ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update llama.py

Browse source
This commit is contained in:
Daniel Han 2025-07-22 03:34:51 -07:00
commit 35dada2a99
1 changed files with 3 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

9
unsloth/models/llama.py
View file

@ -86,16 +86,13 @@ from triton import __version__ as triton_version
HAS_XFORMERS = xformers is not None
BlockDiagonalCausalMask = xformers.attn_bias.BlockDiagonalCausalMask if HAS_XFORMERS else None
def clean_gpu_cache():
if DEVICE_TYPE == "xpu":
torch.xpu.empty_cache()
else:
torch.cuda.empty_cache()
if DEVICE_TYPE == "xpu":
clean_gpu_cache = torch.xpu.empty_cache
get_current_device = torch.xpu.current_device
else:
clean_gpu_cache = torch.cuda.empty_cache
get_current_device = torch.cuda.current_device
pass
def original_apply_qkv(self, X):
Q = self.q_proj(X)