Jijoong Moon
Jijoong Moon
Thanks for sharing the RFC. But I wonder if it is possible to change the model architecture on device at runtime. The first thing about the on-device training we can...
Thanks for your PR. But We recommend to write PR titles and comments to describe what changes included more clear and concise manner ( for example #1547 ). And also...
**TODO List** - [x] set the `dynamic_time_sequence=true` Property for the `addWithReferenceLayers` : #1935 - [ ] set the `Start` and `End Layer` in Recurrent Realizer - [ ] Implement Counting...
> For the backward operation, How can we know that it's _skip_ layer? For example, in diagram 'Option-II', _Layer 3/2_ should be skiped, but it seems that we can know...
This is great work! Using this swapping, we can save much more memory now. We can train the Resnet like model under 100 MB memory!!!
AFAIK, we will decide when ths tensors are off-loaded with tensor life span and execution order.
Tizen 7.0 M2 Release ## Major API Updates 1. Layers - positional encoding layer - multi-head attention layer - layer normalization - mol attention layer - attention layer 2. Tizen...
It could be done in later pr though, but I think you could also add the tensor name since nntrainer tensor has it's own name. And it might help to...
> > :octocat: **cibot**: @mhs4670go, **nntrainer/utils/node_exporter.cpp** does not include Doxygen tags such as **@file @brief @author @bug**. You must include the Doxygen tags in the source code. Please refer to...
We delay this due to the priority.