test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim32-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim40-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim48-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim56-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim64-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim72-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim80-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim88-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim96-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim120-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim128-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-256-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-512-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-2048-1024-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-2048-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-2048-8192-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-4096-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-4096-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-4096-8192-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim8-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-8192-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-8192-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim16-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim96-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim24-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim32-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim40-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim8-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim48-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim16-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim24-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim56-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim64-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim120-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim32-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim40-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim48-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim72-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim128-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim8-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim56-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim80-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOff-8192-8192-hdim88-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim16-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim64-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim24-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim72-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim32-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim40-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim48-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim56-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim64-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim72-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim80-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim88-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim96-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim120-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOff-8192-8192-hdim128-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-256-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-512-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-512-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-1024-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-4096-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim8-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim16-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim24-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim32-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim40-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim80-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim48-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim88-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim56-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim64-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim96-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim8-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim16-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim120-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOff-8192-8192-hdim128-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-256-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim24-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim32-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim40-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim48-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim56-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim64-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim72-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim80-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim88-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim96-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim120-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-8192-8192-hdim128-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim72-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim80-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim88-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim96-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim120-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim128-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-256-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-512-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-512-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-512-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-1024-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-4096-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim8-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-1024-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim8-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim16-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim24-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim32-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim16-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim24-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim40-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim32-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim48-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim40-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim56-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim48-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim56-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim8-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim64-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim72-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim80-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim16-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim88-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim96-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim120-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOn-8192-8192-hdim128-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype2-0.5-CausalOff-8192-8192-hdim64-5-3] (call)	OOM
test/test_backward.py::test_gqa[Split-False-l1-dtype1-0.5-CausalOff-579-2048-hdim248-N_HEADS1-3] (call)	OOM
test/test_backward.py::test_gqa[Split-False-l1-dtype1-0.5-CausalOff-2048-143-hdim248-N_HEADS0-3] (call)	OOM
test/test_backward.py::test_gqa[Split-False-l1-dtype1-0.5-CausalOff-2048-2048-hdim248-N_HEADS0-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-2081-4919-hdim248-5-3] (call)	OOM
test/test_backward.py::test_gqa[Split-False-l1-dtype1-0.5-CausalOff-2048-2048-hdim248-N_HEADS1-3] (call)	OOM
test/test_backward.py::test_gqa[Split-False-l1-dtype2-0.5-CausalOn-579-2048-hdim248-N_HEADS0-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-5237-1033-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-8192-8192-hdim24-5-3] (call)	OOM
test/test_backward.py::test_gqa[Split-False-l1-dtype2-0.5-CausalOn-579-2048-hdim248-N_HEADS1-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-5237-2063-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-5237-4919-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype2-0.5-CausalOn-2081-1033-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim72-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim80-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-337-2063-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-571-4919-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-1063-523-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-1063-4919-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-2081-1033-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-2081-2063-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-2081-4919-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-5237-2063-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-5237-4919-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim88-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim96-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim120-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype2-0.5-CausalOn-8192-8192-hdim128-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-2048-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-2048-8192-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-4096-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-4096-4096-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-4096-8192-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-8192-2048-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-2081-2063-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-5237-4919-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype2-0.5-CausalOn-337-2063-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype2-0.5-CausalOn-571-1033-hdim248-5-3] (call)	OOM
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype2-0.5-CausalOn-2081-1033-hdim248-5-3] (call)	OOM
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-256-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-512-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-512-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-1024-512-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-512-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-512-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-1024-256-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-2048-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-1024-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-1024-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-1024-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-2048-128-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-2048-256-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-2048-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-2048-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-2048-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-2048-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-2048-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-4096-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-2048-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-4096-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-2048-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-4096-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-4096-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-4096-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-4096-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-512-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-8192-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype0-0.5-CausalOff-1024-512-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-4096-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-256-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-256-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-256-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype1-0.5-CausalOff-512-512-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-4096-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-4096-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-8192-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-8192-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype0-0.5-CausalOff-8192-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-256-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-256-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-256-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-512-512-