-
## Motivation
codecov is used to measure the unit test code coverage. However, the ChainerMN tests are seemingly not accounted. See https://github.com/optuna/optuna/pull/2027#discussion_r526573203.…
-
Tensorflow paper (http://arxiv.org/pdf/1605.08695v1.pdf) states that Tensorflow supports multiple communication protocols such as gRPC over TCP, and RDMA over Converged Ethernet. Current repo, on the …
-
## Motivation
It is useful for TensorFlow users if we have `DistributedTensorFlowStudy` and `DistributedTensorFlowTrial` like [`ChainerMNStudy`](https://github.com/optuna/optuna/blob/0b152e1774976e…
-
## Motivation
In many examples, `pruned_trials` and `complete_trials` are obtained by filtering `study.trials`. `study.get_trials` with `states` argument introduced by #1943 will simplify the impleme…
-
This is inspired by #40690 and other requests we saw before. Applications might want to use c10d operations (e.g., `all_gather`, `all_reduce`) in the forward pass and expect they been linked in the sa…
-
This is a work log of a "OSS Gate workshop".
"OSS Gate workshop" is an activity to increase OSS developers.
Here's been discussed in Japanese. Thanks.
## OSS Gateワークショップ関連情報
* [スライド](https:/…
-
Python version: 3.7
OS/Platform: Red Hat 7.6
I am unable to run the chainermn dual parallel example (chainer/examples/chainermn/mnist/train_mnist_dual_parallel.py) with chainer version `7.1.0`. I …
-
Sphinx (`2.2.1` or `master`) produces the following two kinds of warnings in my environment.
### `duplicate object description`
I think cross refs to such object is ambiguous.
### `autosummar…
-
After dropping Python 2.7 support at #710, we can define type hints with function annotation syntax.
~~Do you have a plan to update the coding style guideline?~~
https://github.com/optuna/optuna/wi…
-
When I ran a training of https://github.com/msakai/chainer-formulanet/tree/4389aa09801c739c35a9ae2ceadd18fdbc74d5a5 using ChainerX and ChainerMN, I encountered `chainerx.ChainerxError: cudaErrorIllega…