vrr/kvcache-ai-ktransformers
2
0
Fork 0
mirror of https://github.com/kvcache-ai/ktransformers.git synced 2025-09-09 13:55:27 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

support deepseekv3; runable but have precition problem

Browse source
This commit is contained in:
Azure 2025-01-31 08:27:24 +00:00
parent de7e892f72
commit 476b1d8dc6
13 changed files with 2178 additions and 24 deletions
Download patch file Download diff file Expand all files Collapse all files

18
ktransformers/operators/linear.py
View file

@ -54,15 +54,15 @@ class KLinearBase(ABC):
self.has_bias = False
self.dtype = torch.get_default_dtype()
if orig_module is not None:
self.in_features = orig_module.in_features
self.out_features = orig_module.out_features
else:
shape = self.gguf_loader.tensor_info[key + ".weight"]["shape"]
if len(shape) == 1:
print("Warning: orig_module is not set, but has in_features or out_features equals to 1, can't get in_features and out_features from GGUF")
self.in_features = self.gguf_loader.tensor_info[key + ".weight"]["shape"][0]
self.out_features = self.gguf_loader.tensor_info[key + ".weight"]["shape"][1]
# if orig_module is not None:
# self.in_features = orig_module.in_features
# self.out_features = orig_module.out_features
# else:
shape = self.gguf_loader.tensor_info[key + ".weight"]["shape"]
if len(shape) == 1:
print("Warning: orig_module is not set, but has in_features or out_features equals to 1, can't get in_features and out_features from GGUF")
self.in_features = self.gguf_loader.tensor_info[key + ".weight"]["shape"][0]
self.out_features = self.gguf_loader.tensor_info[key + ".weight"]["shape"][1]
@abstractmethod
def forward(self, x: torch.Tensor) -> torch.Tensor: