pytorch/serve

Having two different endpoints for a single model

Open

#2,058 opened on Jan 5, 2023

View on GitHub
 (2 comments) (0 reactions) (0 assignees)Java (3,844 stars) (790 forks)batch import
enhancementhelp wantedtriaged

Description

Hello!

Is it possible to create two different endpoints for a single model? So it can work in two separate modes and batch accordingly depending on the input (e.g., it can generate an image or perform upscaling given an input). I can make two different models but then the memory will be double, while actually no extra memory is needed (as the model weights are the same). Ideally, I want to have the same model weights but two different endpoints with different batching settings.

Best, Valentin

Contributor guide

Tech stack
pythonpytorchjava
Domain
backendapi
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.
1-2 days
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
TorchServe basicsModel serving conceptsPython/Java knowledge
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30
Research direction
Investigate TorchServe's model registration and endpoint configuration in files like 'frontend/model server.py' or 'backend/...'. Look for how batching is tied to endpoints. The issue may require creating a configuration to allow multiple endpoints per model with separate batching parameters. Links to relevant code should be examined.