ignite
ignite copied to clipboard
pytorch
High-level library to help with training and evaluating neural networks in PyTorch flexibly and transparently.
## 🚀 Feature The idea is to add ```python if torch.cuda.is_available() and torch.backends.cudnn.enabled: torch.backends.cudnn.benchmark = True ``` into native distributed model and clearly state about that in the docs. Once...
Fixes #1089 Description: - This PR improves the API for `Accuracy` according to the suggestions from the issue. Still WIP, to be done: * [x] Support custom binarization threshold. *...
Fixes #1521 Description: - bug fixes - state_dict/load_state_dict refactor - added helper `Engine.debug()` method - more tests Check list: * [x] New tests are added (if a new feature is...
## 🚀 Feature As discussed https://github.com/pytorch/ignite/pull/1919#issuecomment-821959142, some specific features are hardcoded in the `Engine` (reset of batch/output for instance). Usually it is internal concerns and therefore well encapsulated in the...
## 🚀 Feature The first intent of this FR is to address multiple downloading of datasets. The idea is to provide a way to schedule process using context manager. The...
- https://github.com/pytorch/ignite/pull/1878#issuecomment-808041919 cc @trsvchn @ydcjeff
Fixes #1728 Description: This is a WIP to have a discussion about the support of CCL backend, this is my first backend implementation so I'm expecting it will need couple...
## 📚 Documentation Currently we have imported the class in the handlers and metrics docs with `module.class_name` and `class_name`. [handlers docs](https://pytorch.org/ignite/master/handlers.html) [[source](https://github.com/pytorch/ignite/blob/master/docs/source/handlers.rst)] and [metrics docs](https://pytorch.org/ignite/master/metrics.html#complete-list-of-metrics) [[source](https://github.com/pytorch/ignite/blob/master/docs/source/metrics.rst)] This seems inconsistent. The...
## 🚀 Feature Following https://github.com/pytorch/ignite/issues/1709, Ignite is not supporting other modern types of parallelism. The idea is to introduce the following API to make `ignite.distributed` aware of data parallel group:...
## ❓ Questions/Help/Support This is a somewhat general question, but I'd love a detailed response. When wanting to go beyond standard data-parallel training towards hybrid data+model-parallel training (like Megatron-LM), what...
