sinabs 3.1.3__tar.gz → 3.1.3.dev1__tar.gz

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sinabs
-Version: 3.1.3
+Version: 3.1.3.dev1
 Summary: SynSense Spiking Neural Network simulator for deep neural networks (DNNs).
 Author: SynSense (formerly AiCTX)
 Author-email: support@synsense.ai

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/docs/about/release_notes.md

@@ -1,11 +1,5 @@
 # Release notes
 
-## v3.1.3 (04/02/2026)
-
-* Add tests for mapping non-sequential networks
-* Fix networks used for tests, they didnt meet the hardware constraints
-* Update documentation with the non-sequential networks types that have a test
-
 ## v3.1.2 (12/12/2025)
 
 * Update use of visualizer in Sinabs tutorials, previous tutorials were using a deprecated visualizer version.

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/docs/getting_started/iaf_neuron_model.ipynb

@@ -163,6 +163,7 @@
     "import torch\n",
     "import sinabs.layers as sl\n",
     "\n",
+    "\n",
     "# Define a neuron in 'SINABS'\n",
     "neuron = sl.IAF()\n",
     "\n",

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/docs/speck/faqs/available_network_arch.md

@@ -39,31 +39,28 @@ dynapcnn.to(devcie="your device", chip_layers_ordering=[2, 5, 7, 1])
 
 ## What network structure can I define?
 
-`Sinabs` can parse a `torch.nn.Sequential` like architecture, so it is recommended to use a `Sequential` like network.
+`Sinabs` can parse a `torch.nn.Sequential` like architecture, so it is recommended to
+use a `Sequential` like network.
 
 As of `v3.1.0`, we released a network graph extraction feature that helps users deploy their networks with more complex architectures into the devkit.
 Our `Speck` chip, in fact, supports branched architectures. With the graph extraction feature, we support a range of network structures, as shown below:
 
-A network with a merge and a split:
 
-![A network with a merge and a split](imgs/network-with-merge-and-split.png)
+Two independent networks:
+
+![Two independent networks](imgs/two-independent-networks.png)
 
 Two networks with merging outputs:
 
 ![Two networks with merging outputs](imgs/two-networks-merging-output.png)
 
-A network with residual connections:
-
-![A network with residual connections](imgs/network-with-residual-connection.png)
+A network with a merge and a split:
 
-A more complex network:
+![A network with a merge and a split](imgs/network-with-merge-and-split.png)
 
-![A more complex network](imgs/complex-network.png)
 
 Note: with the graph extracture feature it is possible to implement recurrent neural networks. However, this is not recommended or supported as it can result in deadlock on the chip.
 
-Note2: the use of two parallel network although supported by our chip was not fully considered in our sinabs implementation.
-
 ## How to make use of the graph extraction feature?
 
