From bfec2e814bb4d940a6d832db65bbb951f47aa0eb Mon Sep 17 00:00:00 2001
From: Sylwester Fraczek <sylwester.fraczek@intel.com>
Date: Tue, 30 Nov 2021 14:49:14 +0100
Subject: [PATCH] add matmul_v2_transpose_reshape_fuse_pass to
quant2_int8_mkldnn_pass.py (#37619)
---
.../fluid/contrib/slim/quantization/quant2_int8_mkldnn_pass.py | 1 +
1 file changed, 1 insertion(+)
diff --git a/python/paddle/fluid/contrib/slim/quantization/quant2_int8_mkldnn_pass.py b/python/paddle/fluid/contrib/slim/quantization/quant2_int8_mkldnn_pass.py
index 7930923668c7d..bc97e5cf6c958 100644
--- a/python/paddle/fluid/contrib/slim/quantization/quant2_int8_mkldnn_pass.py
+++ b/python/paddle/fluid/contrib/slim/quantization/quant2_int8_mkldnn_pass.py
@@ -419,6 +419,7 @@ def _optimize_fp32_graph(self, graph):
if self._is_fc_quantized(graph):
graph = self._apply_pass(graph, 'fc_mkldnn_pass')
graph = self._apply_pass(graph, 'matmul_transpose_reshape_fuse_pass')
+ graph = self._apply_pass(graph, 'matmul_v2_transpose_reshape_fuse_pass')
# the following pass should be the last one since it will work on all fused ops.
graph = self._apply_pass(graph, 'runtime_context_cache_pass')
return graph