Documentation

System Constraints and Limitations

  • This is a specific guide for TensorFlow 2.6.5. For TensorFlow 1.15 model porting instructions, see TensorFlow 1.15 Model Porting Guide.
  • Data types float64, complex64, complex128, and DT_VARIANT are not supported.
  • Operations related to tf.Variable must be performed on the NPU.
  • Function operators (tf.function) must be executed on the NPU.
  • The tf.compat.v1 API must not be used together with eager APIs in TensorFlow 2.6.5.
  • The TensorFlow 2.6.5 data preprocessing is performed on the host by default. However, variables need to be offloaded to the device for initialization. As such, when variables are used for data preprocessing in the TensorFlow 2.6.5 training script, the NPU training fails. In such case, embed the logic in context.device('CPU:0') and initialize variables for preprocessing on the host.
Parent topic: Before You Start