diff --git a/figures/model_faq48_fig1_0819.png b/figures/model_faq48_fig1_0819.png
new file mode 100644
index 0000000000000000000000000000000000000000..0afa6e00125ce9197a0c60c48978c7172a8bb654
Binary files /dev/null and b/figures/model_faq48_fig1_0819.png differ
diff --git a/figures/model_faq48_fig2_0819.png b/figures/model_faq48_fig2_0819.png
new file mode 100644
index 0000000000000000000000000000000000000000..17dd52ae7dc8ce26db4671365245da047307f88c
Binary files /dev/null and b/figures/model_faq48_fig2_0819.png differ
diff --git a/figures/model_faq48_fig3_0819.png b/figures/model_faq48_fig3_0819.png
new file mode 100644
index 0000000000000000000000000000000000000000..e86c242d7f99f22ceb65d7cd415899b31bde8582
Binary files /dev/null and b/figures/model_faq48_fig3_0819.png differ
diff --git "a/pytorch-train-guide/Pytorch\350\256\255\347\273\203-FAQ.md" "b/pytorch-train-guide/Pytorch\350\256\255\347\273\203-FAQ.md"
index 7c4ed6dafd2827f6d53df673a8b1a7e34ea49e0f..719b0058842ff4e631def440f090a7fcd5428870 100644
--- "a/pytorch-train-guide/Pytorch\350\256\255\347\273\203-FAQ.md"
+++ "b/pytorch-train-guide/Pytorch\350\256\255\347\273\203-FAQ.md"
@@ -51,6 +51,7 @@
     - [FAQ45、模型训练过程DCNv2算子+混合精度报错：excepted scalar type float but found half](#faq45模型训练过程dcnv2算子混合精度报错excepted-scalar-type-float-but-found-half)
     - [FAQ46、模型训练过程中算子PadV3D报错：constant_values value mismatches](#faq46模型训练过程中算子padv3d报错constant_values-value-mismatches)
     - [FAQ47、模型训练过程中python层报错：Expected isFloatingType(grad[i].scalar_type()) to be true, but got false](#faq47模型训练过程中报Expected-isFloatingType(grad[i].scalar_type())-to-be-true-but-got-false)
+    - [FAQ48、模型训练中aicpu算子报错](#faq48模型训练中aicpu算子报错)
   - [2.2 NPU模型分布式运行常见问题FAQ](#22-npu模型分布式运行常见问题faq)
     - [FAQ1、在模型分布式训练时，遇到报错 host not found.](#faq1在模型分布式训练时遇到报错-host-not-found)
     - [FAQ2、在模型运行时，遇到eval模式下loss值特别大，过万.](#faq2在模型运行时遇到eval模式下loss值特别大过万)
@@ -974,6 +975,20 @@ DCNv2安装版本以该链接为例：https://github.com/jinfagang/DCNv2_latest
 
 然后利用模型中的数据构造gather单算子，问题复现，结论是gather算子的输入输非法的，然后npu的gather正常计算，进而在反向报错
 
+### FAQ48、模型训练中aicpu算子报错：Aicpu kernel execute failed
+- 现象描述
+屏幕报错信息如下所示：
+![](https://gitee.com/xiaxia3/ascend-pytorch-crowdintelligence-doc/raw/master/figures/model_faq48_fig1_0819.png)
+从中能看到时aicpu算子报错，然后查看host侧日志，显示如下：
+![](https://gitee.com/xiaxia3/ascend-pytorch-crowdintelligence-doc/raw/master/figures/model_faq48_fig2_0819.png)
+
+- 原因分析
+根据host侧日志信息可以看到是aicpu算子index_put报错，猜测是索引值数据类型不对，aicpu算子实现同CPU算子，索引值类型是int64
+- 处理方法
+打点找到具体报错的函数，将此算子to cpu计算，结果也是报错，打印输入索引类型是int32,转换为int64后，不在报错
+
+  ![](https://gitee.com/xiaxia3/ascend-pytorch-crowdintelligence-doc/raw/master/figures/model_faq48_fig3_0819.png)
+
 ## [2.2 NPU模型分布式运行常见问题FAQ](#22-NPU模型分布式运行常见问题FAQ)
 ### FAQ1、在模型分布式训练时，遇到报错 host not found.