 For general architectures, users need to define their classes, by defining at least the `__init__` method with all the layers, as well as an appropriate `forward` method.
@@ -71,70 +68,64 @@ For general architectures, users need to define their classes, by defining at le
 Here is an example to define a network with a merge and a split:
 
 ```python
-import torch
 import torch.nn as nn
 
 from sinabs.activation.surrogate_gradient_fn import PeriodicExponential
 from sinabs.layers import IAFSqueeze, Merge, SumPool2d
 
-
 class SNN(nn.Module):
     def __init__(self, batch_size) -> None:
         super().__init__()
-        # -- graph node A --
-        self.conv_A = nn.Conv2d(2, 4, 2, 1, bias=False)
-        self.iaf_A = IAFSqueeze(
-            batch_size=batch_size,
-            min_v_mem=-1.0,
-            spike_threshold=1.0,
-            surrogate_grad_fn=PeriodicExponential(),
-        )
-        # -- graph node B --
-        self.conv_B = nn.Conv2d(4, 4, 2, 1, bias=False)
-        self.iaf2_B = IAFSqueeze(
+
+        self.conv1 = nn.Conv2d(2, 1, 2, 1, bias=False)
+        self.iaf1 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
         )
-        self.pool_B = SumPool2d(2, 2)
-        # -- graph node C --
-        self.conv_C = nn.Conv2d(4, 4, 2, 1, bias=False)
-        self.iaf_C = IAFSqueeze(
+
+        self.conv2 = nn.Conv2d(1, 1, 2, 1, bias=False)
+        self.iaf2 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
         )
-        self.pool_C = SumPool2d(2, 2)
-        # -- graph node D --
-        self.conv_D = nn.Conv2d(4, 4, 2, 1, bias=False)
-        self.iaf_D = IAFSqueeze(
+        self.pool2 = SumPool2d(2, 2)
+
+        self.conv3 = nn.Conv2d(1, 1, 2, 1, bias=False)
+        self.iaf3 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
         )
-        # -- graph node E --
-        self.conv_E = nn.Conv2d(4, 4, 2, 1, bias=False)
-        self.iaf3_E = IAFSqueeze(
+        self.pool3 = SumPool2d(2, 2)
+        self.pool3a = SumPool2d(5, 5)
+
+        self.conv4 = nn.Conv2d(1, 1, 2, 1, bias=False)
+        self.iaf4 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
         )
-        self.pool_E = SumPool2d(2, 2)
-        # -- graph node F --
-        self.conv_F = nn.Conv2d(4, 4, 2, 1, bias=False)
-        self.iaf_F = IAFSqueeze(
+        self.pool4 = SumPool2d(3, 3)
+
+        self.flat1 = nn.Flatten()
+        self.flat2 = nn.Flatten()
+
+        self.conv5 = nn.Conv2d(1, 1, 2, 1, bias=False)
+        self.iaf5 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
         )
-        # -- graph node G --
-        self.fc3 = nn.Linear(144, 10, bias=False)
-        self.iaf3_fc = IAFSqueeze(
+
+        self.fc2 = nn.Linear(25, 10, bias=False)
+        self.iaf2_fc = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
@@ -143,47 +134,43 @@ class SNN(nn.Module):
 
         # -- merges --
         self.merge1 = Merge()
-
-        # -- falts --
-        self.flat_D = nn.Flatten()
-        self.flat_F = nn.Flatten()
+        self.merge2 = Merge()
 
     def forward(self, x):
         # conv 1 - A/0
-        convA_out = self.conv_A(x)
-        iaf_A_out = self.iaf_A(convA_out)
+        con1_out = self.conv1(x)
+        iaf1_out = self.iaf1(con1_out)
 
         # conv 2 - B/1
-        conv_B_out = self.conv_B(iaf_A_out)
-        iaf_B_out = self.iaf2_B(conv_B_out)
-        pool_B_out = self.pool_B(iaf_B_out)
+        conv2_out = self.conv2(iaf1_out)
+        iaf2_out = self.iaf2(conv2_out)
+        pool2_out = self.pool2(iaf2_out)
 
         # conv 3 - C/2
-        conv_C_out = self.conv_C(pool_B_out)
-        iaf_C_out = self.iaf_C(conv_C_out)
-        pool_C_out = self.pool_C(iaf_C_out)
-
-        # conv 4 - D/4
-        conv_D_out = self.conv_D(pool_C_out)
-        iaf_D_out = self.iaf_D(conv_D_out)
-        # fc 1 - E/3
-        conv_E_out = self.conv_E(pool_B_out)
-        iaf3_E_out = self.iaf3_E(conv_E_out)
-        pool_E_out = self.pool_E(iaf3_E_out)
-
-        # fc 2 - F/6
-        conv_F_out = self.conv_F(pool_E_out)
-        iaf_F_out = self.iaf_F(conv_F_out)
-
-        # fc 2 - G/5
-        flat_D_out = self.flat_D(iaf_D_out)
-        flat_F_out = self.flat_F(iaf_F_out)
-
-        merge1_out = self.merge1(flat_D_out, flat_F_out)
-        fc3_out = self.fc3(merge1_out)
-        iaf3_fc_out = self.iaf3_fc(fc3_out)
-
-        return iaf3_fc_out
+        conv3_out = self.conv3(iaf1_out)
+        iaf3_out = self.iaf3(conv3_out)
+        pool3_out = self.pool3(iaf3_out)
+        pool3a_out = self.pool3a(iaf3_out)
+
+        # conv 4 - D/3
+        merge1_out = self.merge1(pool2_out, pool3_out)
+        conv4_out = self.conv4(merge1_out)
+        iaf4_out = self.iaf4(conv4_out)
+        pool4_out = self.pool4(iaf4_out)
+        flat1_out = self.flat1(pool4_out)
+
+        # conv 5 - E/4
+        conv5_out = self.conv5(pool3a_out)
+        iaf5_out = self.iaf5(conv5_out)
+        flat2_out = self.flat2(iaf5_out)
+
+        # fc 2 - F/5
+        merge2_out = self.merge2(flat2_out, flat1_out)
+
+        fc2_out = self.fc2(merge2_out)
+        iaf2_fc_out = self.iaf2_fc(fc2_out)
+
+        return iaf2_fc_out
 ```
 