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-512-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-512-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-1024-256-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-1024-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-1024-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-1024-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-2048-128-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-2048-256-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-2048-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-2048-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype1-0.5-CausalOff-2048-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-1024-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-1024-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-2048-128-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-2048-256-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-2048-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-2048-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-2048-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-2048-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-4096-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-4096-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-4096-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-8192-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-256-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-512-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-512-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-1024-512-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-2048-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-2048-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-2048-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-4096-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-4096-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-4096-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-8192-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-256-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-256-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-256-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-512-512-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-512-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-512-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-1024-256-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-1024-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-1024-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-8192-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-1024-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-2048-128-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-2048-256-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-2048-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-2048-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-2048-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-2048-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-4096-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-4096-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-4096-8192-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype1-0.5-CausalOff-8192-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype0-0.5-CausalOff-8192-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-256-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-256-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-256-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-512-512-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-512-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-512-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-1024-256-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l1-dtype1-0.5-CausalOff-1024-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_gqa[Split-False-l2-dtype1-0.5-CausalOff-2048-143-hdim248-N_HEADS1-3] (call)	NAN
test/test_backward.py::test_gqa[Split-False-l2-dtype1-0.5-CausalOff-2048-2048-hdim248-N_HEADS0-3] (call)	NAN
test/test_backward.py::test_gqa[Split-False-l2-dtype1-0.5-CausalOff-2048-2048-hdim248-N_HEADS1-3] (call)	NAN
test/test_backward.py::test_gqa[Split-False-l1-dtype1-0.5-CausalOff-2048-143-hdim248-N_HEADS1-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-223-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-337-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_gqa[Split-False-l2-dtype1-0.5-CausalOff-579-2048-hdim248-N_HEADS1-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-337-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_gqa[Split-False-l2-dtype1-0.5-CausalOff-2048-143-hdim248-N_HEADS0-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-1063-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-2081-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-571-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-571-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-2081-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-571-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-1063-67-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-1063-157-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-1063-523-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-2081-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-1063-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-1063-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-5237-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-5237-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-571-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-1063-157-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-1063-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-223-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-223-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-337-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-337-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-571-157-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-571-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-223-523-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-223-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-337-257-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-337-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-571-157-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-571-