tree 4d9b1fd9d8433e1ff5a993d2d4c7f06bf2a23e4a
parent aa8d896a77995ddf35fc50d9e06e5e121d047610
author Aart Bik <ajcbik@google.com> 1720640345 -0700
committer GitHub <noreply@github.com> 1720640345 -0700
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsFcBAABCAAQBQJmjuNZCRC1aQ7uu5UhlAAA5hoQAHokLNINm3UtwoYPe/l9Zmur
 fnZEPvAC0MXeiwi9ql+7SLKhjXQIN+zkftdsAcZa7rtLBzX7Nf34PwlX7W60D7Xd
 EcU4gSwZKTeQnw8T2sX4xWxKgvoEtv7AkhNqRw62X9EvxExF4FczOF+Q6lt4l5Jk
 3qnZZ0fvYSorjc1vla0Z9OV+k58EgMSXcePxwKSsllLamZFzu0ZrUqT1DOVlt5Rd
 qST6+MhgS4hJarlf2P2+OVV0nbYc1akHFSLGTwY0Uk8K05b/V/uo3M+vBaNkKvUX
 bAh4oBXTip3ceI9tjy27Ps+k6E3d2UTI2/Hyrkcbof8mUjlIBQmLACrMeielo8XZ
 S2eam33X+skNP8VR7XMkem/cJkBQOZQC1Zzof/nrKj1x8ACofpIKiF7kFF9jRdH0
 hc7pqMgLcd4hREeDE4NRZoR599MoTKibp0/gO7PpIa/SVzemqFeeqRGhpC3nKfd4
 ODIc64gSt6hGfogr7B2VgdyQPlgAxWxA1Iwc+0fYybPyrK0jlENE1bS+Sw6BkzYa
 N2xzbcAOriYOCu75nWzznWmJIZ7UV+cWcnFsgTrh4RLbIfw909R4pXGyWNlB2fNA
 Nbcwpiv293/UjIv1NVNvu/uygw7QcGOkhk4XQlyS5h05IhXaXTsje6p+lysabT1D
 rigm3+RBir1xFBFshNJV
 =rKWh
 -----END PGP SIGNATURE-----
 

[mpact][compiler] add training loop to models with simple test (#60)

* [mpact][compiler] add training loop to models with simple test

Note that although MPACT currently does not support autograd yet,
eventually we need to support this too. The current PR adds a very
simple training loop to the models, together with a simple neural
network that uses the training loop to learn classification of
simple sparse/dense tensors in a toy training set.

* linter for darker (I tested with black?!)