 ## Can I achieve a "Residual Connection" like ResNet does?
@@ -193,9 +180,9 @@ change the `samna.speck2f.configuration.CNNLayerDestination.layer` to achieve th
 familiar with the `samna-configuration`.
 You can also make use of our network graph extraction feature, to implement residual networks.
 
-## How can I define "Residual Connection" manually?
+## How to use "Residual Connection" manually?
 
-You can also achieve "Residual Connection" by manually modify the `samna-configuration`.
+Alright! Here I will give an example of achieving the "Residual Connection" by manually modify the `samna-configuration`.
 
 Let's say you want an architecture like below:
 
@@ -234,7 +221,7 @@ class ResidualBlock(nn.Module):
 
 ```
 
-You can write it like:
+Since currently Sinabs can only parse Sequential like network, we need to do some tedious work like below:
 
 ```python
 # define a Sequential first
@@ -277,8 +264,8 @@ devkit.get_model().apply_configuration(samna_cfg)
 
 ```
 
-It is a lot of manual work but it will let you have your Residual Block.
-We recommend to use our network graph extraction feature for residual connections.
+I have to say it is not an elegant solution though, it should help you to achieve an initial Residual Block. We will
+improve this part after Sinabs has the ability for extracting model's graph.
 
 ## What execution order should I be aware of when I am implementing a sequential structure?
 You should be aware with the internal layer order.

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/docs/speck/notebooks/nmnist_quick_start.ipynb

@@ -97,6 +97,7 @@
    "source": [
     "from torch import nn\n",
     "\n",
+    "\n",
     "# define a CNN model\n",
     "cnn = nn.Sequential(\n",
     "    # [2, 34, 34] -> [8, 17, 17]\n",
@@ -1180,6 +1181,7 @@
    "source": [
     "from sinabs.backend.dynapcnn.dynapcnn_visualizer import DynapcnnVisualizer\n",
     "\n",
+    "\n",
     "visualizer = DynapcnnVisualizer(\n",
     "    window_scale=(4, 8),\n",
     "    dvs_shape=(34, 34),\n",

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/docs/speck/notebooks/play_with_speck_dvs.ipynb

@@ -103,6 +103,7 @@
     "from sinabs.backend.dynapcnn.dynapcnn_visualizer import DynapcnnVisualizer\n",
     "from sinabs.backend.dynapcnn import DynapcnnNetwork\n",
     "\n",
+    "\n",
     "# create a dummy snn for DynapcnnNetwork initialization\n",
     "snn = nn.Sequential(\n",
     "    nn.Conv2d(1, 1, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False),\n",

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/docs/tutorials/nmnist.ipynb

@@ -292,6 +292,7 @@
    "source": [
     "from tqdm.notebook import tqdm\n",
     "\n",
+    "\n",
     "n_epochs = 1\n",
     "optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)\n",
     "crit = nn.functional.cross_entropy\n",

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/sinabs/backend/dynapcnn/chips/dynapcnn.py

@@ -101,7 +101,7 @@ class DynapcnnConfigBuilder(ConfigBuilder):
         dimensions["input_shape"]["feature_count"] = channel_count
 
         # dimensions["output_feature_count"] already done in conv2d_to_dict
-        f, h, w = layer.get_neuron_shape()
+        (f, h, w) = layer.get_neuron_shape()
         dimensions["output_shape"]["size"] = {}
         dimensions["output_shape"]["feature_count"] = f
         dimensions["output_shape"]["size"]["x"] = w
@@ -121,7 +121,7 @@ class DynapcnnConfigBuilder(ConfigBuilder):
         config_dict["dimensions"] = dimensions
         # Update parameters from convolution
         if layer.conv_layer.bias is not None:
-            weights, biases = layer.conv_layer.parameters()
+            (weights, biases) = layer.conv_layer.parameters()
         else:
             (weights,) = layer.conv_layer.parameters()
             biases = torch.zeros(layer.conv_layer.out_channels)

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/sinabs/backend/dynapcnn/dynapcnn_visualizer.py

