Pad dynamically when preparing the dataset for batch size autotuning

OpenNMT / OpenNMT-tf

Neural machine translation and sequence learning using TensorFlow

https://opennmt.net/

MIT License

1.46k stars 392 forks source link

Closed guillaumekln closed 2 years ago

guillaumekln commented 2 years ago

We want the length dimension to be unknown when building the graph.