v3.14.0 — Visual SNN Design Studio

Visual SNN Design Studio (Experimental)

A web-based IDE for the complete spiking neural network lifecycle: design neuron models, build networks, train with surrogate gradients, compile to SystemVerilog, and synthesise to FPGA — all from a single browser tab.

pip install sc-neurocore[studio]
sc-neurocore studio

What's New

• 118-model browser with live simulation, parameter sliders, pattern classification
• 20+ analysis views — trace, phase, ISI, f-I, bifurcation, heatmap, sensitivity, STA, frequency response, characterisation, multi-model overlay
• Compiler Inspector — SC IR build/verify/emit, SystemVerilog generation, co-simulation
• Synthesis Dashboard — Yosys synthesis for 4 FPGA targets (ice40, ECP5, Gowin, Xilinx), multi-target comparison
• Training Monitor — live SSE streaming, 6 surrogate gradients, per-layer spike rates
• Network Canvas — React Flow drag-and-drop populations, NIR export/import
• Full pipeline — graph → simulate → compile → synthesise in one click
• Project save/load — persistent JSON workspaces

Rust Engine

• py_simulate_ei_network() — fused E-I network in single Rust call
• py_batch_simulate() — batch model dispatch loop
• 288 Rust tests passing

Security

• 30 CodeQL alerts fixed (traceback exposure + path traversal)
• DOMPurify XSS fix

Other

• Performance: model list caching (<1ms after first load)
• 140+ new Studio tests
• Full documentation

Full changelog: v3.13.3...v3.14.0