Basic SNN functionality in hls4ml (mapped from snntorch)

.gitignore

@@ -13,5 +13,6 @@ docs/_build
 docs/autodoc/*
 hls4mlprj_*
 *~
+*.ipynb
 *.ipynb_checkpoints/
 *.bak

docs/advanced/snn.rst

@@ -0,0 +1,201 @@
+========================================
+Spiking Neural Networks (PyTorch/SNN)
+========================================
+
+This page describes the initial SNN support in the PyTorch frontend.
+
+Install the SNN frontend dependencies with:
+
+.. code-block:: bash
+
+   pip install hls4ml[snn]
+
+Backend support
+===============
+
+The SNN flow currently supports only the ``Vitis`` backend.
+
+Execution model
+===============
+
+Current hls4ml SNN implementations are synchronous (clock-driven). Neuron state
+updates and layer computations run in standard HLS pipelines/streams each cycle
+according to interface handshakes. The generated design is not a native
+asynchronous/event-routed neuromorphic architecture (yet!).
+
+Reuse factor support
+====================
+
+Standard hls4ml layers used inside an SNN, such as ``Dense``/linear layers,
+retain their normal ``ReuseFactor`` support. ``ReuseFactor`` can still be set at
+the model, layer type, or layer name level for these layers, and each dense layer
+uses its own configured value independently of the surrounding spiking neuron
+layers.  The spiking neuron kernels themselves, ``IFNeuron`` and ``LIFNeuron``, do not
+currently expose ``ReuseFactor``. They process one timestep at a time, keep
+internal membrane state across timesteps, and unroll the per-neuron update loop
+across ``n_out`` channels.
+
+Supported PyTorch modules and readout wrappers
+==============================================
+
+The frontend currently supports direct parsing of:
+
+* ``Leaky`` -> ``LIFNeuron`` (or ``IFNeuron`` when ``beta`` is effectively 1)
+
+``SNNReadout`` is an hls4ml layer, not a ``snntorch`` module. To use the
+built-in hls4ml readout from a PyTorch model, instantiate the provided PyTorch
+marker module:
+
+.. code-block:: python
+
+   from hls4ml.contrib.snntorch import SNNReadout
+
+The marker is an identity in PyTorch and is converted to the hls4ml
+``SNNReadout`` layer by the PyTorch frontend.
+
+`snntorch` tracing
+==================
+
+``snntorch`` modules are treated as leaf modules by the hls4ml PyTorch FX tracer.
+This allows conversion models to use ``snntorch.Leaky`` directly without defining
+conversion-only wrapper classes.
+
+For ``Leaky``, the supported reset mechanisms are:
+
+* ``subtract``
+* ``zero``
+
+``threshold`` supports scalar or per-neuron vectors (length ``n_out``) for both ``IFNeuron`` and ``LIFNeuron``.
+``beta`` supports scalar or per-neuron vectors for ``LIFNeuron``.
+
+Conversion selects the most memory-efficient representation automatically:
+
+* scalar values are emitted as compile-time constants
+* per-neuron values are emitted as parameter vectors
+
+For trainable snntorch parameters, conversion uses the current parameter values from the model
+at conversion time.
+
+Readout and Decision Rules
+==========================
+
+The hls4ml ``SNNReadout`` layer implements programmable per-model decision policies.
+By default, ``output_mode="spike"`` preserves the original spike-count behavior:
+
+* ``argmax_spike_count``
+* ``first_to_threshold``
+* ``threshold_then_argmax``
+* ``binary_logit`` (for binary classifiers with ``n_classes == 2``)
+
+The layer accumulates class spikes over a window. For most decision rules it emits
+a class ID. For ``binary_logit``, it emits a score equal to
+``count(class_1) - count(class_0)``.
+
+For non-spiking readout heads, set ``output_mode="membrane"`` and connect
+``SNNReadout`` directly after the final dense/linear layer instead of after a
+final spiking neuron. In this mode the readout owns the final membrane state:
+
+.. code-block:: python
+
+   x = self.fc2(x)
+   return self.readout(x)
+
+At each timestep, the generated readout computes:
+
+.. code-block:: cpp
+
+   mem[i] = beta * mem[i] + input[i];
+
+No threshold or reset-on-spike is applied in membrane mode. The supported
