My Abyss My Workspace

My Cave

A spatial archive of my workspace

CLOUD 9 UAENA · MEMBER ARCHIVE
✳ fly-01resting
RECONSTRUCTED FROM PERSONAL REFERENCE MATERIAL
✳

Loading the environment…

Preparing collection objects

COLLECTION INDEX

Objects in the room.

Albums, editions and reconstructed room objects. Select an entry to locate it in the 3D environment.

Album page archive.

Open this song in Spotify for lyrics ↗

✳

No scanned pages added.

Personal scans can be added locally and viewed alongside playback.

Up to 16 pages per album. Saved only in this browser; keep your original files.

connectome → motor output → code

a nervous system at work.

restingready to run
recorded connectome / build replay

male-cns:v1.0 · loading anatomy

01 / brain viewfly-01 ♂
● optic lobes● central brain● nerve cord● motor / sensory
02 / neuroscopereplay
neural time0.00 s
spikes / tick0
active neurons0
waiting for neural activity

MN9 feeding rate → template keystrokes

sensory input

loading the saved neural run…

◉   neural-network.mjsjavascript
2 → 6 → 1 · gradient descentutf-8

the experiment

learn the exclusive-or gate

epoch—
loss—
correct / 4 rows—
training loss0 → 2,400 epochs
awaiting first forward pass
terminal / browser session
saved experiments, cues & schedule

checking saved build…

The default view replays a measured build run. Use “run live brain” for new simulation and interactive stimuli on your device. Jobs repeat 60 seconds after completion while this room is open. The deployed build runs every six hours via GitHub Actions, subject to scheduling delays. Browser runs stay in this session.

The same 176,422-neuron connectome as fly-brain-minecraft runs in a JavaScript LIF port. MN9 feeding activity is mapped to typing a bounded neural-network template; completing that template triggers execution. This coding mapping, emotional analogies and seated animation are hand-built, not biological programming ability. The brain view uses 141,781 real soma coordinates and up to 256 sampled spikes per tick; the numeric spike count covers the whole model. Model firing rates are not biological measurements. The program’s accuracy covers four training truth-table rows.

work call · 311 LC4/LPLC2 visual looming neurons at 150 Hz for 300 ms of neural time, then 204 sugar pathway neurons at 120 Hz. poor result / long shift · those sugar neurons plus 38 LB1 bitter neurons, both at 120 Hz. IU playing · sugar neurons at 120 Hz plus 78 JO-F antennal neurons at 90 Hz. break · no injected cue. The fly lifts and flutters with escape/flight output, turns with yaw output, and moves to MN9. These are emotional analogies, not measured fly feelings or music recognition.

Adapted from fly-brain-minecraft by Blendi Remade / fal.ai (MIT). Connectome: HHMI Janelia FlyEM, Cambridge / MRC LMB, Google Research; CC BY 4.0. provenance & model limits ↗

saved neural trace ↗saved program ↗saved build result ↗