While submitting I am constantly getting:
WARNING: Your kernel does not support swap limit capabilities or the cgroup is not mounted. Memory limited without swap.
* daemon not running; starting now at tcp:5037
* daemon started successfully
INFO: Initialized TensorFlow Lite runtime.
ERROR: Encountered unresolved custom op: IdentityN.
ERROR: Node number 41 (IdentityN) failed to prepare.
Failed to apply NeuronDelegate delegate.
Benchmarking failed.
Traceback (most recent call last):
File "/tmp/codalab/tmptT034M/run/program/evaluation.py", line 173, in
file = open(LOG_NAME, 'r')
IOError: [Errno 2] No such file or directory: '/tmp/codalab/tmptT034M/run/output/output.csv'
The TFLite model was generated using TF 1.15.0 runtime and without the experimental converter by setting `converter.experimental_new_converter = False`. Could you folks please provide some information on this?
Posted by: sayak @ Feb. 18, 2021, 12:57 p.m.Hi Sayak,
tf.identity is supported in TFLite. However, the tf.identity_n you reported here is not supported by TFLite.
Please find bellow for the ops compatibility of TFLite. Thank you.
https://www.tensorflow.org/lite/guide/ops_compatibility
Is it when we try running inference? Because the conversion works perfectly fine.
Posted by: sayak @ Feb. 19, 2021, 9:29 a.m.