diff --git a/playground/a-immersive.html b/playground/a-immersive.html index e24ef85..3df0926 100644 --- a/playground/a-immersive.html +++ b/playground/a-immersive.html @@ -54,6 +54,7 @@
+
@@ -66,11 +67,9 @@ - - -
- - -
- - - + + @@ -151,6 +149,7 @@
Loss
Noise0.050
Joy X,Y0.50, 0.50
+
@@ -166,19 +165,12 @@

Train a neural network to map 2D joystick positions to output parameters using interactive machine learning.

Input Space

The heatmap shows what the network has learned across the input space. Bright dots show training examples. Drag anywhere on the map to move the input position. Double-click or press Follow to toggle follow mode.

-

Examples Mode

-
    -
  1. Drag on the input space to a position
  2. -
  3. Expand a parameter group and drag bars to set desired outputs
  4. -
  5. Press Add Example
  6. -
  7. Repeat, then Train
  8. -
-

RL Mode

-
    -
  1. Move around the input space — the network produces outputs
  2. -
  3. Like it? Press + Good
  4. -
  5. Not quite? Press − Explore for variation
  6. -
+

Learning

+

Both approaches can be freely mixed:

+

Output Modes

Visual: 20 params control the flow-field particle system. Synth: 126 params control a C15 synthesizer engine.

Tap outside to close

diff --git a/playground/c-journey.html b/playground/c-journey.html index 73aa72b..d429e05 100644 --- a/playground/c-journey.html +++ b/playground/c-journey.html @@ -63,19 +63,22 @@ + +
- -
- - -
- -