Krzysztof Rymski f56d18dd68 Improvements to inference using int8 compressed kv's ... Multiplication is done using int16*int16 multiplication instructions avoid expensive conversion to f32/bf16 x2 speed on zen3 PiperOrigin-RevId: 888690192		2026-03-24 08:51:30 -07:00
..
python	Add ToFloatSlow, move RandomFloat to test_util	2025-11-27 00:14:51 -08:00
BUILD.bazel	Abort if args are unrecognized, refactor argument passing	2025-12-15 03:18:45 -08:00
compress-inl.h	Fix int8	2026-03-12 08:43:18 -07:00
compress.cc	Minor cleanup, on-demand NUQ buffer allocation	2025-04-16 10:49:43 -07:00
compress.h	Minor: rename compression/shared -> types.h	2025-05-13 06:53:21 -07:00
compress_test.cc	Int8 + microscaling support for kv cache formats.	2026-03-09 02:50:08 -07:00
distortion.h	Refactor/cleanup, remove even_odd	2024-09-04 09:25:13 -07:00
distortion_test.cc	Minor: rename compression/shared -> types.h	2025-05-13 06:53:21 -07:00
int-inl.h	Build fix for Arm SVE (explicit namespace qualification)	2026-01-08 13:29:45 -08:00
int_test.cc	Add 8-bit integer quantization (I8Stream) to Gemma.cpp.	2025-10-15 09:25:20 -07:00
nuq-inl.h	Fix NibbleCodec for AVX3_{ZEN4,DL,SPR}	2025-11-11 11:31:25 -08:00
nuq_test.cc	Replace mt19937 with new generator to enable parallel sampling	2025-09-04 23:49:10 -07:00
sfp-inl.h	Minor: rename compression/shared -> types.h	2025-05-13 06:53:21 -07:00
sfp_test.cc	Add ToFloatSlow, move RandomFloat to test_util	2025-11-27 00:14:51 -08:00
test_util-inl.h	Internal change	2026-02-23 08:55:33 -08:00
types.h	Improvements to inference using int8 compressed kv's	2026-03-24 08:51:30 -07:00