Skip to content

Commit 2fc480d

Browse files
committed
examples/qwen3: match config with optimum flow
Sanity checked both flows on 1.0-rc3 and S24
1 parent 6f7d15e commit 2fc480d

File tree

1 file changed

+6
-1
lines changed

1 file changed

+6
-1
lines changed

examples/models/qwen3/config/qwen3_xnnpack_q8da4w.yaml

Lines changed: 6 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -8,8 +8,13 @@ model:
88

99
quantization:
1010
qmode: 8da4w
11+
embedding_quantize: 8,0
12+
13+
export:
14+
max_seq_length: 2048
15+
max_context_length: 2048
1116

1217
backend:
1318
xnnpack:
1419
enabled: True
15-
extended_ops: True
20+
extended_ops: True

0 commit comments

Comments
 (0)