lava-dl is a library of deep learning tools within Lava that support offline training, online training and inference methods for various Deep Event-Based Networks.
There are two main strategies for training Deep Event-Based Networks: direct training and ANN to SNN converison.
Directly training the network utilizes the information of precise timing of events. Direct training is very accurate and results in efficient networks. However, directly training networks take a lot of time and resources.
On the other hand, ANN to SNN conversion is especially suitable for rate coded SNNs where we can leverage fast training of ANNs. These converted SNNs, however, typically require increased latency compared to directly trained SNNs.
Lava-DL provides an improved version of SLAYER for direct training of deep event based networks and a new ANN-SNN accelerated training approach called Bootstrap to mitigate high latency issue of conventional ANN-SNN methods for training Deep Event-Based Networks.
The lava-dl training libraries are independent of the core lava library since Lava Processes cannot be trained directly at this point. Instead, lava-dl is first used to train the model which can then be converted to a network of Lava processes using the netx library using platform independent hdf5 network description.
The library presently consists of
lava.lib.dl.slayerfor natively training Deep Event-Based Networks.lava.lib.dl.bootstrapfor training rate coded SNNs.lava.lib.dl.netxfor training and deployment of event-based deep neural networks on traditional as well as neuromorphic backends.
More tools will be added in the future.
Typical Lava-DL workflow:
- Training: using
lava.lib.dl.{slayer/bootstrap}which results in a hdf5 network description. Training usually follows an iterative cycle of architecture design, hyperparameter tuning, and backpropagation training. - Inference: using
lava.lib.dl.netxwhich generates lava proces from the hdf5 network description of the trained network and enables inference on different backends.
Note: We assume you have already setup Lava with virtual environment. Make sure PYTHONPATH contains path to Lava core library first.
- Linux/MacOS:
echo $PYTHONPATH - Windows:
echo %PYTHONPATH%
The output should contain something like this /home/user/lava
cd $HOME
git clone [email protected]:lava-dl/lava-dl.git
cd lava-dl
pip install "poetry>=1.1.13"
poetry config virtualenvs.in-project true
poetry install
source .venv/bin/activate
pytest# Commands using PowerShell
cd $HOME
git clone git@github.com:lava-dl/lava-dl.git
cd lava-dl
python3 -m venv .venv
.venv\Scripts\activate
pip install -U pip
pip install "poetry>=1.1.13"
poetry config virtualenvs.in-project true
poetry install
pytestYou should expect the following output after running the unit tests:
$ pytest
============================= test session starts ==============================
platform linux -- Python 3.9.10, pytest-7.0.1, pluggy-1.0.0
rootdir: /home/user/lava-dl, configfile: pyproject.toml, testpaths: tests
plugins: cov-3.0.0
collected 86 items
tests/lava/lib/dl/netx/test_blocks.py ... [ 3%]
tests/lava/lib/dl/netx/test_hdf5.py ... [ 6%]
tests/lava/lib/dl/slayer/neuron/test_adrf.py ....... [ 15%]
...... pytest output ...
tests/lava/lib/dl/slayer/neuron/dynamics/test_adaptive_threshold.py .... [ 80%]
. [ 81%]
tests/lava/lib/dl/slayer/neuron/dynamics/test_leaky_integrator.py ..... [ 87%]
tests/lava/lib/dl/slayer/neuron/dynamics/test_resonator.py ..... [ 93%]
tests/lava/lib/dl/slayer/utils/filter/test_conv_filter.py .. [ 95%]
tests/lava/lib/dl/slayer/utils/time/test_replicate.py . [ 96%]
tests/lava/lib/dl/slayer/utils/time/test_shift.py ... [100%]
=============================== warnings summary ===============================
...... pytest output ...
src/lava/lib/dl/slayer/utils/time/__init__.py 4 0 100%
src/lava/lib/dl/slayer/utils/time/replicate.py 6 0 100%
src/lava/lib/dl/slayer/utils/time/shift.py 59 16 73% 22-43, 50, 55, 75, 121, 128, 135, 139
src/lava/lib/dl/slayer/utils/utils.py 13 8 38% 14, 35-45
--------------------------------------------------------------------------------------------
TOTAL 4782 2535 47%
Required test coverage of 45.0% reached. Total coverage: 46.99%
======================= 86 passed, 3 warnings in 46.56s ========================
If you only need the lava-dl package in your python environment, we will publish Lava releases via GitHub Releases. Please download the package and install it.
Open a python terminal and run:
$ python3 -m venv python3_venv
$ pip install -U pip
$ pip install lava-dl-0.2.0.tar.gzEnd to end training tutorials
- Oxford spike train regression
- MNIST digit classification
- NMNIST digit classification
- PilotNet steering angle prediction
Deep dive training tutorials
Inference tutorials
lava.lib.dl.slayer is an enhanced version of SLAYER. Most noteworthy enhancements are: support for recurrent network structures, a wider variety of neuron models and synaptic connections (a complete list of features is here). This version of SLAYER is built on top of the PyTorch deep learning framework, similar to its predecessor. For smooth integration with Lava, lava.lib.dl.slayer supports exporting trained models using the platform independent hdf5 network exchange format.
In future versions, SLAYER will get completely integrated into Lava to train Lava Processes directly. This will eliminate the need for explicitly exporting and importing the trained networks.
