固件类型：自行编译的固件
SDK包名称：rv1126_rv1109_linux_release_20220224_v2.2.5a
SDK修改内容：rv1126_rv1109_linux_release_20220224_v2.2.5a
Log日志：log.txt
本帖最后由 晴朗多肉 于 2023-7-6 10:57 编辑
pytorch框架，基于yolov7的模型。在pc端可正常推理，在模拟器转换为rknn模型的过程中，提示
I squeeze_to_4d_transpose: remove node = , add node = [‘2569_Transpose_squeeze0’, ‘2569_Transpose_squeeze1’]
I squeeze_to_4d_transpose: remove node = , add node = [‘2544_Transpose_squeeze0’, ‘2544_Transpose_squeeze1’]
I squeeze_to_4d_transpose: remove node = , add node = [‘2519_Transpose_squeeze0’, ‘2519_Transpose_squeeze1’]
E build: Catch exception when building RKNN model!
E build: Traceback (most recent call last):
E build: File “rknn/api/rknn_base.py”, line 1546, in rknn.api.rknn_base.RKNNBase.build
E build: File “rknn/api/graph_optimizer.py”, line 1276, in rknn.api.graph_optimizer.GraphOptimizer.fuse_ops
E build: File “rknn/api/fuse_rules.py”, line 5449, in rknn.api.fuse_rules._p_tiling_conv_channelin
E build: File “rknn/api/fuse_rules.py”, line 5387, in rknn.api.fuse_rules._judge_filter_limit
E build: ZeroDivisionError: division by zero
Build model failed!
其他模型均可正常转换，猜测可能是有些op不支持，请教如何进一步分析？
tchip_askquestions
log.txt (12.2 KB)