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-571-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-1063-67-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-1063-157-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-1063-523-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-1063-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-1063-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-223-523-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-2081-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-2081-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-5237-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-256-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-512-2048-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-512-4096-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-1024-512-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-True-l2-dtype0-0.5-CausalOff-2048-1024-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-223-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-223-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-337-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-337-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-571-157-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-571-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-1063-67-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-1063-157-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-337-257-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-1063-523-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-1063-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-1063-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-571-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-1063-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-2081-1033-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-2081-4919-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-5237-2063-hdim248-5-3] (call)	NAN
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-223-523-hdim248-5-3] (call)	NAN
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-256-4096-hdim248-5-3] (call)	{"adiff": 0.20015489339360404, "grads_adiff": [0.6824409769353413, 0.08474079879345497, 3.54301037059912e-05, 0.0]}
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-512-2048-hdim248-5-3] (call)	{"adiff": 0.17340913055245044, "grads_adiff": [0.6425962511132479, 0.04485006029574823, 1.253193262584773e-05, 0.0]}
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-512-4096-hdim248-5-3] (call)	{"adiff": 0.18029874641996202, "grads_adiff": [0.6345338762819732, 0.04802263580171506, 2.1622720017155928e-05, 0.0]}
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-1024-2048-hdim248-5-3] (call)	{"adiff": 0.1785553175082495, "grads_adiff": [0.6620021633757794, 0.029938592844443557, 9.612298077943393e-06, 0.0]}
test/test_backward.py::test_regular_bwd[Split-False-l2-dtype2-0.5-CausalOn-4096-2048-hdim248-5-3] (call)	{"adiff": 0.15719180530055488, "grads_adiff": [0.5663895855169662, 0.030292836046205474, 9.377420890466226e-06, 0.0]}
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-512-2048-hdim248-5-3] (call)	{"adiff": 0.1669282003326329, "grads_adiff": [0.03771064263632657, 0.0020036206266136726, 8.66260158138843e-06, 0.0]}
test/test_backward.py::test_regular_bwd[Split-False-l1-dtype2-0.5-CausalOn-4096-2048-hdim248-5-3] (call)	{"adiff": 0.16061849760576735, "grads_adiff": [0.0351832636222187, 0.0013726963678113782, 8.19814050689871e-06, 0.0]}
test/test_backward.py::test_gqa[Split-False-l2-dtype2-0.5-CausalOn-579-2048-hdim248-N_HEADS0-3] (call)	{"adiff": 0.17533733453839606, "grads_adiff": [0.6515664956476684, 0.053120247091798856, 3.690390485466821e-05, 0.0]}
test/test_backward.py::test_gqa[Split-False-l1-dtype1-0.5-CausalOff-579-143-hdim248-N_HEADS0-3] (call)	{"adiff": 120.87503051757812, "grads_adiff": [5791.99658203125, 493.9849853515625, 19.074329376220703, 0.0]}
test/test_backward.py::test_gqa[Split-False-l1-dtype1-0.5-CausalOff-579-143-hdim248-N_HEADS1-3] (call)	{"adiff": 118.88042449951172, "grads_adiff": [5856.0, 1048.010986328125, 37.322750091552734, 0.0]}
test/test_backward.py::test_gqa[Split-False-l1-dtype1-0.5-CausalOff-579-2048-hdim248-N_HEADS0-3] (call)	{"adiff": 121.86238861083984, "grads_adiff": [5920.00146484375, 6816.01953125, 205.19342041015625, 0.0]}
test/test_backward.py::test_gqa[Split-False-l2-dtype2-0.5-CausalOn-579-2048-hdim248-N_HEADS1-3] (call)	{"adiff": 0.1699317499159354, "grads_adiff": [0.6286548863488897, 0.07790125596731468, 0.00011670183872780626, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-223-1033-hdim248-5-3] (call)	{"adiff": 4.445913791656494, "grads_adiff": [7.589638710021973, 3.7546427249908447, 2.9212563037872314, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-337-1033-hdim248-5-3] (call)	{"adiff": 12.472078323364258, "grads_adiff": [61.72141647338867, 31.423587799072266, 10.042353630065918, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-2081-2063-hdim248-5-3] (call)	{"adiff": 1197.432373046875, "grads_adiff": [NaN, NaN, 332.9591064453125, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-337-2063-hdim248-5-3] (call)	{"adiff": 13.033522605895996, "grads_adiff": [62.99995803833008, 58.86396408081055, 18.446569442749023, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-571-257-hdim248-5-3] (call)	{"adiff": 12.39471435546875, "grads_adiff": [66.55937194824219, 5.910293102264404, 1.9693784713745117, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-1063-523-hdim248-5-3] (call)	{"adiff": 88.90946197509766, "grads_adiff": [3523.999755859375, 315.7469177246094, 14.512910842895508, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-1063-2063-hdim248-5-3] (call)	{"adiff": 91.08583068847656, "grads_adiff": [3510.0, 1166.9896240234375, 47.60105895996094, 0.0]}
test/test_backward.py::test_gqa[Split-False-l2-dtype1-0.5-CausalOff-579-143-hdim248-N_HEADS0-3] (call)	{"adiff": 134.40869140625, "grads_adiff": [121344.0078125, 13504.1455078125, 26.469541549682617, 0.0]}