Import modules
import lava.lib.dl.slayer as slayerNetwork Description
# like any standard pyTorch network
class Network(torch.nn.Module):
def __init__(self):
...
self.blocks = torch.nn.ModuleList([# sequential network blocks
slayer.block.sigma_delta.Input(sdnn_params),
slayer.block.sigma_delta.Conv(sdnn_params, 3, 24, 3),
slayer.block.sigma_delta.Conv(sdnn_params, 24, 36, 3),
slayer.block.rf_iz.Conv(rf_params, 36, 64, 3, delay=True),
slayer.block.rf_iz.Conv(sdnn_cnn_params, 64, 64, 3, delay=True),
slayer.block.rf_iz.Flatten(),
slayer.block.alif.Dense(alif_params, 64*40, 100, delay=True),
slayer.block.cuba.Recurrent(cuba_params, 100, 50),
slayer.block.cuba.KWTA(cuba_params, 50, 50, num_winners=5)
])
def forward(self, x):
for block in self.blocks:
# forward computation is as simple as calling the blocks in a loop
x = block(x)
return x
def export_hdf5(self, filename):
# network export to hdf5 format
h = h5py.File(filename, 'w')
layer = h.create_group('layer')
for i, b in enumerate(self.blocks):
b.export_hdf5(layer.create_group(f'{i}'))Training
net = Network()
assistant = slayer.utils.Assistant(net, error, optimizer, stats)
...
for epoch in range(epochs):
for i, (input, ground_truth) in enumerate(train_loader):
output = assistant.train(input, ground_truth)
...
for i, (input, ground_truth) in enumerate(test_loader):
output = assistant.test(input, ground_truth)
...Export the network
net.export_hdf5('network.net')In general ANN-SNN conversion methods for rate based SNN result in high latency of the network during inference. This is because the rate interpretation of a spiking neuron using ReLU acitvation unit breaks down for short inference times. As a result, the network requires many time steps per sample to achieve adequate inference results.
lava.lib.dl.bootstrap enables rapid training of rate based SNNs by translating them to an equivalent dynamic ANN representation which leads to SNN performance close to the equivalent ANN and low latency inference. More details here. It also supports hybrid training with a mixed ANN-SNN network to minimize the ANN to SNN performance gap. This method is independent of the SNN model being used.
It has similar API as lava.lib.dl.slayer and supports exporting trained models using the platform independent hdf5 network exchange format.
Import modules
import lava.lib.dl.bootstrap as bootstrapNetwork Description
# like any standard pyTorch network
class Network(torch.nn.Module):
def __init__(self):
...
self.blocks = torch.nn.ModuleList([# sequential network blocks
bootstrap.block.cuba.Input(sdnn_params),
bootstrap.block.cuba.Conv(sdnn_params, 3, 24, 3),
bootstrap.block.cuba.Conv(sdnn_params, 24, 36, 3),
bootstrap.block.cuba.Conv(rf_params, 36, 64, 3),
bootstrap.block.cuba.Conv(sdnn_cnn_params, 64, 64, 3),
bootstrap.block.cuba.Flatten(),
bootstrap.block.cuba.Dense(alif_params, 64*40, 100),
bootstrap.block.cuba.Dense(cuba_params, 100, 10),
])
def forward(self, x, mode):
...
for block, m in zip(self.blocks, mode):
x = block(x, mode=m)
return x
def export_hdf5(self, filename):
# network export to hdf5 format
h = h5py.File(filename, 'w')
layer = h.create_group('layer')
for i, b in enumerate(self.blocks):
b.export_hdf5(layer.create_group(f'{i}'))Training
net = Network()
scheduler = bootstrap.routine.Scheduler()
...
for epoch in range(epochs):
for i, (input, ground_truth) in enumerate(train_loader):
mode = scheduler.mode(epoch, i, net.training)
output = net.forward(input, mode)
...
loss.backward()
for i, (input, ground_truth) in enumerate(test_loader):
mode = scheduler.mode(epoch, i, net.training)
output = net.forward(input, mode)
...Export the network
net.export_hdf5('network.net')For inference using Lava, lava.lib.dl.netx provides an automated API for loading SLAYER-trained models as Lava Processes, which can be directly run on a desired backend. lava.lib.dl.netx imports models saved via SLAYER using the hdf5 network exchange format. The details of hdf5 network description specification can be found here.
Import modules
from lava.lib.dl.netx import hdf5Load the trained network
# Import the model as a Lava Process
net = hdf5.Network(net_config='network.net')Attach Processes for Input-Output interaction
from lava.proc import io
# Instantiate the processes
dataloader = io.dataloader.SpikeDataloader(dataset=test_set)
output_logger = io.sink.RingBuffer(shape=net.out_layer.shape, buffer=num_steps)
gt_logger = io.sink.RingBuffer(shape=(1,), buffer=num_steps)
# Connect the input to the network:
dataloader.ground_truth.connect(gt_logger.a_in)
dataloader.s_out.connect(net.in_layer.neuron.a_in)
# Connect network-output to the output process
net.out_layer.out.connect(output_logger.a_in)Run the network
from lava.magma import run_configs as rcfg
from lava.magma import run_conditions as rcnd
net.run(condition=rcnd.RunSteps(total_run_time), run_cfg=rcfg.Loihi1SimCfg())
output = output_logger.data.get()
gts = gt_logger.data.get()
net.stop()