CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-20 12:07:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

optimize deepstack structure

Browse Source
This commit is contained in:
JJJYmmm
2025-10-28 21:58:46 +08:00
parent 0443a098f3
commit 3271877207
5 changed files with 73 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

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

@@ -278,7 +278,7 @@ class Keys:
USE_GELU = "clip.use_gelu"
USE_SILU = "clip.use_silu"
N_WA_PATTERN = "clip.vision.n_wa_pattern" # used by qwen2.5vl
DEEPSTACK_LAYERS = "clip.vision.deepstack_layers"
IS_DEEPSTACK_LAYERS = "clip.vision.is_deepstack_layers"
class Attention:
HEAD_COUNT = "clip.vision.attention.head_count"

4
gguf-py/gguf/gguf_writer.py
View File

@@ -1074,8 +1074,8 @@ class GGUFWriter:
def add_vision_n_wa_pattern(self, value: int) -> None:
self.add_uint32(Keys.ClipVision.N_WA_PATTERN, value)
def add_vision_deepstack_layers(self, layers: Sequence[int]) -> None:
self.add_array(Keys.ClipVision.DEEPSTACK_LAYERS, layers)
def add_vision_is_deepstack_layers(self, layers: Sequence[bool]) -> None:
self.add_array(Keys.ClipVision.IS_DEEPSTACK_LAYERS, layers)
# audio models