@@ -455,7 +455,7 @@ class DynapcnnVisualizer:
 
         # Streamer graph
         # Dvs node
-        _, dvs_member_filter, _, streamer_node = self.streamer_graph.sequential(
+        (_, dvs_member_filter, _, streamer_node) = self.streamer_graph.sequential(
             [
                 dynapcnn_network.samna_device.get_model_source_node(),
                 samna.graph.JitMemberSelect(),
@@ -526,7 +526,7 @@ class DynapcnnVisualizer:
         ## Readout node
         if "r" in self.gui_type:
             if self.readout_node == "JitMajorityReadout":
-                _, majority_readout_node, _ = self.streamer_graph.sequential(
+                (_, majority_readout_node, _) = self.streamer_graph.sequential(
                     [
                         spike_collection_node,
                         samna.graph.JitMajorityReadout(samna.ui.Event),
@@ -544,7 +544,7 @@ class DynapcnnVisualizer:
                     self.readout_default_threshold_high
                 )
             else:
-                _, majority_readout_node, _ = self.streamer_graph.sequential(
+                (_, majority_readout_node, _) = self.streamer_graph.sequential(
                     [
                         spike_collection_node,
                         self.readout_node,

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/sinabs/backend/dynapcnn/nir_graph_extractor.py

@@ -504,7 +504,7 @@ class GraphExtractor:
             A handler to the newly added `DVSLayer` instance.
         """
 
-        features, height, width = dvs_input_shape
+        (features, height, width) = dvs_input_shape
         if features > 2:
             raise ValueError(
                 f"A DVSLayer istance can have a max feature dimension of 2 but {features} was given."
@@ -578,7 +578,7 @@ class GraphExtractor:
                 f"A DVSLayer node exists and there are {nb_entries} entry nodes in the graph: the DVSLayer should be the only entry node."
             )
 
-        features, _, _ = dvs_input_shape
+        (features, _, _) = dvs_input_shape
 
         if features > 2:
             raise ValueError(
@@ -752,7 +752,7 @@ class GraphExtractor:
                 # different input sources to a core to have the same output shapes.
                 if any(inp.shape != inputs[0].shape for inp in inputs):
                     raise ValueError(
-                        f"Layer `sinabs.layers.merge.Merge` (node {node}) requires input tensors with the same shape."
+                        f"Layer `sinabs.layers.merge.Merge` (node {node}) requires input tensors with the same shape"
                     )
 
                 # forward input through the node.
@@ -828,7 +828,6 @@ class GraphExtractor:
         if len(sources) == 0:
             return -1
         if len(sources) > 1:
-            # return -1 #TODO: why throw a runtime error when the documentation explicitly say -1 in case of more than one input node?
             raise RuntimeError(f"Node {node} has more than 1 input")
         return sources.pop()
 

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/sinabs/layers/pool2d.py

@@ -80,7 +80,7 @@ class SpikingMaxPooling2dLayer(nn.Module):
         Returns:
             (channelsOut, height_out, width_out)
         """
-        channels, height, width = input_shape
+        (channels, height, width) = input_shape
 
         height_out = conv_output_size(
             height + sum(self.padding[2:]), self.pool_size[0], self.strides[0]

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/sinabs.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sinabs
-Version: 3.1.3
+Version: 3.1.3.dev1
 Summary: SynSense Spiking Neural Network simulator for deep neural networks (DNNs).
 Author: SynSense (formerly AiCTX)
 Author-email: support@synsense.ai

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/sinabs.egg-info/SOURCES.txt

@@ -121,9 +121,8 @@ docs/speck/faqs/index.rst