test/test_backward.py::test_gqa[Split-False-l2-dtype1-0.5-CausalOff-579-143-hdim248-N_HEADS1-3] (call)	{"adiff": 129.3883056640625, "grads_adiff": [107007.9921875, 26879.623046875, 47.72488784790039, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-223-523-hdim248-5-3] (call)	{"adiff": 4.3600850105285645, "grads_adiff": [7.592602252960205, 1.9889391660690308, 1.535754680633545, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-223-2063-hdim248-5-3] (call)	{"adiff": 4.437762260437012, "grads_adiff": [8.133810043334961, 7.092381477355957, 4.933204174041748, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-337-257-hdim248-5-3] (call)	{"adiff": 12.761295318603516, "grads_adiff": [59.99742889404297, 9.058525085449219, 3.174912929534912, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-1063-4919-hdim248-5-3] (call)	{"adiff": 92.57709503173828, "grads_adiff": [3621.995849609375, 2718.007568359375, 105.12247467041016, 0.0]}
test/test_backward.py::test_gqa[Split-False-l2-dtype1-0.5-CausalOff-579-2048-hdim248-N_HEADS0-3] (call)	{"adiff": 149.4048309326172, "grads_adiff": [142336.0, 227327.0625, 356.2010498046875, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype0-0.5-CausalOff-2081-1033-hdim248-5-3] (call)	{"adiff": 1193.4395751953125, "grads_adiff": [NaN, NaN, 189.8391571044922, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype1-0.5-CausalOff-571-157-hdim248-5-3] (call)	{"adiff": 78.53720092773438, "grads_adiff": [66.50302124023438, 3.389111280441284, 1.276880145072937, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype2-0.5-CausalOn-571-1033-hdim248-5-3] (call)	{"adiff": 0.14951028274432487, "grads_adiff": [0.034434443955957225, 0.0013274186490902125, 4.991536516119055e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-223-1033-hdim248-5-3] (call)	{"adiff": 4.525271892547607, "grads_adiff": [123.05451202392578, 85.7261962890625, 3.7782468795776367, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-337-1033-hdim248-5-3] (call)	{"adiff": 13.25483226776123, "grads_adiff": [1126.005615234375, 763.0446166992188, 14.080952644348145, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-1063-4919-hdim248-5-3] (call)	{"adiff": 104.76374816894531, "grads_adiff": [Infinity, Infinity, 176.13133239746094, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-571-257-hdim248-5-3] (call)	{"adiff": 12.930094718933105, "grads_adiff": [1150.95166015625, 154.6791229248047, 2.6314384937286377, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-1063-523-hdim248-5-3] (call)	{"adiff": 97.81470489501953, "grads_adiff": [Infinity, 8807.9677734375, 23.050994873046875, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-2081-1033-hdim248-5-3] (call)	{"adiff": 1802.4647216796875, "grads_adiff": [NaN, NaN, 316.8727111816406, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype2-0.5-CausalOn-223-2063-hdim248-5-3] (call)	{"adiff": 0.18076703669854266, "grads_adiff": [0.03363647709013536, 0.004060464487759735, 1.2525058477486084e-05, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-1063-2063-hdim248-5-3] (call)	{"adiff": 108.1191177368164, "grads_adiff": [Infinity, 39359.73046875, 77.14659118652344, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-2081-2063-hdim248-5-3] (call)	{"adiff": 1836.45556640625, "grads_adiff": [NaN, NaN, 604.6317138671875, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-1063-67-hdim248-5-3] (call)	{"adiff": 92.9439468383789, "grads_adiff": [61183.9921875, 1144.0020751953125, 4.017114162445068, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype2-0.5-CausalOn-337-1033-hdim248-5-3] (call)	{"adiff": 0.16818575232821947, "grads_adiff": [0.03558964166390553, 0.0024261573712394055, 5.2858889345586135e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-1063-1033-hdim248-5-3] (call)	{"adiff": 99.95183563232422, "grads_adiff": [65535.98046875, 16895.92578125, 37.486839294433594, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l1-dtype2-0.5-CausalOn-337-2063-hdim248-5-3] (call)	{"adiff": 0.1707080446951964, "grads_adiff": [0.03559618492502229, 0.0019179752072506426, 9.842573445162373e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype1-0.5-CausalOff-337-257-hdim248-5-3] (call)	{"adiff": 13.38608455657959, "grads_adiff": [1088.0072021484375, 226.0345916748047, 4.364029407501221, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-2081-4919-hdim248-5-3] (call)	{"adiff": 1803.4644775390625, "grads_adiff": [NaN, NaN, 1372.04345703125, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype2-0.5-CausalOn-223-2063-hdim248-5-3] (call)	{"adiff": 0.1862605302665361, "grads_adiff": [0.6729442532730694, 0.053583462817752364, 1.4648277824491629e-05, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-337-2063-hdim248-5-3] (call)	{"adiff": 12.99111270904541, "grads_adiff": [64.67747497558594, 59.72299575805664, 18.44368553161621, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-571-257-hdim248-5-3] (call)	{"adiff": 12.466618537902832, "grads_adiff": [65.12035369873047, 5.7116217613220215, 1.9901319742202759, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-1063-523-hdim248-5-3] (call)	{"adiff": 89.72150421142578, "grads_adiff": [3485.998779296875, 313.4931335449219, 14.267143249511719, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-5237-1033-hdim248-5-3] (call)	{"adiff": 152.9676971435547, "grads_adiff": [Infinity, 9800.015625, 14.192126274108887, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype2-0.5-CausalOn-337-1033-hdim248-5-3] (call)	{"adiff": 0.1662994315378301, "grads_adiff": [0.5587013056622334, 0.04438827156946601, 6.4710278491197926e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-1063-2063-hdim248-5-3] (call)	{"adiff": 90.9024658203125, "grads_adiff": [3514.00244140625, 1158.9962158203125, 48.29338836669922, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype2-0.5-CausalOn-337-2063-hdim248-5-3] (call)	{"adiff": 0.1791898374612002, "grads_adiff": [0.5742219614048262, 0.051978012518899774, 1.2571444054643166e-05, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-1063-4919-hdim248-5-3] (call)	{"adiff": 90.34230041503906, "grads_adiff": [3579.9970703125, 2719.966552734375, 103.25476837158203, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-5237-2063-hdim248-5-3] (call)	{"adiff": 147.2229766845703, "grads_adiff": [Infinity, 19103.939453125, 26.98251724243164, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-2081-1033-hdim248-5-3] (call)	{"adiff": 1181.4449462890625, "grads_adiff": [NaN, NaN, 190.58868408203125, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype2-0.5-CausalOn-571-1033-hdim248-5-3] (call)	{"adiff": 0.15728074732797315, "grads_adiff": [0.5482990466054384, 0.040491597677908775, 6.237375450801608e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-2081-2063-hdim248-5-3] (call)	{"adiff": 1199.432373046875, "grads_adiff": [NaN, NaN, 333.7194519042969, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype2-0.5-CausalOn-2081-1033-hdim248-5-3] (call)	