OpenNMT/OpenNMT-tf

Copy Attention feature is needed

Open

#95 opened on Apr 2, 2018

View on GitHub
 (3 comments) (0 reactions) (0 assignees)Python (1,428 stars) (394 forks)batch import
enhancementhelp wanted

Description

Opennmt-py has Copy Attention feature, which is very useful mechanism for many applications. Is there any plan to add this feature on Opennmt-tf? The link of copy attention in opennmt-py, http://opennmt.net/OpenNMT-py/onmt.modules.html?highlight=copy#copy-attention

Contributor guide

Tech stack
python
Domain
machine learningai
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
knowledge of attention mechanismsfamiliarity with TensorFlowunderstanding of OpenNMT tf codebase
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
20
Research direction
The issue requests adding a copy attention mechanism similar to OpenNMT py (reference: http://opennmt.net/OpenNMT py/onmt.modules.html?highlight=copy#copy attention). To implement, first study the copy attention implementation in OpenNMT py's source code, then explore OpenNMT tf's model structure (likely in the 'models' or 'layers' directory) to identify where to integrate. Check existing attention implementations in OpenNMT tf to understand the pattern. No linked PRs or assignees, so the feature is available for implementation.