Hello! First, thank you for creating this excellent visualization app with fully repository, It's been very helpful for understanding neural network behavior.
I noticed a small labeling inconsistency that might cause confusion for users.
In the neuron detail panel, when clicking on output layer neurons, the activation function is displayed as "Linear". However, the visualization correctly applies SoftMax to display probabilities in the prediction chart.
In mlp_train.py, the comment correctly notes that SoftMax is , but the label stored in the exported weights shows "linear" rather than "SoftMax" because of writing "linear" inside labeling bracket.
I think this problem can be easily solved by changing "linear" to "SoftMax" and retraining the model.
with bests
Hello! First, thank you for creating this excellent visualization app with fully repository, It's been very helpful for understanding neural network behavior.
I noticed a small labeling inconsistency that might cause confusion for users.
In the neuron detail panel, when clicking on output layer neurons, the activation function is displayed as "Linear". However, the visualization correctly applies SoftMax to display probabilities in the prediction chart.
In mlp_train.py, the comment correctly notes that SoftMax is , but the label stored in the exported weights shows "linear" rather than "SoftMax" because of writing "linear" inside labeling bracket.
I think this problem can be easily solved by changing "linear" to "SoftMax" and retraining the model.
with bests