diff options
| author | Iwan Kawrakow <iwan.kawrakow@gmail.com> | 2024-07-29 12:38:46 +0300 |
|---|---|---|
| committer | Kawrakow <48489457+ikawrakow@users.noreply.github.com> | 2024-08-01 09:38:06 +0200 |
| commit | 5d341757bc73efee0deba07a17679a965037753b (patch) | |
| tree | 33db7a06b6b3ae1067a1d3db36192926705c678f /ggml/include | |
| parent | 06e255ac9da49cabde466b9ef8b3c09c0f8d8dd1 (diff) | |
iq5_k: Basics
Quantize/dequantize, CUDA dequantize
Diffstat (limited to 'ggml/include')
| -rw-r--r-- | ggml/include/ggml.h | 10 |
1 files changed, 6 insertions, 4 deletions
diff --git a/ggml/include/ggml.h b/ggml/include/ggml.h index 2cb4af32..b7585ad6 100644 --- a/ggml/include/ggml.h +++ b/ggml/include/ggml.h @@ -389,8 +389,9 @@ extern "C" { GGML_TYPE_IQ1_BN = 34, GGML_TYPE_IQ2_BN = 35, GGML_TYPE_Q8_K64 = 36, - GGML_TYPE_IQ4_K = 37, - GGML_TYPE_IQ2_K = 38, + GGML_TYPE_IQ2_K = 37, + GGML_TYPE_IQ4_K = 38, + GGML_TYPE_IQ5_K = 39, GGML_TYPE_COUNT, }; @@ -437,8 +438,9 @@ extern "C" { GGML_FTYPE_MOSTLY_Q4_0_8_8 = 27, // except 1d tensors GGML_FTYPE_MOSTLY_IQ1_BN = 28, // except 1d tensors GGML_FTYPE_MOSTLY_IQ2_BN = 29, // except 1d tensors - GGML_FTYPE_MOSTLY_IQ4_K = 30, // except 1d tensors - GGML_FTYPE_MOSTLY_IQ2_K = 31, // except 1d tensors + GGML_FTYPE_MOSTLY_IQ2_K = 30, // except 1d tensors + GGML_FTYPE_MOSTLY_IQ4_K = 31, // except 1d tensors + GGML_FTYPE_MOSTLY_IQ5_K = 32, // except 1d tensors }; // available tensor operations: |