The ModelCIMs framework has been developed for the training of CNNs/DNNs aware of SRAM-based compute in-memory accelerators non-idealities.
\ No newline at end of file
The ModelCIMs framework has been developed for the training of CNNs/DNNs aware of SRAM-based compute in-memory accelerators non-idealities.
--- Copyright/referencing ---
Usage of this framework is subject to copyright policies. Please cite the original research paper
"A. Kneip et al., A 1-to-4b 16.8-POPS/W 473-TOPS/mm2 6T-based In-Memory Computing SRAM in 22nm FD-SOI with Multi-Bit Analog Batch-Normalization, Proceedings of IEEE ESSCIRC 2022, pp. 1-4" when using it for academic or industrial purpose.