Keisuke OGAKI

Results 4 comments of Keisuke OGAKI

Yes. I used about 0.2M illustrations accumulated from web to train the model.

Cool feature :) Mostly I agree to this proposal. > It may violate the design philosophy of gokart and luigi to call dump() only once at the end of run()....

I think we only need to inject `workspace_directory` of final task. As following ``` class TaskA(gokart.TaskOnKart): def run(self): print(self.output().path()) self.dump("") class TaskB(TaskA): def requires(self): return TaskA() with tempfile.TemporaryDirectory() as d:...

I have trained my model with about 1M images with 96x96 resolution. So I could only raise my model to stage 12. I haven't use any augmentation.