openppl-public ppq issues

openppl-public / ppq

PPL Quantization Tool (PPQ) is a powerful offline neural network quantization tool.

Apache License 2.0

1.44k stars 223 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Quantization for tflite

#567 liamsun2019 opened 2 weeks ago
0
RuntimeError: Op Execution Error: /Slice(Type: Slice, Num of Input: 5, Num of Output: 1)

#566 yehu3d opened 2 weeks ago
0
ProgramEntrance_2.py读取imagenet图片reshape报错

#565 huanFigure opened 3 weeks ago
0
求助为什么量化以后模型没有减小？

#564 stricklandye opened 1 month ago
0
AttributeError: 'NoneType' object has no attribute 'copy'

#563 undcloud opened 1 month ago
0
PPQ在画质类任务上(如低光降噪)量化效果咋样呢？

#562 666DZY666 opened 1 month ago
0
PPQ已经支持了不少后端推理框架(TensorRT,ncnn,mnn等），这些后端的量化规格细节都是怎么获取的呢？现在PPQ与后端量化推理可以完全对齐吗？

#561 666DZY666 opened 1 month ago
0
VIT类模型量化感知训练loss跑飞

#560 XingchenSong1116 opened 1 month ago
1
能否支持权重和激活位宽不同的量化？

#559 frankgt opened 1 month ago
0
这个工具链已经收敛了吗？我看已经比较久没有特性更新了，后续还有新特性计划吗？

#558 666DZY666 opened 1 month ago
1
ppq使用torch_mlir导出tosa格式的mlir

#557 zccyman opened 2 months ago
1
是否有计划支持高通QNN(比较新的版本)和海思芯片呢？

#556 666DZY666 opened 2 months ago
1
关于QAT中的BN？

#555 666DZY666 opened 2 months ago
1
Op Execution Error MMCVModulatedDeformConv2d

#554 JXCab closed 2 months ago
0
PPQint8CPU量化bug

#553 37ETs opened 2 months ago
1
[视频资料]使用profiler来分析网络性能，tensorboard无法解析日志文件

#552 JXCab opened 2 months ago
2
请问PPQ支持一维网络模型的量化吗

#551 JiawenDeng opened 2 months ago
1
yolov5量化，在ppq/samples/yolo中运行02_Quantization.py文件出现警告和报错，输出模型只有onnx没有engine文件

#550 zjhleaning opened 2 months ago
1
Op Execution Error: Einsum

#549 heqqian opened 2 months ago
1
请问如何使用ppq量化模型的部分模块，而不是整个模型呢？

#548 heqqian closed 2 months ago
0
Fix Resize and Gridsample ops

#547 liwei109 closed 3 months ago
0
请问目前支持pytorch模型直接量化（而不是转成onnx），并生成量化后pytorch模型吗

#546 JiliangNi opened 3 months ago
1
请问为什么在运行 ppq/ppq/samples/TensorRT/Benchmark_with_onnx.py 文件时会出现Segmentation Fault的可能原因是什么？

#545 wmWangmiao closed 3 months ago
1
How to get a Qoperator format ONNX model after quantization?

#544 JiliangNi opened 3 months ago
1
执行quantize_torch_model.py后，pytorch模型export出的onnx模型并不是QNN模型

#543 JiliangNi opened 3 months ago
9
BatchNormalization量化出错

#542 18liumin opened 3 months ago
2
请问是否可以通过量化后产生的onnx 推理结果可以直观看出量化的情况？

#541 fakerhbj opened 4 months ago
2
加减类算子类型匹配问题

#540 ycwnuoyan opened 4 months ago
3
关于导出平台的报错信息

#539 dpf0522 opened 5 months ago
2
修正了resize op与Prelu op执行时可能出现的问题

#538 ZhangZhiPku closed 4 months ago
0
修正了一个 onnx domain 的相关问题

#537 ZhangZhiPku closed 5 months ago
0
[fix] Remove restriction on calibration dataset length

#536 liwei109 closed 4 months ago
1
yolov5_sample报错

#535 uidq2956 opened 5 months ago
1
量化模型的时候出现TypeError

#534 Francis235 opened 5 months ago
1
Update README.md

#533 LJoson closed 5 months ago
0
关于resize op的一个bug

#532 Nusselder9 opened 5 months ago
1
Op Execution Error: PRelu_1

#531 qiankl opened 6 months ago
4
调度部分算子至FP32时，生成engine失败

#530 zhjw0927 opened 6 months ago
2
请问model zoo的模型在哪里？

#529 YilanWang opened 6 months ago
1
NotImplementedError: Graph op: /quantize/ArgMin(ArgMin) has no backend implementation

#528 YilanWang opened 6 months ago
2
量化时出现RandomNormalLike这个算子没有注册跟实现的问题，请问有没有简单的注册实现例子可以参考？

#527 Uwenisme opened 6 months ago
1
QLinearMatmul : weight zero point must be a scalar, 1D tensor of size 1, or last to second dimension is 1

#526 fhahaha opened 6 months ago
0
yolov5 模型量化后的精度损失

#525 zhjw0927 opened 6 months ago
2
内存泄露

#524 KompressorSC opened 7 months ago
3
能不能量化自己的模型啊？有教程吗

#523 couakua opened 7 months ago
1
YOLOv6量化后仅能预测一类

#522 zhujiajian98 opened 7 months ago
3
yolov8量化后无结果

#521 kjgfcdb opened 7 months ago
3
PPQ yolov5-7.0 ：return torch.cat(concat_view, axis=axis) RuntimeError: Sizes of tensors must match except in dimension 1. Expected size 28 but got size 27 for tensor number 1 in the list.

#520 YellenYeason opened 7 months ago
1
PPQ installtion error: `TypeError: Descriptors cannot be created directly. If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.`

#519 kyrie2to11 opened 7 months ago
5
LSQ算法微调后，输出的scale量化参数存在负值，请问如何修改源码，使其输出的scale为正值？

#518 fengyanWang opened 7 months ago
1