diff options
| author | Kawrakow <iwankawrakow@gmail.com> | 2025-02-06 14:08:52 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-02-06 14:08:52 +0200 |
| commit | 7f61b3068e18728e5e7e2b95546ff03dd2fd41ac (patch) | |
| tree | f175a942a6ebd2d2d8b08c46fa71d9f6fbad50e7 /ggml/include | |
| parent | a6f9f2ec9af92b5a13f035db054aac2fd2efaee7 (diff) | |
IQ1_M_R4: better 1.75 bpw quants (#187)
* iq1_m_r4: basics (quantize/dequantize)
* iq1_m_r4: Zen4 gemm
* iq1_m_r4: neon gemm
* iq1_m_r4: switch to q8_0_x4 also on AVX2/Zen4
With the deltas being per group of 8, we cannot make use
of the q8 sums stored in q8_1, so we get a tiny gain by
using q8_0_x4.
* iq1_m_r4: rename mul_mat_iq1_m_r4_q8_1 to mul_mat_iq1_m_r4_q8_0
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'ggml/include')
| -rw-r--r-- | ggml/include/ggml.h | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/ggml/include/ggml.h b/ggml/include/ggml.h index 9668dc32..77ac33a9 100644 --- a/ggml/include/ggml.h +++ b/ggml/include/ggml.h @@ -432,6 +432,7 @@ extern "C" { GGML_TYPE_IQ3_S_R4 = 221, GGML_TYPE_IQ2_S_R4 = 222, GGML_TYPE_IQ4_XS_R4 = 223, + GGML_TYPE_IQ1_M_R4 = 229, GGML_TYPE_BF16_R16 = 230, GGML_TYPE_Q6_0_R4 = 233, GGML_TYPE_IQ2_BN_R4 = 335, @@ -516,6 +517,7 @@ extern "C" { GGML_FTYPE_MOSTLY_IQ3_S_R4 = 220, // except 1d tensors GGML_FTYPE_MOSTLY_IQ2_S_R4 = 221, // except 1d tensors GGML_FTYPE_MOSTLY_IQ4_XS_R4 = 222, // except 1d tensors + GGML_FTYPE_MOSTLY_IQ1_M_R4 = 223, // except 1d tensors GGML_FTYPE_MOSTLY_BF16_R16 = 224, // except 1d tensors GGML_FTYPE_MOSTLY_Q6_0_R4 = 227, // except 1d tensors GGML_FTYPE_MOSTLY_IQ2_BN_R4 = 329, // except 1d tensors |