Skip to content
Snippets Groups Projects
Commit 181d0b8f authored by Michele Alberti's avatar Michele Alberti
Browse files

[release] Correct some typos and spacing in CHANGES.md file.

parent 0c10669b
No related branches found
No related tags found
No related merge requests found
# 0.2 (19-06-2023) # 0.2 (19-06-2023)
- [prover] Integration of the [nnenum](https://github.com/stanleybak/nnenum) prover. - [prover] Integration of the [nnenum](https://github.com/stanleybak/nnenum)
- [prover] Integration of the [$\alpha-\beta-$CROWN](https://github.com/stanleybak/nnenum) prover. prover.
- [prover] Integration of the
[$\alpha-\beta-$CROWN](https://github.com/stanleybak/nnenum) prover.
- [prover] Integration of the AIMOS metamorphic testing prover. - [prover] Integration of the AIMOS metamorphic testing prover.
- [prover] Add printer for VNN-LIB format for property specification as supported by nnenum and $\alpha-\beta-$CROWN provers.
- [prover] Support for multiple configurations of provers, using the prover-altern command line option. - [prover] Add printer for VNN-LIB format for property specification as
An example of registering an alternate configuration is available under config/caisar-detection.conf supported by nnenum and $\alpha-\beta-$CROWN provers.
- [prover] Add transformation to translate ONNX format into SMTLIB format for allowing SMTLIB2 compliant provers to work on neural networks.
- [doc] The first version of the CAISAR manual is available - [prover] Support for multiple configurations of provers, using the
publicly on our [website](caisar-platform.com/documentation). prover-altern command line option. An example of registering an alternate
It includes detailed installation instructions, configuration is available under config/caisar-detection.conf
examples on the ACAS-Xu and MNIST local
robustness benchmarks. Those examples use an experimental - [prover] Add transformation to translate ONNX format into SMTLIB format for
interpretation language that is not fully documented yet. allowing SMTLIB2 compliant provers to work on neural networks.
- [deps] Upgraded Why3 to 1.6.0 version
- [doc] The first version of the CAISAR manual is publicly available on our
[website](caisar-platform.com/documentation). It includes detailed
installation instructions, examples on the ACAS-Xu and MNIST local robustness
benchmarks. Those examples use an experimental interpretation language that is
not fully documented yet.
- [deps] Upgraded Why3 to 1.6.0 version.
- Add verify-json command for verifying a robustness property via a JSON - Add verify-json command for verifying a robustness property via a JSON
configuration file. configuration file.
- Add verification of datasets for classification tasks in terms of a specific - Add verification of datasets for classification tasks in terms of a specific
CSV format: each line provides the label in 1st column, and data features in CSV format: each line provides the label in 1st column, and data features in
the other columns. the other columns.
- Add builtin implementations of min-max scaling and z-score normalization for data normalization.
- Add support for memory limitation and timeout. - Add support for memory and time limits.
- Add debug logging options. - Add debug logging options.
# 0.1 (13-07-2022) # 0.1 (13-07-2022)
First public release of CAISAR. - First public release of CAISAR.
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment