|
Quick mode
|
Use default param settings
|
|
Update mode
|
Update default param settings
|
|
Analysis mode
|
Find hyper params
|
|
Expert mode
|
Set params from scratch
|
|
Compare mode
|
Compare against output metrics
|
|
Pytorch
|
v1.2.0, v1.3.0
|
|
Keras
|
v2.2.5, v2.3.1
|
|
GluonCV
|
v0.5.0
|
|
Mxnet
|
v1.5.1
|
|
Cuda - Support
|
v9.0, v10.0
|
|
Ubuntu
|
v16.04, v18.04
|
|
Mac-OS
|
-
|
|
Windows
|
v10
|
|
Python
|
v3.5, v3.6, v3.7
|
|
Write syntax invariant code
|
|
Resume interrupted experiments
|
|
Move workspaces across systems and continue training/inferencing
|
|
Compare experiments visually
|
|
Auto find hyper params
|
|
Run training/inferencing on sub-sample of original datasets
|
$ git clone https://github.com/Tessellate-Imaging/monk_v1
$ pip install -r monk_v1/installation/requirements_cu10.txt
import sys
sys.path.append("monk_v1/monk")from pytorch_prototype import prototypefrom gluon_prototype import prototypefrom keras_prototype import prototypeptf = prototype(verbose=1);
ptf.Prototype("sample-project-1", "sample-experiment-1");
ptf = prototype(verbose=1);
ptf.Prototype("sample-project-1", "sample-experiment-2");ptf.Default(dataset_path="train",
model_name="resnet18",
num_epochs=2);
ptf.Train();out = ptf.Infer(img_name="path_to_img");
out = ptf.Infer(img_dir="path_to_dataset");
No comments yet.
Cheatography n. The art of mapping complex subjects into clear, instantly navigable reference guides.