 slaren
|
017e6999b5
add basic tensor data validation function (#6884)
|
1 year ago |
 Georgi Gerganov
|
54770413c4
ggml : fix MIN / MAX macros (#6904)
|
1 year ago |
|
Georgi Gerganov
|
c0d1b3e03e
ggml : move 32-bit arm compat in ggml-impl.h (#6865)
|
1 year ago |
 Justine Tunney
|
8cc91dc63c
ggml : add llamafile sgemm (#6414)
|
1 year ago |
 Carolinabanana
|
5dc9dd7152
llama : add Command R Plus support (#6491)
|
1 year ago |
 Kawrakow
|
cbc8343619
Make IQ1_M work for QK_K = 64 (#6327)
|
1 year ago |
|
Kawrakow
|
55c1b2a3bb
IQ1_M: 1.75 bpw quantization (#6302)
|
1 year ago |
 Justine Tunney
|
7733f0c760
ggml : support AVX512VNNI (#6280)
|
1 year ago |
|
Kawrakow
|
cfd3be76e3
ggml : same IQ4_NL quantization for CPU/CUDA/Metal (#6196)
|
1 year ago |
|
Georgi Gerganov
|
8030da7afe
ggml : reuse quantum structs across backends (#5943)
|
1 year ago |
|
Georgi Gerganov
|
184215e783
ggml : fix UB in IQ2_S and IQ3_S (#6012)
|
1 year ago |
|
Kawrakow
|
44ca159faf
1.5 bit: we can do even better (#5999)
|
1 year ago |
 Michael Podvitskiy
|
3202361c5b
ggml, ci : Windows ARM runner and build fixes (#5979)
|
1 year ago |
|
Kawrakow
|
be858f6205
Better 1.5 bit quantization (#5971)
|
1 year ago |
|
Georgi Gerganov
|
df4dc3e7cb
ggml : try fix 32-bit arm compat (whisper/1938)
|
1 year ago |
|
Georgi Gerganov
|
8380ecfb21
ggml : fix unnecessary f32 -> f16 -> f32 casts (mmla) (#5951)
|
1 year ago |
|
Georgi Gerganov
|
5b09797321
ggml : remove old quantization functions (#5942)
|
1 year ago |
|
Georgi Gerganov
|
8a3012a4ad
ggml : add ggml-common.h to deduplicate shared code (#5940)
|
1 year ago |
 bobqianic
|
e25fb4b18f
ggml : use `uint8x16_t` return type for `ggml_vqtbl1q_u8` (#5894)
|
1 year ago |
 Jared Van Bortel
|
bd836944f8
quants : use MM256_SET_M128I consistently to fix gcc 7 build (#5889)
|
1 year ago |
|
Georgi Gerganov
|
494c870326
ggml : fix IQ3_S AVX implementation (#5834)
|
1 year ago |
|
Kawrakow
|
bbde6eb256
ggml : IQ3_S improvements (#5829)
|
1 year ago |
|
Kawrakow
|
7c4263d426
ggml : make i-quants work with super-blocks of 64 (CPU,Metal) (#5760)
|
1 year ago |
|
Kawrakow
|
cb49e0f8c9
Attempt to fix android build (#5752)
|
1 year ago |
|
Kawrakow
|
0becb22ac0
IQ4_XS: a 4.25 bpw quantization (#5747)
|
1 year ago |
 Engininja2
|
1f30b7a9f1
ggml-quants : fix avx2 iq1_s vec_dot when compiled with gcc (#5742)
|
1 year ago |
|
Kawrakow
|
a33e6a0d2a
Adding IQ2_S and IQ2_M to complete coverage of the 2-3 bit quantization range (#5721)
|
1 year ago |
 Radosław Gryta
|
abbabc5e51
ggml-quants : provide ggml_vqtbl1q_u8 for 64bit compatibility (#5711)
|
1 year ago |
|
Kawrakow
|
4c4cb30736
IQ3_S: a much better alternative to Q3_K (#5676)
|
1 year ago |
|
Georgi Gerganov
|
7e4f339c40
ggml : always define ggml_fp16_t as uint16_t (#5666)
|
1 year ago |
