pytorch/examples

A question about weight initialization of embedding layer.

Open

#595 opened on Jul 18, 2019

View on GitHub
 (1 comment) (0 reactions) (0 assignees)Python (9,429 forks)batch import
good first issue

Repository metrics

Stars
 (21,634 stars)
PR merge metrics
 (No merged PRs in 30d)

Description

I have a question about the weight intialization of embedding layer.

In source code of PyTorch, the weight of embedding layer is initialized by N(0, 1). In this code, the weight of embedding layer is intialized by uniform.

I trained the model two times with different initialization method, and I found that the default initialization makes the model converges too slow.

So, why the embedding is initialized by N(0, 1), which seems not a good start point?

Contributor guide

Research direction
Investigate the rationale behind using normal distribution N(0,1) for embedding weight initialization in PyTorch vs uniform initialization used in examples; trace relevant discussions and literature.
Tech stack
pythonpytorch
Domain
machine learning
Issue type
Research
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
2
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Clear
Prerequisites
PyTorch
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
75