Skip to content

Commit 7b47912

Browse files
authored
add config for infer (#2379)
1 parent ec1303c commit 7b47912

File tree

1 file changed

+6
-0
lines changed

1 file changed

+6
-0
lines changed

paddleformers/quantization/quantization_config.py

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -74,6 +74,9 @@ def __init__(
7474
actscale_moving_rate=0.01,
7575
fp8_format_type="hybrid",
7676
scale_epsilon=1e-8,
77+
dense_quant_type="",
78+
moe_quant_type="",
79+
quantization="",
7780
**kwargs,
7881
):
7982
if weight_quantize_algo is not None:
@@ -154,6 +157,9 @@ def __init__(
154157
self.actscale_moving_rate = actscale_moving_rate
155158
self.fp8_format_type = fp8_format_type
156159
self.scale_epsilon = scale_epsilon
160+
self.dense_quant_type = dense_quant_type
161+
self.moe_quant_type = moe_quant_type
162+
self.quantization = quantization
157163

158164
@property
159165
def fp8_format(self):

0 commit comments

Comments
 (0)