+membrane decision policies are:
+
+* ``argmax_membrane``
+* ``binary_logit`` (emits ``mem(class_1) - mem(class_0)`` for binary classifiers)
+
+This will be explained in a tutorial in the hls4ml-tutorials repo.
+
+Do not place a final spiking neuron before ``SNNReadout(output_mode="membrane")``
+unless you intentionally want the readout to consume that neuron's spike output.
+The membrane mode does not recover or expose the internal membrane state of a
+preceding ``Leaky``/``IFNeuron``/``LIFNeuron`` layer. If a final output neuron
+has a learnable ``beta``, that learnable neuron membrane is not the same state
+as the readout-owned membrane. The readout uses its own scalar ``beta``.
+
+When using the default PyTorch parser, the wrapper module should expose these
+attributes as needed:
+
+* ``n_classes`` (defaults to the input feature count if omitted)
+* ``window_size`` or ``stream_length`` (defaults to ``1``)
+* ``class_threshold`` (defaults to ``1``)
+* ``output_mode`` (defaults to ``spike``; use ``membrane`` for readout-owned membrane accumulation)
+* ``beta`` (defaults to ``1.0`` for membrane readout)
+* ``decision_rule`` (defaults to ``argmax_spike_count``)
+* ``reset_policy`` or ``state_reset_policy`` (defaults to ``fixed_window``)
+
+Window Boundary Semantics
+=========================
+
+The current implementation uses ``window_size`` timesteps as the sequence boundary
+for generated HLS. During PyTorch conversion, the first fixed-window
+``SNNReadout``'s ``window_size`` is propagated to all converted ``IFNeuron`` and
+``LIFNeuron`` layers in the graph.
+
+At each boundary:
+
+* the class decision is emitted
+* internal readout counters or readout membrane state are reset for the next sequence
+* internal ``IFNeuron``/``LIFNeuron`` membrane state is reset for the next sequence
+
+The reset happens after the final timestep has been processed and has contributed
+to the output. This behavior is compatible with fixed-length time windows.
+
+Only fixed-window reset is implemented in generated layer kernels today.
+``state_reset_policy`` accepts future-facing values such as ``tlast``,
+``host_pulse``, and ``never``, but the current layer kernels still use fixed
+``window_size`` reset behavior.
+
+Running ``hls_model.predict()``
+==============================
+
+Compiled SNN models are stateful across top-function calls. For fixed-window
+SNN inference, call the compiled model once per timestep and pass exactly
+``window_size`` timesteps for each independent sequence:
+
+.. code-block:: python
+
+   last = None
+   for step in range(timesteps):
+       x_step = x_sequence[step].astype("float32")[None, :]
+       last = hls_model.predict(x_step)
+
+After the last call in the window, generated HLS resets the neuron and readout
+state for the next sequence. Avoid making stray single-timestep ``predict``
+calls before evaluating a sequence, because those calls advance the state.
+
+For membrane readout, the PyTorch reference should match the generated readout
+accumulation:
+
+.. code-block:: python
+
+   mem = torch.zeros_like(currents[:, 0, :])
+   for step in range(currents.shape[1]):
+       mem = beta * mem + currents[:, step, :]
+   pred = mem.argmax(dim=1)
+
+Using only the final dense current, or using spike-count reduction for a
+membrane readout, does not match generated HLS behavior.
+
+Precision note
+==============
+
+Membrane readout accumulates dense currents over the full window, so very narrow
+fixed-point types can reduce accuracy even when the floating-point PyTorch model
+looks good.
+
+``TLAST`` note
+==============
+
+True AXI sideband ``TLAST`` boundary handling requires top-level writer/interface support for packetized AXI stream types.
+The current implementation does not yet expose ``TLAST`` to layer kernels directly.
+
+For variable-length windows, a practical workaround is to keep the hls4ml core unchanged and perform ``TLAST`` to boundary conversion in a thin wrapper IP around the generated project.