 docs/speck/faqs/output_monitoring.md
 docs/speck/faqs/save_hardware_config_as_binary.md
 docs/speck/faqs/tips_for_training.md
-docs/speck/faqs/imgs/complex-network.png
 docs/speck/faqs/imgs/network-with-merge-and-split.png
-docs/speck/faqs/imgs/network-with-residual-connection.png
+docs/speck/faqs/imgs/two-independent-networks.png
 docs/speck/faqs/imgs/two-networks-merging-output.png
 docs/speck/notebooks/leak_neuron.ipynb
 docs/speck/notebooks/nmnist_quick_start.ipynb
@@ -229,7 +228,6 @@ sinabs/layers/functional/__init__.py
 sinabs/layers/functional/alif.py
 sinabs/layers/functional/lif.py
 tests/__init__.py
-tests/hw_utils.py
 tests/mnist_params.pt
 tests/requirements.txt
 tests/test_activations.py
@@ -255,8 +253,8 @@ tests/inputs_and_results/hooks/input_diffs.pth
 tests/inputs_and_results/hooks/model_synops.pth
 tests/models/README.txt
 tests/models/synop_hook_model.pth
-tests/test_dynapcnn/__init__.py
 tests/test_dynapcnn/custom_jit_filters.py
+tests/test_dynapcnn/hw_utils.py
 tests/test_dynapcnn/test_auto_mapping.py
 tests/test_dynapcnn/test_compatible_layer_build.py
 tests/test_dynapcnn/test_config_making.py
@@ -289,7 +287,6 @@ tests/test_dynapcnnnetwork/model_dummy_1.py
 tests/test_dynapcnnnetwork/model_dummy_2.py
 tests/test_dynapcnnnetwork/model_dummy_3.py
 tests/test_dynapcnnnetwork/model_dummy_4.py
-tests/test_dynapcnnnetwork/model_dummy_5.py
 tests/test_dynapcnnnetwork/model_dummy_seq.py
 tests/test_dynapcnnnetwork/test_dynapcnnnetwork.py
 tests/test_dynapcnnnetwork/test_failcases.py

sinabs-3.1.3.dev1/sinabs.egg-info/pbr.json

@@ -0,0 +1 @@
+{"git_version": "70cee7a", "is_release": false}

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/tests/test_dynapcnn/test_device_movement.py

@@ -3,7 +3,7 @@ import torch.nn as nn
 
 from sinabs.backend.dynapcnn import DynapcnnNetwork
 from sinabs.from_torch import from_model
-from tests.hw_utils import find_open_devices
+from hw_utils import find_open_devices
 
 ann = nn.Sequential(
     nn.Conv2d(1, 20, 5, 1, bias=False),

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/tests/test_dynapcnn/test_discover_device.py

@@ -2,7 +2,7 @@ import pytest
 import samna
 
 from sinabs.backend.dynapcnn import io
-from tests.hw_utils import find_open_devices
+from hw_utils import find_open_devices
 
 
 def test_is_device_type():

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/tests/test_dynapcnn/test_large_net.py

@@ -14,7 +14,7 @@ from torch import nn
 from sinabs.backend.dynapcnn.dynapcnn_network import DynapcnnNetwork
 from sinabs.from_torch import from_model
 from sinabs.layers import NeuromorphicReLU
-from tests.hw_utils import find_open_devices
+from hw_utils import find_open_devices
 
 
 class DynapCnnNetA(nn.Module):

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/tests/test_dynapcnn/test_neuron_leak.py

@@ -3,7 +3,12 @@ import time
 import pytest
 import samna
 import torch
-from tests.hw_utils import find_open_devices, is_any_samna_device_connected
+from hw_utils import (
+    find_open_devices,
+    get_ones_network,
+    is_any_samna_device_connected,
+    is_device_connected,
+)
 from torch import nn
 
 from sinabs.backend.dynapcnn import DynapcnnNetwork

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/tests/test_dynapcnn/test_single_neuron_hardware.py

@@ -1,8 +1,9 @@
 import pytest
-from tests.hw_utils import (
+from hw_utils import (
     find_open_devices,
     get_ones_network,
     is_any_samna_device_connected,
+    is_device_connected,
     reset_all_connected_boards,
 )
 

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/tests/test_dynapcnn/test_visualizer.py

@@ -3,8 +3,9 @@ from typing import Callable, Union
 
 import pytest
 import samna
-from .custom_jit_filters import majority_readout_filter as custom_filter
-from tests.hw_utils import find_open_devices
+from custom_jit_filters import majority_readout_filter as custom_filter
