fly-brain (browser WASM)
Embodied whole-CNS MaleCNS LIF in the browser (WASM) plus MuJoCo flybody and flyvis vision.
Neurons are simulated — leaky integrate-and-fire or better — with documented, tuned parameters.
Dispute this rating →The evidence
No evidence, no level.The LIF core states its neuron model and every parameter, including membrane and synaptic time constants, threshold, reset and refractory period.
https://github.com/Lulzx/fly-brain/blob/4a8a8ebe2b8713106b605f5e32bc8458d65e0f16/src/lif.js#L1-L10Octopamine acts by lowering the spike threshold of its targets rather than through fast synapses, so neuromodulation changes excitability, not weights.
https://github.com/Lulzx/fly-brain/blob/4a8a8ebe2b8713106b605f5e32bc8458d65e0f16/src/sim/neuromod.js#L13-L18The project states in its own documentation that a connectome is not synaptic strengths, neuromodulation or plasticity.
https://github.com/Lulzx/fly-brain/blob/4a8a8ebe2b8713106b605f5e32bc8458d65e0f16/docs/guide/what-the-wiring-gives.md#L3-L7
A 165,122-neuron MaleCNS LIF core after Shiu et al. with every parameter stated in one place (dt 0.5 ms, rest and reset -52 mV, threshold -45 mV, tau_m 20 ms, tau_syn 5 ms, refractory 2.2 ms, delay 1.8 ms, 0.275 mV per contact), plus spike-frequency adaptation - membrane voltages change under a documented neuron model. There is also an alternative whole-connectome firing-rate model after Pugliese et al. 2025 with its ODE written out. Not L2: nothing in the source ever writes to the weight array. The two things that look like plasticity are not. Neuromodulation is modelled honestly but acts by lowering the spike threshold of octopamine targets, an excitability change, not a synaptic one - the code says so in its own header. Short-term depression is a per-presynaptic-neuron resource variable that scales that neuron's output and recovers on a 200 ms constant, non-associative and not a stored weight. The mushroom-body 'three-factor plasticity' text in structures.js is a description of the anatomy for the visualiser, with no implementation behind it. The project states the boundary itself: 'A connectome is wiring. It is not synaptic strengths, neuromodulation, gap junctions or plasticity.'
Record
Nearby
All projects →Research code for the Shiu et al. connectome-based leaky integrate-and-fire model, including activation/silencing experiments, notebooks, and FlyWire data configuration.
FlyWire whole-brain LIF implementation based on Shiu et al., with multiple simulation backends and benchmarking tools. This repository supplies the neural model; it is not a complete embodied demo package.
Ratings on thelearningfly.com are proposals, not verdicts. Every one of them is arguable in public.