# Retraining with new data
**URL:** https://discuss.flower.ai/t/retraining-with-new-data/1193
**Category:** Flower Help - Intermediate
**Created:** [January 15, 2026, 11:58am UTC](https://discuss.flower.ai/t/retraining-with-new-data/1193 "2026-01-15T11:58:46Z")
**Posts on this page:** 3
**Page:** 1
### Author:  [@neonduck](https://discuss.flower.ai/u/neonduck)
#### Post date: [January 15, 2026, 11:58am UTC](https://discuss.flower.ai/t/retraining-with-new-data/1193/1 "2026-01-15T11:58:46Z")
Hi,
I’m currently implementing an autoencoder with PyTorch (client with Libtorch in C++) and I want to ask if there is an example or some guide on how to retrain model using Flower with new data to improve the prediction performance.
---
### Author:  [@pan-h](https://discuss.flower.ai/u/pan-h)
#### Post date: [January 27, 2026, 11:01am UTC](https://discuss.flower.ai/t/retraining-with-new-data/1193/2 "2026-01-27T11:01:24Z")
Hi @neonduck,
Flower doesn’t have a special “retrain” mode. Retraining with new data is done by starting a new run and initializing it with the previously trained global model.
---
### Author:  [@system](https://discuss.flower.ai/u/system)
#### Post date: [March 24, 2026, 11:45am UTC](https://discuss.flower.ai/t/retraining-with-new-data/1193/3 "2026-03-24T11:45:18Z")
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.