imatrix : use GGUF to store imatrix data

This commit is contained in:
Francis Couture-Harpin 2024-09-06 17:17:25 -04:00
parent 347247a24e
commit 3de9300c37
4 changed files with 352 additions and 149 deletions

View file

@ -167,6 +167,12 @@ class Keys:
TYPE = "adapter.type"
LORA_ALPHA = "adapter.lora.alpha"
class IMatrix:
CHUNK_COUNT = "imatrix.chunk_count"
CHUNK_SIZE = "imatrix.chunk_size"
DATASET = "imatrix.dataset"
#
# recommended mapping of model tensor names for storage in gguf
#
@ -175,6 +181,7 @@ class Keys:
class GGUFType:
MODEL = "model"
ADAPTER = "adapter"
IMATRIX = "imatrix"
class MODEL_ARCH(IntEnum):