{"adiff": 0.15244527729322233, "grads_adiff": [0.5736883445989924, 0.04351589377637973, 6.230950343244501e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-223-1033-hdim248-5-3] (call)	{"adiff": 4.45311975479126, "grads_adiff": [7.66373348236084, 3.6760408878326416, 2.9202218055725098, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-337-1033-hdim248-5-3] (call)	{"adiff": 12.595605850219727, "grads_adiff": [64.12251281738281, 31.11073875427246, 9.960864067077637, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-223-1033-hdim248-5-3] (call)	{"adiff": 21.173797607421875, "grads_adiff": [9.370746612548828, 3.602043867111206, 2.9111015796661377, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-2081-4919-hdim248-5-3] (call)	{"adiff": 1195.4354248046875, "grads_adiff": [NaN, NaN, 736.73046875, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-337-1033-hdim248-5-3] (call)	{"adiff": 38.07740020751953, "grads_adiff": [61.25391387939453, 23.11181640625, 7.451603889465332, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-571-1033-hdim248-5-3] (call)	{"adiff": 24.03269386291504, "grads_adiff": [66.00072479248047, 16.000818252563477, 4.9672088623046875, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-1063-1033-hdim248-5-3] (call)	{"adiff": 93.04186248779297, "grads_adiff": [3456.001220703125, 463.9943542480469, 20.111583709716797, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-5237-1033-hdim248-5-3] (call)	{"adiff": 134.5836181640625, "grads_adiff": [8959.998046875, 334.7486572265625, 8.905959129333496, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-False-l2-dtype0-0.5-CausalOff-5237-4919-hdim248-5-3] (call)	{"adiff": 144.34437561035156, "grads_adiff": [Infinity, 40416.1171875, 55.5188102722168, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-2081-1033-hdim248-5-3] (call)	{"adiff": 1191.4395751953125, "grads_adiff": [638976.0, 39424.00390625, 138.72076416015625, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-5237-2063-hdim248-5-3] (call)	{"adiff": 135.20657348632812, "grads_adiff": [9096.0, 637.0072021484375, 15.98451042175293, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype2-0.5-CausalOn-571-1033-hdim248-5-3] (call)	{"adiff": 0.16653801656989808, "grads_adiff": [0.03491764768734624, 0.0014986443681774334, 5.69497042679501e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype2-0.5-CausalOn-2081-1033-hdim248-5-3] (call)	{"adiff": 0.16512287200188314, "grads_adiff": [0.0368673779276776, 0.0016772184173279928, 5.69715195197773e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-223-1033-hdim248-5-3] (call)	{"adiff": 4.496992588043213, "grads_adiff": [130.31471252441406, 88.9173355102539, 3.9426300525665283, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-337-1033-hdim248-5-3] (call)	{"adiff": 12.962925910949707, "grads_adiff": [1088.113525390625, 811.7464599609375, 13.695953369140625, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-1063-2063-hdim248-5-3] (call)	{"adiff": 104.62861633300781, "grads_adiff": [Infinity, 36352.66015625, 79.05854797363281, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype1-0.5-CausalOff-5237-4919-hdim248-5-3] (call)	{"adiff": 99840.515625, "grads_adiff": [917504.0, 1408.006591796875, 32.757530212402344, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype2-0.5-CausalOn-223-2063-hdim248-5-3] (call)	{"adiff": 0.17450350037249618, "grads_adiff": [0.041031789856926855, 0.0028171933838232466, 1.2157543087809586e-05, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-1063-4919-hdim248-5-3] (call)	{"adiff": 105.30293273925781, "grads_adiff": [Infinity, Infinity, 174.12643432617188, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype0-0.5-CausalOff-5237-4919-hdim248-5-3] (call)	{"adiff": 134.95779418945312, "grads_adiff": [9191.998046875, 1487.9996337890625, 34.848045349121094, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-337-2063-hdim248-5-3] (call)	{"adiff": 13.496801376342773, "grads_adiff": [1177.8331298828125, 1626.846923828125, 27.25987434387207, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-2081-1033-hdim248-5-3] (call)	{"adiff": 1891.4742431640625, "grads_adiff": [NaN, NaN, 329.3557434082031, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-571-257-hdim248-5-3] (call)	{"adiff": 13.159988403320312, "grads_adiff": [1119.968994140625, 152.2781982421875, 2.917426347732544, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype2-0.5-CausalOn-337-1033-hdim248-5-3] (call)	{"adiff": 0.16412890516941386, "grads_adiff": [0.03563390742622774, 0.001524224987692776, 5.182912305556897e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-1063-523-hdim248-5-3] (call)	{"adiff": 99.4869384765625, "grads_adiff": [Infinity, 9071.9208984375, 22.221437454223633, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-2081-2063-hdim248-5-3] (call)	{"adiff": 1759.4613037109375, "grads_adiff": [NaN, NaN, 594.1522827148438, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l1-dtype2-0.5-CausalOn-337-2063-hdim248-5-3] (call)	{"adiff": 0.1684949349454713, "grads_adiff": [0.037507796204058114, 0.002387794019745537, 1.0037088933323446e-05, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-5237-1033-hdim248-5-3] (call)	{"adiff": 142.7222442626953, "grads_adiff": [Infinity, 9463.9619140625, 13.926084518432617, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-2081-4919-hdim248-5-3] (call)	{"adiff": 1735.4595947265625, "grads_adiff": [NaN, NaN, 1311.1170654296875, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-5237-2063-hdim248-5-3] (call)	{"adiff": 144.72325134277344, "grads_adiff": [Infinity, 17760.18359375, 25.620296478271484, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype1-0.5-CausalOff-571-2063-hdim248-5-3] (call)	{"adiff": 13.427008628845215, "grads_adiff": [1199.9952392578125, 1128.091064453125, 17.43039321899414, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype2-0.5-CausalOn-337-1033-hdim248-5-3] (call)	{"adiff": 0.17706218346235364, "grads_adiff": [0.5612329307752161, 0.040944134123277465, 6.661512470529374e-06, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype2-0.5-CausalOn-223-2063-hdim248-5-3] (call)	{"adiff": 0.20389187230071426, "grads_adiff": [0.644698277527453, 0.07605776623279126, 1.575064458947395e-05, 0.0]}
test/test_backward.py::test_irregulars[Split-BiasOff-True-l2-dtype0-0.5-CausalOff-5237-4919-hdim248-5-3] (call)	{"adiff": 147.9653778076172, "grads_adiff": [Infinity, 40703.83984375, 56.673912048339844, 0.0]}
