nwpie/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2026-02-05 13:53:23 +02:00
Code Issues Packages Projects Releases Wiki Activity

mtmd: add min/max pixels gguf metadata (#19273)

Browse Source
This commit is contained in:
Xuan-Son Nguyen
2026-02-02 20:59:06 +01:00
committed by GitHub
parent 9f682fb640
commit 07a7412a3b
3 changed files with 10 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

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

@@ -284,6 +284,8 @@ class Keys:
class ClipVision:
PROJECTOR_TYPE = "clip.vision.projector_type" # for mixed modality models
IMAGE_SIZE = "clip.vision.image_size"
IMAGE_MIN_PIXELS = "clip.vision.image_min_pixels"
IMAGE_MAX_PIXELS = "clip.vision.image_max_pixels"
PREPROC_IMAGE_SIZE = "clip.vision.preproc_image_size"
PATCH_SIZE = "clip.vision.patch_size"
EMBEDDING_LENGTH = "clip.vision.embedding_length"

6
gguf-py/gguf/gguf_writer.py
View File

@@ -1113,6 +1113,12 @@ class GGUFWriter:
def add_vision_image_size(self, value: int) -> None:
self.add_uint32(Keys.ClipVision.IMAGE_SIZE, value)
def add_vision_max_pixels(self, value: int) -> None:
self.add_uint32(Keys.ClipVision.IMAGE_MAX_PIXELS, value)
def add_vision_min_pixels(self, value: int) -> None:
self.add_uint32(Keys.ClipVision.IMAGE_MIN_PIXELS, value)
def add_vision_preproc_image_size(self, value: int) -> None:
self.add_uint32(Keys.ClipVision.PREPROC_IMAGE_SIZE, value)

2
tools/mtmd/clip-impl.h
View File

@@ -36,6 +36,8 @@
// vision-specific
#define KEY_VISION_PROJ_TYPE "clip.vision.projector_type" // for models with mixed modalities
#define KEY_IMAGE_SIZE "clip.vision.image_size"
#define KEY_IMAGE_MIN_PIXELS "clip.vision.image_min_pixels"
#define KEY_IMAGE_MAX_PIXELS "clip.vision.image_max_pixels"
#define KEY_PREPROC_IMAGE_SIZE "clip.vision.preproc_image_size"
#define KEY_PATCH_SIZE "clip.vision.patch_size"
#define KEY_IMAGE_MEAN "clip.vision.image_mean"