Ride

Fixed
- Device transfer in benchmark.

Added
- Defensive fallback for FLOPs measurement.
- Add MultiStepLR optimizers.

Changed
- Profiling to use `pytorch_benchmark` package.

Fixed
- WandB logger log_dir extraction.

Changed
- Profile only warms up on first inference.

Added
- Memory profiling.

Fixed
- Tune DeprecationWarning.

Added
- Add pred and target dict support in Lifecycle.

Fixed
- Avoid detaching loss in step.

Added
- Add preprocess_batch method to Lifecycle.
- Add option for string type in utils.name.
- Add Metric Selector.

Fixed
- Weight freezing during model loading.
- Fix discriminative_lr param selection for NoneType parameters.
- Fix wandb project naming during hparamsearch.
- Optimizer Schedulers take `accumulate_grad_batches` into account.

Changed
- Key debug statements while loading models to include both missing and unexpected keys.
- Bumped PL to version 1.4. Holding back on 1.5 due to Tune integration issues.
- Bumped Tune to version 1.8.