Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / toodef/neural-pipeline issues and pull requests
#33 - Metrics computation is wrong?
Issue -
State: open - Opened by elch10 over 5 years ago
- 1 comment
Labels: wontfix
#32 - Add DataProducerPredictor
Issue -
State: open - Opened by toodef over 5 years ago
#31 - Let user choose a device by himself
Issue -
State: open - Opened by toodef over 5 years ago
#30 - Improve metrics
Issue -
State: open - Opened by toodef over 5 years ago
#29 - Add the ability to select a subpart from dataset
Issue -
State: open - Opened by toodef over 5 years ago
Labels: good first issue
#28 - Extract all console output to Console Monitor
Issue -
State: open - Opened by toodef over 5 years ago
Labels: good first issue, feature
#27 - Typo in setup.py
Issue -
State: closed - Opened by Balykovsky over 5 years ago
- 1 comment
Labels: bug
#26 - StandartStage doesn't exists in train_config/__init__.py
Issue -
State: open - Opened by toodef over 5 years ago
Labels: bug
#25 - Add Comparable Trainer
Issue -
State: open - Opened by toodef over 5 years ago
- 1 comment
#24 - Disappearance of metrics_log when resuming
Issue -
State: open - Opened by Balykovsky over 5 years ago
Labels: bug
#23 - TensorboardMonitor is_continue flag
Issue -
State: open - Opened by Balykovsky over 5 years ago
Labels: bug
#22 - fix typo
Pull Request -
State: closed - Opened by pfriesch almost 6 years ago
#21 - Add continue from best checkpoint to Predictor
Issue -
State: open - Opened by toodef almost 6 years ago
#20 - Add a guide about execution performance
Issue -
State: open - Opened by toodef almost 6 years ago
- 1 comment
Labels: docs
#19 - from neural_pipeline.train_config import TrainConfig for line 104
Pull Request -
State: closed - Opened by cclauss almost 6 years ago
- 4 comments
Labels: bug, help wanted
#18 - Add trainer with asynchronous metrics processing
Issue -
State: open - Opened by toodef almost 6 years ago
#17 - Add system monitor
Issue -
State: open - Opened by toodef almost 6 years ago
#16 - Add `turbojpeg` for images fast loading
Issue -
State: open - Opened by toodef almost 6 years ago
#15 - Add built-in metrics for classification and segmentation tasks
Issue -
State: open - Opened by toodef almost 6 years ago
#14 - Add FP16 support
Issue -
State: open - Opened by toodef almost 6 years ago
#13 - Add models constructor
Issue -
State: open - Opened by toodef almost 6 years ago
#12 - Add TTA to Predictor
Issue -
State: open - Opened by toodef almost 6 years ago
Labels: good first issue
#11 - Add `jit` training
Issue -
State: open - Opened by toodef almost 6 years ago
#10 - Profile code and optimize as possible
Issue -
State: open - Opened by toodef almost 6 years ago
#9 - Add batch gradients accumulating (for increase batch without increasing memory usage)
Issue -
State: open - Opened by toodef almost 6 years ago
Labels: good first issue
#8 - Move to PyTorch 1.0.1
Issue -
State: closed - Opened by toodef almost 6 years ago
#7 - Add best practices guide to documentation
Issue -
State: open - Opened by toodef almost 6 years ago
Labels: docs
#6 - Think about parallel losses computation. May be needed when losses computed on CPU.
Issue -
State: open - Opened by toodef almost 6 years ago
#5 - Add delayed losses computation for increase batch size without memory increasing.
Issue -
State: open - Opened by toodef almost 6 years ago
#4 - Add loss composer for stack losses with weights.
Issue -
State: open - Opened by toodef almost 6 years ago
#3 - Add example for multitask
Issue -
State: open - Opened by toodef almost 6 years ago
#2 - Experiments comparision
Issue -
State: open - Opened by toodef almost 6 years ago
#1 - Typo in docstrings
Issue -
State: closed - Opened by Balykovsky almost 6 years ago
Labels: docs