docs/frontend/pytorch.rst

@@ -18,3 +18,6 @@ of the model. If the ``io_parallel`` I/O type (see :ref:`Concepts`) is used, a t
 Outputs are not transposed back by default, but in ``io_parallel`` case, a transpose node can be added. If not needed, these adjustments can also be switched off. See :py:class:`~hls4ml.utils.config.config_from_pytorch_model` for details.
 
 The equivalent of Keras extension API is not yet available for PyTorch parser, and will be provided in the future.
+
+.. note::
+    Experimental spiking layer support is available for selected modules. See :doc:`../advanced/snn` for details.

docs/index.rst

@@ -51,6 +51,7 @@
     advanced/precision
     advanced/fifo_depth
     advanced/extension
+    advanced/snn
     advanced/model_optimization
     advanced/bramfactor
     advanced/plugins

hls4ml/backends/vivado/passes/snn_templates.py

@@ -0,0 +1,128 @@
+from hls4ml.backends.template import FunctionCallTemplate, LayerConfigTemplate
+from hls4ml.model.layers import IFNeuron, LIFNeuron, SNNReadout
+
+if_config_template = """struct config{index} : nnet::if_neuron_config {{
+    static const unsigned n_in = {n_in};
+    static const unsigned n_out = {n_out};
+    static const unsigned io_type = nnet::{iotype};
+    static const unsigned window_size = {window_size};
+    static const bool threshold_is_vector = {threshold_is_vector};
+    static constexpr float threshold = {threshold};
+    static const nnet::snn_reset_mode reset_mode = nnet::snn_reset_mode::{reset_mechanism};
+    typedef {threshold_t.name} threshold_t;
+    typedef {membrane_t.name} membrane_t;
+}};\n"""
+
+if_function_template = 'nnet::if_neuron<{input_t}, {output_t}, {config}>({input}, {output}, {threshold});'
+snn_include_list = ['nnet_utils/nnet_snn.h', 'nnet_utils/nnet_snn_stream.h']
+
+
+class IFNeuronConfigTemplate(LayerConfigTemplate):
+    def __init__(self):
+        super().__init__(IFNeuron)
+        self.template = if_config_template
+
+    def format(self, node):
+        params = self._default_config_params(node)
+        params['threshold_is_vector'] = 'true' if node.get_attr('threshold_mode', 'scalar') == 'vector' else 'false'
+        return self.template.format(**params)
+
+
+class IFNeuronFunctionTemplate(FunctionCallTemplate):
+    def __init__(self):
+        super().__init__(IFNeuron, include_header=snn_include_list)
+        self.template = if_function_template
+
+    def format(self, node):
+        params = self._default_function_params(node)
+        params['threshold'] = (
+            node.get_weights('threshold_vec').name if node.get_attr('threshold_mode', 'scalar') == 'vector' else 'nullptr'
+        )
+        return self.template.format(**params)
+
+
+lif_config_template = """struct config{index} : nnet::lif_neuron_config {{
+    static const unsigned n_in = {n_in};
+    static const unsigned n_out = {n_out};
+    static const unsigned io_type = nnet::{iotype};
+    static const unsigned window_size = {window_size};
+    static const bool beta_is_vector = {beta_is_vector};
+    static const bool threshold_is_vector = {threshold_is_vector};
+    static constexpr float threshold = {threshold};
+    static constexpr float beta = {beta};
+    static const nnet::snn_reset_mode reset_mode = nnet::snn_reset_mode::{reset_mechanism};
+    typedef {beta_t.name} beta_t;
+    typedef {threshold_t.name} threshold_t;
+    typedef {membrane_t.name} membrane_t;
+}};\n"""
+
+lif_function_template = 'nnet::lif_neuron<{input_t}, {output_t}, {config}>({input}, {output}, {beta}, {threshold});'
+
+
+class LIFNeuronConfigTemplate(LayerConfigTemplate):
+    def __init__(self):
+        super().__init__(LIFNeuron)
+        self.template = lif_config_template
+
+    def format(self, node):
+        params = self._default_config_params(node)
+        params['beta_is_vector'] = 'true' if node.get_attr('beta_mode', 'scalar') == 'vector' else 'false'
+        params['threshold_is_vector'] = 'true' if node.get_attr('threshold_mode', 'scalar') == 'vector' else 'false'
+        return self.template.format(**params)
+
+
+class LIFNeuronFunctionTemplate(FunctionCallTemplate):
+    def __init__(self):
+        super().__init__(LIFNeuron, include_header=snn_include_list)
+        self.template = lif_function_template
+
+    def format(self, node):
+        params = self._default_function_params(node)
+        params['beta'] = node.get_weights('beta_vec').name if node.get_attr('beta_mode', 'scalar') == 'vector' else 'nullptr'
+        params['threshold'] = (
+            node.get_weights('threshold_vec').name if node.get_attr('threshold_mode', 'scalar') == 'vector' else 'nullptr'
+        )
+        return self.template.format(**params)
+
+
+readout_config_template = """struct config{index} : nnet::snn_readout_config {{
+    static const unsigned n_classes = {n_classes};
+    static const unsigned io_type = nnet::{iotype};
+    static const unsigned window_size = {window_size};
+    static const unsigned class_threshold = {class_threshold};
+    static constexpr float beta = {beta};
+    static const nnet::snn_readout_mode output_mode = nnet::snn_readout_mode::{output_mode};
+    static const nnet::snn_decision_rule decision_rule = nnet::snn_decision_rule::{decision_rule};
+    typedef {membrane_t.name} membrane_t;
+}};\n"""
+
+readout_function_template = 'nnet::snn_readout<{input_t}, {output_t}, {config}>({input}, {output});'
+
+
+class SNNReadoutConfigTemplate(LayerConfigTemplate):
+    def __init__(self):
+        super().__init__(SNNReadout)
+        self.template = readout_config_template
+
+    def format(self, node):
+        params = self._default_config_params(node)
+        return self.template.format(**params)
+
+
+class SNNReadoutFunctionTemplate(FunctionCallTemplate):
+    def __init__(self):
+        super().__init__(SNNReadout, include_header=snn_include_list)
+        self.template = readout_function_template
+
+    def format(self, node):
+        params = self._default_function_params(node)
+        return self.template.format(**params)
+
+
+def register_snn_templates(backend):
+    backend.register_template(IFNeuronConfigTemplate)
+    backend.register_template(IFNeuronFunctionTemplate)
+    backend.register_template(LIFNeuronConfigTemplate)
+    backend.register_template(LIFNeuronFunctionTemplate)
+    backend.register_template(SNNReadoutConfigTemplate)
+    backend.register_template(SNNReadoutFunctionTemplate)