unit8co/darts

Retrieve last hidden state for RNNModel and BlockRNNModel

Open

#1,649 opened on Mar 16, 2023

View on GitHub
 (2 comments) (1 reaction) (0 assignees)Python (6,832 stars) (762 forks)batch import
good first issueimprovement

Description

Is your feature request related to a current problem? Please describe. The last hidden state is an encoding of the information contained in the time series and is useful to retrieve in some projects.

Describe proposed solution Add a method that retrieve the last hidden state. Or maybe there is already an easy way to retrieve it? I couldn't find it.

Contributor guide

Tech stack
pythonpytorch
Domain
machine learning
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
half day
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
needs maintainer response
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
PyTorch RNN basicsdarts model structure
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
40
Research direction
The issue asks to add a method to retrieve the last hidden state from RNNModel and BlockRNNModel. Start by examining the source code for these models in darts/models/forecasting/ directory, particularly the forward pass output. Check if there is already a way to access hidden states (e.g., via return state parameter). Look at similar implementations in PyTorch Lightning or other codebases. Since there are no comments from maintainers, propose a design in the issue before coding.