.tar.gz (21.0 kB view details)

Uploaded Source

Built Distribution

pytorch_mcrf-0.0.3-py3-none-any.whl (28.9 kB view details)

Uploaded Python 3

File details

Details for the file pytorch-mcrf-0.0.3.tar.gz.

File metadata

  • Download URL: pytorch-mcrf-0.0.3.tar.gz
  • Upload date:
  • Size: 21.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/50.3.2 requests-toolbelt/0.9.1 tqdm/4.53.0 CPython/3.7.7

File hashes

Hashes for pytorch-mcrf-0.0.3.tar.gz
Algorithm Hash digest
SHA256 7405ea38737b5dfa783f7895e99266129430db5caa4bee5db1f27e837c50a0fa
MD5 77f365743cd5f7b933fef0473dc2cd61
BLAKE2b-256 39812849d48596d7144bb546b675d8f4cba570423455cdbf39746fff86a248e1

See more details on using hashes here.

File details

Details for the file pytorch_mcrf-0.0.3-py3-none-any.whl.

File metadata

  • Download URL: pytorch_mcrf-0.0.3-py3-none-any.whl
  • Upload date:
  • Size: 28.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/50.3.2 requests-toolbelt/0.9.1 tqdm/4.53.0 CPython/3.7.7

File hashes

Hashes for pytorch_mcrf-0.0.3-py3-none-any.whl
Algorithm Hash digest
SHA256 2451e53e6b9216b231f52f68778d16c6bda1b64e06e23a5fa7e9a594e77dbfcc
MD5 e9afda1b4548f362b4dce464b95f5a57
BLAKE2b-256 9db6c44bc6e47d01f9696074d003c0ccc486b9ace94f8e27b0d69b2685b0aa67

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page