Choose a validated model in the selector, write a text prompt, and press Send. The model streams its response into the chat pane while the connectome responds to each received token.
Infinite Mode treats the compose text as the first World event. When the compose box is empty, the send action becomes Random and chooses one of ten premade World events. When it contains text, the arrow sends that text directly as the first World event. The Participant always responds first, then World and Participant turns alternate. Infinite Mode applies an internal high-randomness configuration (temperature at least 1.25 and top-p at least .96), regardless of the Normal Chat controls. Switching between Normal Chat and Infinite Mode clears both the conversation and the replay. The compose action becomes Stop while either mode is running; Continue is available only for a stopped Infinite scenario.
Use Advanced settings to open generation, context, GPU offload, replay-speed, and Analysis+ memory controls in a separate scrollable window. Reopened Mode Actions and View Settings panels return to their first control so previously scrolled content cannot remain out of view.
Open Mode Actions, enable Infinite Mode, then use the compose arrow to start an open-ended local roleplay. AIBrain loads two separate copies of the selected GGUF: the World produces only external events; the Participant produces only first-person responses. Every stream update is tied to both its role and turn number, so World tokens cannot append to a Participant bubble. Continue reconstructs the two role histories from the stopped transcript rather than restarting the scenario. The UI clearly labels this as a simulation—the model is generating roleplay text, not a real sentient being. The session continues until Stop is pressed, while each model keeps only a bounded recent context window so it can run without unbounded prompt growth. The right-side connectome renders and records Participant inference telemetry only; World narration remains in the chat and never creates connectome frames.
NN Analysis+ replaces the compact analysis action. It creates a compact JSON or compressed JSON data file with the full conversation and a trained online autoencoder's findings: reconstruction/coherence metrics, channel patterns, temporal segments, and selected high-novelty events. It processes every recorded telemetry frame but does not dump every display-node vector. Its inputs are measured raw-logit summaries and observed generation timing, context, progress, and recent-output values. It does not receive hidden states, attention matrices, MLP activations, or literal transformer neurons.
- Stop requests cancellation after the current native generation step.
- Escape does the same from anywhere in the application.
- Regenerate removes the latest assistant answer and resends its preceding user prompt.
- Clear removes visible messages, replay/analysis readiness, and in-memory conversation history.
Ctrl+Cin the launch terminal requests a clean application shutdown.
The latest completed answer has replay controls beneath it. Replay becomes Stop replay while it owns the graph; the frame arrows and Exit Rewind action are disabled during that run. Manual Previous/Next inspection holds the selected telemetry frame steady until replay or rewind exits. A new response, model selection, graph-quality change, or spacing change replaces that one-response replay buffer.
The controls at the bottom of the chat pane are saved in Windows application settings when a response starts.
| Control | Effect |
|---|---|
| Temperature | Sampling randomness, from 0 to 2. |
| Top-p | Nucleus-sampling cutoff, from 0.05 to 1. |
| Max tokens | Target upper limit; AIBrain permits a short grace window to finish the current sentence. |
| Context | Context window requested from llama.cpp. |
| GPU layers | -1 requests automatic/all-layer offload where the installed backend supports it; 0 requests CPU inference. |
| Analysis+ RAM | Infinite-mode RAM budget for the rewind tail. Older compact analysis records move to the temporary cache. |
| Analysis+ cache | Temporary .cache/temp budget for paged Analysis+ records. Defaults to 1 GB; 0 disables it. |
| Generation speed | UI token pacing from 0.1x to 1.0x. It does not change model sampling. |
Open View settings for Signal-map detail, Rendering GPU, Spacing, and Signal-node borders (0–1). Reset view is always available beside the connectome title. In 2D, drag to pan and use the wheel to zoom; picking follows the translated view. Full 2D keeps all nine named clusters in separate positions. Choose Rewind after a normal response to lock chat actions and move Previous, Replay, and Next below the graph. The same left-side control becomes Exit Rewind while inspection is active; replay works in both 2D and 3D.
The right pane is a native QOpenGLWidget renderer. It supports left-drag to orbit, mouse-wheel zoom, and Reset
view to restore the camera. Pause stops visual updates; press Resume to continue. Low, Medium, and
High quality graphs contain respectively 5,000, 11,000, and 22,000 display nodes. The Spacing slider shows its
selected multiplier immediately and rebuilds the deterministic display layout after you pause movement, avoiding
repeated expensive graph rebuilds while dragging. Channel colors remain visible before telemetry begins.
The GPU selector asks Windows to use the high-performance adapter for the project Python executable on the next launch. The overlay always identifies the renderer actually selected by OpenGL.
Click a signal node to inspect its channel, unmodified normalized measurement, rendered intensity, and display peak.
You can hide it or change its display importance from 0 to 3. Click the displayed Selected signal node number to
enter an exact node index. Signal-node borders outline every display node; use the adjacent numeric control to tune
the outline width. Channel clusters use a stable high-contrast palette for AIBrain's dark default background. Choose Colour
settings beneath the connectome title to customize every semantic interface colour with Qt's wheel, RGB, and hexadecimal
controls; settings are saved to .cache/aibrain.color.json. Sector/channel mapping colours remain stable. All message
bubbles and status/inspector text can be highlighted and copied. Hide, importance, spacing, and colour controls affect
only the display. They never change recorded measurements, Analysis+ input, model weights, or inference.
After generating a response, choose NN Analysis+ to create its compact telemetry-analysis JSON. Use Export chat for a complete conversation-only JSON or readable text file. Full semantics and file layout are documented in Connectome and analysis. Chat bubbles render safe Markdown for headings, emphasis, inline code, and HTTP (S) links while preserving selectable source text. Generated chat follows the newest token only while the message list is already at the bottom; scroll up to read or copy earlier text without being pulled back down.