vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge branch 'master' into compilade/lazy-convert-hf

Browse source
This commit is contained in:
Francis Couture-Harpin 2024-05-08 10:56:03 -04:00
commit bffdaf4010
43 changed files with 1646 additions and 179 deletions
Download patch file Download diff file Expand all files Collapse all files

2
gguf-py/gguf/constants.py
View file

@ -817,6 +817,7 @@ class GGMLQuantizationType(IntEnum):
I64 = 27
F64 = 28
IQ1_M = 29
BF16 = 30
class GGUFEndian(IntEnum):
@ -888,6 +889,7 @@ GGML_QUANT_SIZES: dict[GGMLQuantizationType, tuple[int, int]] = {
GGMLQuantizationType.I64: (1, 8),
GGMLQuantizationType.F64: (1, 8),
GGMLQuantizationType.IQ1_M: (256, QK_K // 8 + QK_K // 16 + QK_K // 32),
GGMLQuantizationType.BF16: (1, 2),
}