adjoint_channels listlengths 2 2 | adjoint_scales dict | bin_ms float64 1 1 | chunks listlengths 10 10 | complete bool 1
class | dt_ms float64 0.1 0.1 | format stringclasses 1
value | located_neurons int64 140k 140k | metadata dict | n_neurons int64 167k 167k | neuron_file stringclasses 1
value | playback_direction stringclasses 1
value | sample_phase stringclasses 1
value | superclasses listlengths 27 27 | total_bins int64 100 100 | trace_kind stringclasses 1
value | units stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
[
"voltage_adjoint",
"drive_adjoint"
] | {
"drive_adjoint": 0.00008255933789769188,
"voltage_adjoint": 0.0005098350811749697
} | 1 | [
{
"bin_steps": 10,
"first_bin": 0,
"first_end_step": 10,
"num_bins": 10,
"url": "adjoints/b5cdb20337f2f8af-0000000000-0000000010.bin"
},
{
"bin_steps": 10,
"first_bin": 10,
"first_end_step": 110,
"num_bins": 10,
"url": "adjoints/b5cdb20337f2f8af-0000000010-0000000020.bin"... | true | 0.1 | njx-trace-v1 | 139,662 | {
"active_neurons": 21652,
"adjoint_recording": {
"bin_ms": 1,
"bin_steps": 10,
"n_bins": 100,
"objective": "mean(log1p(exp-filtered spikes at trial end)^2)",
"time_convention": "Cotangents of state after the final tick before each exclusive end_step; chronological storage, reverse-time playback... | 166,700 | neurons.bin | backward | bin_end | [
"ENS",
"ascending_neuron",
"cb_efferent",
"cb_endocrine",
"cb_intrinsic",
"cb_motor",
"cb_sensory",
"cb_sensory_tbc",
"descending_neuron",
"descending_neuron_tbc",
"efferent_ascending",
"efferent_descending",
"ol_intrinsic",
"ol_sensory",
"sensory_ascending",
"sensory_ascending_tbc",
... | 100 | adjoint | loss/mV |
[
"voltage_adjoint",
"drive_adjoint"
] | {
"drive_adjoint": 0.00008255933789769188,
"voltage_adjoint": 0.0005098350811749697
} | 1 | [
{
"bin_steps": 10,
"first_bin": 0,
"first_end_step": 10,
"num_bins": 10,
"url": "adjoints/b5cdb20337f2f8af-0000000000-0000000010.bin"
},
{
"bin_steps": 10,
"first_bin": 10,
"first_end_step": 110,
"num_bins": 10,
"url": "adjoints/b5cdb20337f2f8af-0000000010-0000000020.bin"... | true | 0.1 | njx-trace-v1 | 139,662 | {
"active_neurons": 21652,
"adjoint_recording": {
"bin_ms": 1,
"bin_steps": 10,
"n_bins": 100,
"objective": "mean(log1p(exp-filtered spikes at trial end)^2)",
"time_convention": "Cotangents of state after the final tick before each exclusive end_step; chronological storage, reverse-time playback... | 166,700 | neurons.bin | backward | bin_end | [
"ENS",
"ascending_neuron",
"cb_efferent",
"cb_endocrine",
"cb_intrinsic",
"cb_motor",
"cb_sensory",
"cb_sensory_tbc",
"descending_neuron",
"descending_neuron_tbc",
"efferent_ascending",
"efferent_descending",
"ol_intrinsic",
"ol_sensory",
"sensory_ascending",
"sensory_ascending_tbc",
... | 100 | adjoint | loss/mV |
malecns-eventprop-zero-trace
A recorded simulation trace in njx-trace-v1 format, covering 100 ms of
simulated activity. Replay is deterministic playback of stored states; opening
this dataset does not run a simulation.
| Property | Value |
|---|---|
| Recorded duration | 100 ms |
| Bins | 100 x 1 ms |
| Integration step | 0.1 ms |
| Neurons | 166,700 |
| Located somata | 139,662 |
| Recorded passes | adjoint |
| Trace format | njx-trace-v1 |
Replay
Chunks are immutable, content-addressed files fetched directly from the Hugging Face CDN. Pin a commit revision rather than a branch so the recording cannot change under a reader.
<script type="module" src="/connectome-viewer/connectome-viewer.js"></script>
<connectome-viewer trace="https://huggingface.co/datasets/ranvier-labs/malecns-eventprop-zero-trace/resolve/73cefb2e944361912b826ad74dbf3844520cc531/manifest.json" style="height:720px"></connectome-viewer>
The viewer resolves neuron_file and every chunk URL relative to the manifest
URL, so no rewriting is needed when the bundle moves between hosts.
- Downloads last month
- 101