I missed giving --export_dir parameter while running training, so at the end DeepSpeech did not generated model. I could not figure out a way, so I ran for another epoch with the export_dir (but of course that takes a lot of time).
Given a checkpoint, what is the command to generate PB and tflite model?
Thanks,
+satish
lissyx
((slow to reply) [NOT PROVIDING SUPPORT])
2
The same command. Just don’t pass --train_files or dev ot test, and set --export_dir and if needed --export_tflite. This is all well documented, so please give feedback on what is unclear in the documentation for that.