toodef neural-pipeline issues

toodef / neural-pipeline

Neural networks training pipeline based on PyTorch

https://neural-pipeline.readthedocs.io

MIT License

312 stars 24 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Dev

#34 toodef opened 4 years ago
0
Metrics computation is wrong?

#33 elch10 opened 5 years ago
1
Add DataProducerPredictor

#32 toodef opened 5 years ago
0
Let user choose a device by himself

#31 toodef opened 5 years ago
0
Improve metrics

#30 toodef opened 5 years ago
0
Add the ability to select a subpart from dataset

#29 toodef opened 5 years ago
0
Extract all console output to Console Monitor

#28 toodef opened 5 years ago
0
Typo in setup.py

#27 Balykovsky closed 5 years ago
1
StandartStage doesn't exists in train_config/__init__.py

#26 toodef opened 5 years ago
0
Add Comparable Trainer

#25 toodef opened 5 years ago
1
Disappearance of metrics_log when resuming

#24 Balykovsky opened 5 years ago
0
TensorboardMonitor is_continue flag

#23 Balykovsky opened 5 years ago
0
fix typo

#22 pfriesch closed 5 years ago
0
Add continue from best checkpoint to Predictor

#21 toodef opened 5 years ago
0
Add a guide about execution performance

#20 toodef opened 5 years ago
1
from neural_pipeline.train_config import TrainConfig for line 104

#19 cclauss closed 5 years ago
4
Add trainer with asynchronous metrics processing

#18 toodef opened 5 years ago
0
Add system monitor

#17 toodef opened 5 years ago
0
Add `turbojpeg` for images fast loading

#16 toodef opened 5 years ago
0
Add built-in metrics for classification and segmentation tasks

#15 toodef opened 5 years ago
0
Add FP16 support

#14 toodef opened 5 years ago
0
Add models constructor

#13 toodef opened 5 years ago
0
Add TTA to Predictor

#12 toodef opened 5 years ago
0
Add `jit` training

#11 toodef opened 5 years ago
0
Profile code and optimize as possible

#10 toodef opened 5 years ago
0
Add batch gradients accumulating (for increase batch without increasing memory usage)

#9 toodef opened 5 years ago
0
Move to PyTorch 1.0.1

#8 toodef closed 5 years ago
0
Add best practices guide to documentation

#7 toodef opened 5 years ago
0
Think about parallel losses computation. May be needed when losses computed on CPU.

#6 toodef opened 5 years ago
0
Add delayed losses computation for increase batch size without memory increasing.

#5 toodef opened 5 years ago
0
Add loss composer for stack losses with weights.

#4 toodef opened 5 years ago
0
Add example for multitask

#3 toodef opened 5 years ago
0
Experiments comparision

#2 toodef opened 5 years ago
0
Typo in docstrings

#1 Balykovsky closed 5 years ago
0