happyz
/
gemma.cpp
mirror of https://github.com/google/gemma.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
gemma.cpp/ops
History
Jan Wassenberg 5e812f07f5 Use f64 Dot and sum in softmax - faster than Cascaded 
Also let the kernel specify the Raw and State types,
rename WeightT/VecT -> WT/VT.

PiperOrigin-RevId: 680464427
 		2024-09-30 01:22:09 -07:00
..
dot-inl.h Use f64 Dot and sum in softmax - faster than Cascaded 2024-09-30 01:22:09 -07:00
dot_test.cc Use f64 Dot and sum in softmax - faster than Cascaded 2024-09-30 01:22:09 -07:00
fp_arith-inl.h Cascaded summation for Softmax 2024-09-20 10:31:23 -07:00
gemma_matvec_test.cc Major compression update, arbitrary-len unpack + new Dot 2024-09-10 02:22:19 -07:00
matmul-inl.h Major compression update, arbitrary-len unpack + new Dot 2024-09-10 02:22:19 -07:00
matmul.h Major MatMul update, 1.9-2.3x speedup on Zen4 via bf16 mul 2024-08-16 07:52:20 -07:00
matmul_test.cc 1.6x speedup of MatMulSlow using compensated Dot 2024-09-26 02:42:53 -07:00
matvec-inl.h Major compression update, arbitrary-len unpack + new Dot 2024-09-10 02:22:19 -07:00
ops-inl.h Use f64 Dot and sum in softmax - faster than Cascaded 2024-09-30 01:22:09 -07:00
ops_test.cc Add tests for SampleTopK that highlight existing problems and fix those: 2024-09-18 10:32:01 -07:00