gemma.cpp/gemma
Jan Wassenberg 4f9155d8c6 Add bf16 matmul support, update naming+test
Avoid int32, which can easily overflow for large matrices.
Also fix IDE warning in sfp-inl.

PiperOrigin-RevId: 640149845
2024-06-04 07:41:46 -07:00
..
evals Add MMLU eval to github 2024-05-20 10:20:53 -07:00
activations.cc Add first version of backpropagation support. 2024-06-04 08:37:49 +00:00
activations.h Adress review comments 2024-06-04 08:37:54 +00:00
benchmark.cc Add MMLU eval to github 2024-05-20 10:20:53 -07:00
common-inl.h Adress review comments 2024-06-04 08:37:54 +00:00
common.h Add first version of backpropagation support. 2024-06-04 08:37:49 +00:00
compress_weights.cc Improve documentation for compress_weights flags 2024-04-29 06:49:50 -07:00
configs.h Add first version of backpropagation support. 2024-06-04 08:37:49 +00:00
gemma.cc Use HWY_EXPORT_AND_DYNAMIC_DISPATCH_T where possible. 2024-06-04 09:18:56 +00:00
gemma.h Add first version of backpropagation support. 2024-06-04 08:37:49 +00:00
gemma_test.cc Add first version of backpropagation support. 2024-06-04 08:37:49 +00:00
ops.h Add bf16 matmul support, update naming+test 2024-06-04 07:41:46 -07:00
ops_test.cc Add bf16 matmul support, update naming+test 2024-06-04 07:41:46 -07:00
run.cc Update to Highway 1.2 for topology/VQSelect 2024-05-31 12:29:10 -07:00
run_mmlu.cc Add MMLU eval to github 2024-05-20 10:20:53 -07:00
weights.cc Adress review comments 2024-06-04 08:37:54 +00:00
weights.h Adress review comments 2024-06-04 08:37:54 +00:00