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

Update utils.py

Browse source
This commit is contained in:
Daniel Han 2025-03-02 23:41:35 -08:00
commit 4b2b345cdd
1 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

3
unsloth/kernels/utils.py
View file

@ -451,7 +451,7 @@ pass
def matmul_lora(X, W, W_quant, A, B, s, out = None):
dtype = X.dtype
W = fast_dequantize(W.t(), W_quant, use_global_buffer = True)
W = fast_dequantize(W.t(), W_quant, use_global_buffer = False)
if X.dim() == 3:
batch, seq_len, d = X.shape
@ -461,6 +461,7 @@ def matmul_lora(X, W, W_quant, A, B, s, out = None):
reshape = False
pass
print(X.device, W.device, torch.cuda.current_device())
out = torch_matmul(X, W, out = out)
if W_quant is not None: del W