Teaching Inspector

CodefyUI can be used as an interactive lesson — students see the exact tensor that flows through every node. The Teaching Inspector captures node outputs during a run and renders them in the right-hand panel.

Walkthrough

1. Drag a TensorInput node onto the canvas (Data category). Set value_mode: explicit and fill the inline grid with the numbers you want the pipeline to see.
2. Wire it through any chain of tensor-op nodes (e.g. Reshape → Softmax → Print).
3. Add a Start node and connect its trigger output to the first node you want executed — typically the TensorInput. Without this, the graph is a draft and Run is rejected (see Your First Graph).
4. Open the toolbar Settings popover and switch Record outputs ON, then click Run. Every completed node's full output is captured in server memory, keyed by the run.
5. Click any node — the Inspector panel fetches that node's input and output, showing shape, dtype, min/max/mean and the actual values stacked top-to-bottom. Cells that changed are heat-coloured.
6. Shift-select two nodes and use Compare Segment (under Settings → Inspection) to focus on just the head-input and tail-output; the canvas wraps them in a light-orange bubble with HEAD / TAIL badges.
7. Switch Record outputs OFF before a heavy training run if you don't want each epoch captured — previously captured runs stay fetchable until the server restarts.

note

Captured data lives in per-session RAM (LRU, last 20 runs). Segment markers are saved with the graph JSON.

Settings popover toggles

The toolbar Settings popover groups every per-tab teaching/training switch in one place — same idea as VS Code's Settings UI:

Toggle	What it does
Record outputs	Capture each completed node's full output for the Inspector. Off by default; turn off for heavy training runs.
Verbose mode	Backend records intermediate algorithmic steps (attention scores, softmax temperatures, etc.) alongside outputs — feeds the Inspector Steps tab.
Compare Segment	Wraps two shift-selected nodes in a HEAD/TAIL bubble so the Inspector shows only that subgraph's boundaries.
Persist weights between runs	Keep Conv2d/Linear/Attention weights across Run clicks (so the model actually learns). When off, every run reinitialises.
Reset all weights now	Drop every cached weight for this tab; the next Run starts fresh.
Capture gradients	Run forward + .backward() and store each layer's gradient for the Inspector Backward tab.
Auto-synthesize loss	When the graph has no Loss/BackwardOnce node, synthesize one so .backward() can run.
Grid snap	Snap dragged nodes to the canvas grid.
Show node tooltips	Reveal the description card when hovering nodes on the canvas.
Node category mode	Basic shows only essential categories in the sidebar; All shows every category.

Step traces (Verbose mode)

With Verbose mode on, instrumented nodes emit a __steps__ trace that the Inspector renders one row at a time. Educational plugin nodes lean on this heavily — e.g. Edu-ColumnStats shows the population-std formula as sum → divide → deviations² → variance → sqrt. See Plugins.

Gradient capture (Backward tab)

With Capture gradients on, the engine runs a forward pass, calls .backward(), and stores each layer's gradient. Open a node's Backward tab in the Inspector to see gradient magnitudes per layer — useful for diagnosing vanishing/exploding gradients.