search

tensorflow / tflite-micro

Infrastructure to enable deployment of ML models to low-power resource-constrained embedded targets (including microcontrollers and digital signal processors).
Apache License 2.0
1.74k stars 768 forks source link

Support for rolled LSTM #2600

Open zixiaoli0206 opened 1 month ago

zixiaoli0206 commented 1 month ago

Dear contributors,

I am trying to deploy a LSTM-based network on STM32. However, I noticed that the deployment is only successful when I used unrolled LSTM. When I use the rolled LSTM, the Allocator always complains "failed to get the registration from op code CUSTOM". Is it true the framework cannot support rolled LSTM up to now? The unrolled LSTM taks much more storage than rolled LSTM.

rascani commented 1 month ago

From your description, I do not know what type of LSTM you're trying to deploy, but currently, the only LSTM we support is a unidirectional sequence LSTM.

github-actions[bot] commented 4 days ago

"This issue is being marked as stale due to inactivity. Remove label or comment to prevent closure in 5 days."