progress
Function
Executes a step in the training pipeline, including forward propagation, backward propagation, and parameter update.
Prototype
1 | def progress(self, dataloader_iter: Iterator[In]) -> Out: |
Parameters
Parameter |
Data Type |
Mandatory/Optional |
Description |
|---|---|---|---|
dataloader_iter |
Iterator[In] |
Mandatory |
Data loader iterator. |
Return Value
- Success: The training output result is returned.
- Failure: StopIteration or RuntimeError is thrown.
Parent topic: EmbCacheTrainPipelineSparseDist