NN
Simulator
Research
Design
Analyze
Diagram
Compare
⟳ Reset
▶ Train
Layers
Baselines
Custom
Custom Layer
Name
Description
Color
Params (name:type,…)
Forward stub
return self.op(x)
+ Register
No saved architectures.
Empty
↑
↓
Remove
Layers
0
Params
0
Memory
0MB
FLOPs
0
Depth
0
Status
—
Inspector
Validate
Analysis
Diagram
Code
Select a layer to inspect
⟳ Refresh
⬇ SVG
⬇ PNG
Publication-ready · Export SVG for papers
TensorFlow.js Experiment
Runs real browser-side gradient updates on generated tensors and reports measured metrics.
Optimizer
Adam
AdamW
SGD
RMSprop
Learning Rate
Loss Function
CrossEntropyLoss
MSELoss
BCELoss
NLLLoss
Epochs
Batch Size
Dataset
Synthetic
MNIST (sim)
CIFAR-10 (sim)
ImageNet (sim)
IMDb (sim)
Close
▶ Start Training
⬇ Export Training .py
Save Architecture
Save a snapshot for comparison or reuse.
Model name
Research Notes
Cancel
💾 Save