Modify exllama weight
This commit is contained in:
parent
8af4a7a0b0
commit
a9838bba2f
|
@ -199,7 +199,7 @@ def get_linear(weight, bias, quantize):
|
|||
)
|
||||
elif quantize == "gptq":
|
||||
try:
|
||||
qweight, qzeros, scales, g_idx, bits, groupsize = weight
|
||||
qweight, qzeros, scales, g_idx, bits, groupsize, use_exllama = weight
|
||||
except Exception:
|
||||
raise NotImplementedError(
|
||||
f"The passed weight is not `gptq` compatible, loader needs to be updated."
|
||||
|
|
Loading…
Reference in New Issue