rknn-toolkit2
rknn-toolkit2 copied to clipboard
onnx转rknn过程中,onnx1.4.0是支持gather的,但有报错 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,...
在最新的rknn-toolkit2-1.3.0中转换ONNX模型,运行到rknn.build出现段错误 转换过程的log  [lstm.log](https://github.com/rockchip-linux/rknn-toolkit2/files/9466218/lstm.log) 转模型脚本以及模型文件 [lstm.zip](https://github.com/rockchip-linux/rknn-toolkit2/files/9466228/lstm.zip)
问题描述: --------------------------- 单batch输入的yolov5s只能占用45%的NPU性能,试图用多batch推理提高整体速度。 24+batch推理时,npu性能出现10%-89%的波动,随后报错killed。  模型信息: --------------------------- 模型:YOLOV5S batch:28 推理模式:单核(NPU_0) 内存评估: 
[bug report] E load_onnx: Traceback (most recent call last): E load_onnx: File "rknn/api/rknn_base.py", line 1157, in rknn.api.rknn_base.RKNNBase.load_onnx E load_onnx: UnboundLocalError: local variable 'inp' referenced before assignment
你好, 有什么方法可以将 .rknn 二进制文件转换为 .txt 以查看它的结构。 我试图找到类似于 onnx-proto 的东西。 或者也许有一种方法可以解码 rknn 二进制文件? 如果有任何帮助,我将不胜感激。
你好, 我试图找到有关如何将 .onnx 模型转换为 .rknn 模型的描述。 在 lib 目录中没有找到关于这个问题的任何源代码。 有什么办法可以查看库的完整源代码吗?
Hello, Is there any way to convert .rknn binary file to .txt in order to see the structure of it. I'm trying to find something similar to onnx-proto. Or maybe...
Hello, I'm trying to find description of how .onnx models are converted to .rknn models. Didn't find any source codes on this issue in lib directory. Is there any way...
Apols guys as a dumb englishman quite impressed by the rk3588 npu and would like to bench some models. I have been running the basic resnet-18 model but found its...
Can you supply a wheel for ubuntu as you have rknn_toolkit_lite2-1.4.0-cp37-cp37m-linux_aarch64.whl rknn_toolkit_lite2-1.4.0-cp39-cp39-linux_aarch64.whl Which is ok for Debian but no whl for Ubuntu?