Nic Ma

Hi @vfdev-5 and @sdesrozis , Thanks for your discussion. I added `self.engine` to the metrics in this draft MONAI PR: https://github.com/Project-MONAI/MONAI/pull/1497. Will delete it in the future when you guys...

Hi @vfdev-5 , MONAI has a model-parallel tutorial: https://github.com/Project-MONAI/research-contributions/tree/master/lamp-automated-model-parallelism But I think it's not based on ignite workflow. Thanks.

Hi @vfdev-5 , I think @SachidanandAlle already provided the simple program in the above comment? Thanks.

Hi @SachidanandAlle , Could you please help share more information about the simple program to reproduce? Thanks in advance.

Hi @vfdev-5 , Thanks for raising this ticket. The idea is that: currently, MONAI post transforms(independent with ignite) execute on the `engine.state.output` dict, the post transforms can't get some useful...

Hi @vfdev-5 , May I know do you guys have any updates on this topic? We start to develop MONAI v0.6 now, making a clear data flow for all the...

Hi @vfdev-5 , In my simple mind, the easiest way for MONAI is to change this line: https://github.com/Project-MONAI/MONAI/blob/master/monai/engines/trainer.py#L160 into: ```py engine.state.output = dict(engine.state.batch) engine.state.output.update({Keys.PRED: preds, Keys.LOSS: loss}) ``` Then we...

Hi @vfdev-5 , Thanks for your confirm, let me move forward in MONAI with `engine.state.output = dict(engine.state.batch)` first. And I think your proposal of `engine.state.data` can't work for our case,...

Based on our online discussion with @vfdev-5 , we have a draft idea for dataflow: 1. **Engine** as the app level dict, contains all the components of the program: network,...

Hi @vfdev-5 , Do you have any updates about your ideas or plans for this feature request? Thanks in advance.