mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2025-04-16 03:26:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

convert : fix squeeze for ssm_conv tensors (#12573)

Browse Source 
* convert : fix squeeze for ssm_conv tensors

* convert : match ssm_conv tensors by type

---------

Co-authored-by: Francis Couture-Harpin <git@compilade.net>
This commit is contained in:
Georgi Gerganov 2025-03-26 14:21:05 +02:00 committed by GitHub
parent 5ed38b6852
commit df4d20cd53
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 4 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
convert_hf_to_gguf.py
View File

@ -3803,8 +3803,6 @@ class MambaModel(Model):
_tok_embd = None
def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
del bid # unused
output_name = self.format_tensor_name(gguf.MODEL_TENSOR.OUTPUT)
tok_embd_name = self.format_tensor_name(gguf.MODEL_TENSOR.TOKEN_EMBD)
@ -3814,6 +3812,10 @@ class MambaModel(Model):
logger.debug("A_log --> A ==> " + new_name)
data_torch = -torch.exp(data_torch)
# [4 1 8192 1] -> [4 8192 1 1]
if self.match_model_tensor_name(new_name, gguf.MODEL_TENSOR.SSM_CONV1D, bid):
data_torch = data_torch.squeeze()
# assuming token_embd.weight is seen before output.weight
if self._tok_embd is not None and new_name == output_name:
if torch.equal(self._tok_embd, data_torch):