Sinabs

------

* Fix typos and broken links in the documentation
* Move instructions for adding new device to FAQ
* Move tutorials from speck to top level
* Fix build of documentation with new sphinx version
* Update tests to new API
* Upgrade black to v24.1.0
* Remove ONNX support in favor of NIR
* Add notebooks to ignore execution list
* Fix unit test to check for deprecation warnings correctly
* Include firing rate analysis in synop loss tutorial
* Add deprecation warnings to SNNAnalyzer and SynOpsCouner
* Add missing typehint import
* Add tutorial about nir deployment on speck
* Add NIR api to docs
* Add test for extended readout layer
* Add test for NIR submodule
* Add test for raster to events
* Merge DynapCNN and Sinabs **breaking change**
* Update synop loss tutorials for snns and anns
* Ensure connection map for synop model hook is on same device as input
* Add test for synop model hook on cuda
* Add unit tests for firing rate and input diff hooks
* Speed up synops hook
* Add unit tests for model-to-layer conversion
* Allow ignoring layers when converting to dynapcnn network
* Add unit tests for synops hooks
* Add import to Visualizer back
* Add hook for novel 'diff input' metric
* Remove easy-install command
* Add graphviz dependency

-------

* Update codecov.yml
* Update ci-pipeline.yml
* Create codecov.yml
* Update ci-pipeline.yml
* Update ci-pipeline.yml
* Update ci-pipeline.yml
* Fix wrong variable name
* Update .gitlab-ci.yml file
* Add installation of pytest
* Update docker image to be used in the pipeline
* Add modifications to Authors and ChangeLog
* corrected data formats while loading nir
* added ignore dims to extract nir
* flatten layer dims updated to NIR speck
* added in input\_shape
* Install CPU version of torch for CI pipeline
* Update .readthedocs.yaml
* added method to documents
* added method to change batch size
* lower threshold equal to threshold while loading
* added identity transformations for IO
* removed debug message
* bug fix in stride of sumpool
* added citation to readme file
* added cff file
* set default value to 1 for stride in SumPool2d
* updated nir to support Flatten and SumPool2d

------

* Add unit test to ensure that subract value of MembraneSubtract can be a tensor
* Correct test whether subtract value is None
* add figure for input data flow
* fix confusion doc string of func ChipFactory.xytp\_to\_events about the key of the structured array
* update error in available pooling size
* add api indexing for specksim and visualizer
* add indexing for specksim
* fix typos and invalid url links
* add play with dvs on speck notebook
* Update requirements.txt
* Update ChangeLog to resolve conflict
* Unit test for correct duration of forward pass
* Record for at least the duration of the input
* fix typo in the quick start notebook
* commit jupyter notebook change
* add description for dvs visualization
* add N-MNIST quick start tutorial notebook
* complete contents for training tips section
* add items for device management
* Only publish pre-release when on develop branch
* device management
* fix terminology for samna configuration obj
* fix invalid urls in the docs
* Additional information on the architecture restrictions in the documentation
* sinabs/backend/dynapcnn/specksim.py
* Update the markdown with state monitoring
* Fixed the reset states test
* Changed the forward pass to make sure we receive all the events as fast as possible
* fix warnings related with overview.md can not be found when build doc
* Specksim documentation added
* Add tests to the monitoring
* Change readout and add a delay. (WIP: This will be reworked.)
* Several bugfixes, inclusion of the ability to reset network states and ability to monitor internal events
* Add unit test for conversion from events to raster
* Change order of dimensions in events\_to\_raster output to TxCxWxH
* Rename cpp binding functionality test
* Update gesture\_viz to work with the new structure
* Change the MNIST example to work with the new structure
* Fix rasterization time step
* Add unit test for monitor='all'
* change name from examples/dynapcnn\_visualizer to examples/visualizer update the visualizer documentation
* examples/mnist/specksim\_network.py
* Restructure the MNIST example
* Restructure examples
* Move the test specksim network under examples
* Divide examples into subfolders
* Support DynapcnnNetwork conversion. Fix the bugs related to AvgPool2d conversion
* Fix bug when monitor=all and dvs layer is present. Clarify some docstrings
* Bugfixes based on the MNIST script. Still not seeing any output
* Implementation and API is complete. Not tested yet
* Initial commit on the structure
* Namespace change on the tests
* add output monitoring introction for FAQs
* add example of implementing Residual Block
* add example of implementing Residual Block
* reform the FAQ section of the doc
* Tests related to specksim cpp bindings are implemented and added
* deal with v\_leak / r of size 0
* use nirtorch.load and use str indexing where possible
* added Conv2d to NIR
* remove superfluous comments
* use deprecated torch.testing.assert\_allclose to make tests for older pytorch pass
* add nir IF support
* updated Sinabs to latest NIR

-----

* add nir and nirtorch to requirements
* implemented sequential for from\_nir
* add test for to\_nir
* deleted old files
* added channel shift layer
* removed graph (migrated to nirtorch)
* Add missing nodes
* minor fixes in node conversion
* wip: restructure code
* wip: continue import from nir
* wip conversion from nir to sinabs
* Fix issue 99, add unit test

------

* change nmnist tutorial title
* added NMNIST tutorial
* modified Add to Morph
* function call bug fix in ignore\_nodes
* back to previous version
* back to previous code
* refactor code, change the extended method from encrypted function to extra function
* fixed variable assignment issue
* modify the documentation for extend method
* fixed some typo
* remove unneeded package importing
* add readout layer extend for Speck2E and 2F
* update ChangeLog
* blackened
* added method to ignore tensors in graph
* Unit test for periodic exponential gradient
* Generate periodic exponential surrogate gradient correctly for different thresholds. Solves issue 97
* removed torchvision from test requirements
* using cat instead of concat
* added doc string
* added assert
* balckened
* added optional model name to ignore it from graph
* added test for branched SNN
* added modules for addition and concatenation
* added convenience method for graph extraction
* removed torchview imports
* added sensible tests conditions in place of prints
* only saving index of last used tensor id
* Added methods to simplify the graph
* updated graph definition
* removing redundant incoming nodes attribute
* added context manager
* wip. basic graph exteaction and tracing added
* Warning about running the script for MacOS users added to the documentation
* minor changes and corrections
* Updated info on biases/leak
* added how to list connected devices
* Fix plot axes labels
* Removed commented kwargs from methods
* updated files
* Add unit test for DynapcnnNetwork.is\_compatible\_with
* Bugfix in DynapcnnNetwork.is\_compatible\_with where it would raise a ValueError instead of returning False
* Added documentaion in FAQ
* added is\_compatible\_with method; blackened
* updated DynapcnnVisualizer connect method
* moved configuration settings order
* blacked and updaetd test
* updated dynapcnn visualizer with latest samna api
* wip, dynapcnn visualizer refractoring
* wip updating dynapcnn visualizer
* blackened; updated to use run\_vusualizer method
* Added method for launching samna visualizer
* Updated notebook to the new samna graph api
* complete the neuron-leak example notebook
* add leak-neuron link in the FAQs
* merge develop to 76-add-leak-neruon-option
* fixed links to static images in notebooks
* fixed links
* shuffled the order of documentation
* setup file params use \_ instead of - now
* Fixed link to samna
* Make sure DVSLayer is added to DynapcnnNetwork if and only if necessary. Fix unit test about single neuron on hardware
* Turn off DVS sensor when DynapcnnNetwork has no DVSLayer
* Restructure unit tests
* Fix: Allow input shape to DVS layer to have 2 channels
* More flexible DVSLayer generation
* Remove 'compatible\_layers' attribute from DynapcnnNetwork and use 'sequence' instead
* documentation improvement
* replace curl with wget
* update ChangeLog
* replace all math equation
* Added board resetting before second test. Note that the device will be opened already before the second test starts, so it is not actually necessary
* Implemented resetting the boards before running the test
* add getting started with speck section
* Add none type comparison to spike\_threshold and min\_v\_mem
* Updated dangers.md with suggestion of using MultiSpike for training models to be deployed to chip
* skip visualizer test
* warn if layer with pooling is monitored
* add speck image at the doc homepage
* Add notes in docstrings and in the documentation about the fact that attribute modifications after the call have no effects
* specify inline math latex for Overview tutorial
* make tutorial notebook that contains latex a markdown documentsince it doesn't contain any code
* add skipif condition for test-case that need hardware connected to the CI machine
* update changelog
* add neuron leak tutorial notebook
* fix bug for neuron address mapping and add unit test case
* update ChangeLog
* add neruon leak option and related unit test
* remove system package install cmd
* fix add-apt-repository can't be found error
* add apt repository for graphviz localization
* try another way install graphviz
* fix CI pipeline failure

------

* make contact title bold like the others
* add contact section to documentation
* checked tests pass on mac