+from hw_utils import find_open_devices, is_any_samna_device_connected
+
 from sinabs.backend.dynapcnn.dynapcnn_visualizer import DynapcnnVisualizer
 
 
@@ -88,7 +89,7 @@ def test_jit_compilation():
             streamer_graph = samna.graph.EventFilterGraph()
             # Streamer graph
             # Dvs node
-            _, dvs_member_filter, _, streamer_node = streamer_graph.sequential(
+            (_, dvs_member_filter, _, streamer_node) = streamer_graph.sequential(
                 [
                     # samna.graph.JitSource(samna.speck2e.event.OutputEvent),
                     dynapcnn_network.samna_device.get_model_source_node(),

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/tests/test_dynapcnnnetwork/conftest_dynapcnnnetwork.py

@@ -14,11 +14,6 @@ from .model_dummy_4 import batch_size as batch_size_4
 from .model_dummy_4 import expected_output as expected_output_4
 from .model_dummy_4 import input_shape as input_shape_4
 from .model_dummy_4 import snn as snn_4
-from .model_dummy_5 import batch_size as batch_size_5
-from .model_dummy_5 import expected_output as expected_output_5
-from .model_dummy_5 import input_shape as input_shape_5
-from .model_dummy_5 import snn as snn_5
-
 from .model_dummy_seq import (
     expected_seq_1,
     expected_seq_2,
@@ -35,13 +30,3 @@ args_DynapcnnNetworkTest = [
     (seq_1, input_shape_seq, 1, expected_seq_1),
     (seq_2, input_shape_seq, 1, expected_seq_2),
 ]
-
-args_DynapcnnNetworkMappingTest = [
-    (snn_1, input_shape_1, batch_size_1),
-    (snn_2, input_shape_2, batch_size_2),
-    (snn_3, input_shape_3, batch_size_3),
-    (snn_4, input_shape_4, batch_size_4),
-    (seq_1, input_shape_seq, 1),
-    (seq_2, input_shape_seq, 1),
-    # (snn_5, input_shape_5, batch_size_5), #TODO: forward method needs implementation
-]

{sinabs-3.1.3 → sinabs-3.1.3.dev1}/tests/test_dynapcnnnetwork/model_dummy_1.py

@@ -13,49 +13,49 @@ class SNN(nn.Module):
     def __init__(self, batch_size) -> None:
         super().__init__()
 
-        self.conv1 = nn.Conv2d(2, 10, 2, 1, bias=False)
+        self.conv1 = nn.Conv2d(2, 10, 2, 1, bias=False)  # node 0
         self.iaf1 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
-        )
-        self.pool1 = nn.AvgPool2d(2, 2)
-        self.pool1a = nn.AvgPool2d(2, 2)
+        )  # node 1
+        self.pool1 = nn.AvgPool2d(3, 3)  # node 2
+        self.pool1a = nn.AvgPool2d(4, 4)  # node 3
 
-        self.conv2 = nn.Conv2d(10, 10, 1, 1, bias=False)
+        self.conv2 = nn.Conv2d(10, 10, 4, 1, bias=False)  # node 4
         self.iaf2 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
-        )
+        )  # node 6
 
-        self.conv3 = nn.Conv2d(10, 1, 2, 1, bias=False)
+        self.conv3 = nn.Conv2d(10, 1, 2, 1, bias=False)  # node 8
         self.iaf3 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
-        )
+        )  # node 9
 
         self.flat = nn.Flatten()
 
-        self.fc1 = nn.Linear(225, 200, bias=False)
+        self.fc1 = nn.Linear(49, 500, bias=False)  # node 10
         self.iaf4 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
-        )
+        )  # node 11
 
-        self.fc2 = nn.Linear(200, 10, bias=False)
+        self.fc2 = nn.Linear(500, 10, bias=False)  # node 12
         self.iaf5 = IAFSqueeze(
             batch_size=batch_size,
             min_v_mem=-1.0,
             spike_threshold=1.0,
             surrogate_grad_fn=PeriodicExponential(),
-        )
+        )  # node 13
 
         self.adder = Merge()
 
@@ -67,13 +67,14 @@ class SNN(nn.Module):
 
         conv2_out = self.conv2(pool1_out)
         iaf2_out = self.iaf2(conv2_out)
+
         conv3_out = self.conv3(self.adder(pool1a_out, iaf2_out))
         iaf3_out = self.iaf3(conv3_out)
+
         flat_out = self.flat(iaf3_out)
 
         fc1_out = self.fc1(flat_out)
         iaf4_out = self.iaf4(fc1_out)
-
         fc2_out = self.fc2(iaf4_out)
         iaf5_out = self.iaf5(fc2_out)