Open algorithmdog opened 5 years ago
@algorithmdog good question. It would great to be able to use this on a Spark cluster. I hope someone answers or looks into this since it hasn't been looked at in 4 months.
Sorry for not responding earlier but I'm currently busy with multiple other projects and my graduation. In principle you should be able to implement a distributed trainer for your models relatively easily (especially if you're using data parallelism), but TF Scala does not provide any such pre-existing trainers.
I am using tensorflow_scala to train a model. The data is large and I want to use distributed training. Does tensorflow_scala support distributed training. If so, how can I get some documents and examples? Thanks in advance.