Deep learningyou can see inside.

# Submit a graph to the server's queue.
# The run keeps going after the terminal closes.
$ cdui run examples/Usage_Example/CNN-MNIST/TrainCNN-MNIST/graph.json

# Run one graph over a grid of parameter values.
$ curl -X POST http://127.0.0.1:8000/api/sweeps \
    -H "X-CodefyUI-Token: $TOKEN" \
    --data "@sweep.json"

# A published graph answers HTTP like any other API.
$ curl -s -X POST "http://127.0.0.1:8000/api/apps/classifier/invoke" \
    -H "Authorization: Bearer cdui_YOUR_KEY" \
    -H "Content-Type: application/json" \
    --data "@payload.json"

# payload.json
{"inputs": {"x": "hello"}}

# A project directory is plain files, so it is plain git.
$ cdui project init my-project
$ cdui start --project my-project

# In the editor: Source Control tab
#   stage · commit · branch · stash · pull · push
$ git log --oneline
a41c2e0 Lower the learning rate to 3e-4
9be07d1 Add RandAugment before the DataLoader

# ResNet-18 on CIFAR-10, from the examples folder.
$ cdui run examples/Usage_Example/ResNet18-CIFAR10-Baseline/graph.json

# Top-1 test accuracy        95.48%
# Same-seed re-run           95.48%
# Recorded metric points     601 of 601 bitwise identical