Actions: f-dangel/curvlinops
Actions
605 workflow runs
605 workflow runs
bfloat16
to float32
before calling .numpy
Test
#829:
Commit 5ed730d
pushed
by
f-dangel
tol
for power iteration
Test
#822:
Commit 683ea1a
pushed
by
f-dangel
gradient_and_loss
function from CurvatureLinearOperator
Test
#815:
Commit bfa1ad9
pushed
by
f-dangel
torch_matvec
and torch_matmat
Test
#809:
Commit 7b7f746
pushed
by
f-dangel