SNNLang diagram review

Inspect labels, group boundaries, arrow routing, overlaps, and training annotations. Open an SVG for full-size inspection; PNGs are available alongside each view.

all_box_types

All six authoring box types plus a neutral external-encoder annotation: inputs, spiking and analogue populations, a mean operation, output, trainable/frozen parameter groups, and a cross-entropy objective. The E/I group is always explicit. The external encoder illustrates a manual diagram annotation, outside the compiled network.

training

view=training; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
all_box_types: training

circuit

view=circuit; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
all_box_types: circuit

circuit weights

W: projection weights; J: recurrent weights; E/I: excitatory/inhibitory populations; in/out: input/readout; rec: recurrent. Subscripts use source→target order.

view=circuit; weight labels=True; all groups explicit; PNG scale=2 SVG · PNG
all_box_types: circuit_weights

two_layer_classifier

32 spike inputs feed two grouped E/I layers (64/16 and 32/8 neurons), with excitatory recurrence, inhibitory feedback, a four-class spike-count readout, and a cross-entropy training recipe using AdamW.

circuit

view=circuit; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
two_layer_classifier: circuit

training

view=training; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
two_layer_classifier: training

circuit weights

W: projection weights; E1/I1: excitatory/inhibitory populations in layer 1; E2/I2: populations in layer 2; in: input weights. Subscripts use source→target order.

view=circuit; weight labels=True; all groups explicit; PNG scale=2 SVG · PNG
two_layer_classifier: circuit_weights

exp124_coba

Ungrouped two-layer COBA classifier with analogue readout, reductions, score normalization, and trainable/frozen parameter groups. Frozen authoring snapshot from experiments/exp124/recipe.py.

circuit

view=circuit; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp124_coba: circuit

training

view=training; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp124_coba: training

exp124_ping

The same classifier with two local E/I recurrent loops; exposes ungrouped population placement and training annotations. Frozen authoring snapshot from experiments/exp124/recipe.py.

circuit

view=circuit; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp124_ping: circuit

training

view=training; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp124_ping: training

exp131_coupled

Two independently driven 80-E/20-I PING circuits with reciprocal E-to-E and E-to-I coupling. Frozen authoring snapshot from experiments/exp131/recipe.py.

circuit

view=circuit; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp131_coupled: circuit

training

view=training; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp131_coupled: training

circuit weights

W: projection weights; EA/IA: excitatory/inhibitory populations in PING A; EB/IB: populations in PING B; inA/inB: their external drives. Subscripts use source→target order.

view=circuit; weight labels=True; all groups explicit; PNG scale=2 SVG · PNG
exp131_coupled: circuit_weights

exp131_phase_response

One PING circuit with separate drive, E probe, and I probe inputs. Frozen authoring snapshot from experiments/exp131/recipe.py.

circuit

view=circuit; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp131_phase_response: circuit

training

view=training; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp131_phase_response: training

exp133_coupled

Two PING circuits with 1,024-channel drives and reciprocal, delayed, step-modulated coupling. Frozen authoring snapshot from experiments/exp133/recipe.py.

circuit

view=circuit; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp133_coupled: circuit

training

view=training; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp133_coupled: training

exp133_sender_receiver

Independently driven sender and receiver PING circuits with one-way, delayed E-to-E and E-to-I coupling. Frozen authoring snapshot from experiments/exp133/sender_receiver.py.

circuit

view=circuit; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp133_sender_receiver: circuit

training

view=training; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp133_sender_receiver: training

exp136_sustained_cue_ping

Two independently driven sender PING circuits converge on a receiver, each with its own non-recurrent neuronal cue population. Five authored groups include sensory and cue inputs. Uncalibrated circuit proposal; no simulation or training results. Frozen authoring snapshot from experiments/exp136/recipe.py.

circuit

view=circuit; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp136_sustained_cue_ping: circuit

training

view=training; weight labels=False; all groups explicit; PNG scale=2 SVG · PNG
exp136_sustained_cue_ping: training