Syencil/mobile-yolov5-pruning-distillation

Issues

• AttributeError: 'Model' object has no attribute 'names'

  #82 opened 2 years ago by YongpeiZhu
  0

• 'python3 train.py --type vocl' meet the error

  #81 opened 2 years ago by ResidualNS
  0

• 蒸馏代码问题

  #80 opened 2 years ago by ftl123456789
  0

• 蒸馏模块无法跑起来

  #79 opened 2 years ago by nayenizhenmei
  2

• 关于train后识别不到任何结果

  #78 opened 2 years ago by yakexi-cmd
  0

• 蒸馏

  #43 opened 4 years ago by yyyooohao
  4

• model文件夹中yaml文件

  #77 opened 2 years ago by zy597337447
  0

• 求助帖

  #73 opened 3 years ago by DoAlone
  3

• feature蒸馏的一些疑惑

  #75 opened 2 years ago by xxxwuwq
  0

• no fuse

  #74 opened 2 years ago by gy-zkj
  0

• mAP值一直不变，从一开始就是接近0

  #59 opened 3 years ago by wxz1996
  4

• 旋转框蒸馏

  #72 opened 3 years ago by xing-bing
  6

• 蒸馏后Student Model的准确率下降

  #71 opened 3 years ago by WenzhiShao
  1

• 使用python3 train.py --type mvocs训练出来的pt模型是20.7M，这正常吗？我看baseline的参数量是3.60 Params(M)，这是为什么呢

  #51 opened 4 years ago by tommy3266
  2

• 关于mobile-yolov5s的mAP

  #58 opened 3 years ago by EEEricLiu
  2

• 'Got {}, but numpy array, torch tensor, or caffe2 blob name are expected.'.format(type(x))) NotImplementedError: Got <class 'NoneType'>, but numpy array, torch tensor, or caffe2 blob name are expected.

  #42 opened 4 years ago by ahhhxxx
  2

• 大佬好，ignore_idx 是不是写错了

  #68 opened 3 years ago by Veal98
  1

• 请问如何配置apex进行混合多精度加速训练

  #70 opened 3 years ago by mengxiangqiqin
  0

• 作者你好，请问剪枝过程是参考的哪篇论文

  #69 opened 3 years ago by leiqing110
  0

• before_pruning.jpg和after_pruning.jpg

  #67 opened 3 years ago by yuzhixiaobo
  0

• 关于TA策略与d_online

  #66 opened 3 years ago by StrongerzZ
  1

• 大佬好，注意到 yolo head 和原版的有所不同，想问下具体做了哪些改变

  #65 opened 3 years ago by Veal98
  4

• AttributeError: Can't get attribute 'ModelPruned' on <module 'models.yolo' from 'models\\yolo.py'>

  #64 opened 3 years ago by hzc5095
  0

• 模型剪枝之后，参数减少很多，但是模型推理速度并没有明显区别

  #63 opened 3 years ago by Ylifei
  0

• 稀疏训练时遇到一些问题

  #62 opened 3 years ago by yezechen
  1

• 大佬好，求教小白如何顺利运行，感觉缺失了一些数据文件

  #61 opened 3 years ago by Veal98
  1

• pytorch版本

  #60 opened 3 years ago by Xingxu1996
  0

• Prune的ignore_idx指的是哪几层

  #32 opened 4 years ago by JasperKirk
  2

• 权重问题

  #57 opened 3 years ago by 1012207826
  0

• 常见的一些问题汇总

  #38 opened 4 years ago by Syencil
  3

• mobilenet的预训练权重

  #56 opened 3 years ago by jzh-Joshua
  1

• How to fix this problem 'Detect'object has no attribute 'm'

  #55 opened 3 years ago by Niki173
  0

• Mobilev2-Yolov5 的throughput

  #54 opened 4 years ago by EEEricLiu
  1

• 请问bn参数分布图如何绘制

  #53 opened 4 years ago by qq302639209
  0

• 权重文件不能下载

  #52 opened 4 years ago by MoXQian
  3

• 、

  #50 opened 4 years ago by chengzi999
  0

• 1

  #49 opened 4 years ago by zack2020-star
  0

• 请问稀疏参数的选取有什么原则吗？

  #48 opened 4 years ago by 15050188022
  0

• 可以私聊点问题吗

  #47 opened 4 years ago by neasfss
  1

• 这个可以扩展至5x吧？

  #46 opened 4 years ago by Bobby-youngking
  0

• 关于teacher model预训练权重

  #45 opened 4 years ago by Sharpiless
  1

• 遇到了点问题

  #44 opened 4 years ago by Lihong1997
  0

• RuntimeError: shape '[1, 3, 85, 80, 64]' is invalid for input of size 655360

  #41 opened 4 years ago by ahhhxxx
  3

• 关于finetune

  #40 opened 4 years ago by 15050188022
  2

• 策略

  #39 opened 4 years ago by yyyooohao
  0

• 蒸馏时Precision下降太多的问题

  #34 opened 4 years ago by CangHaiQingYue
  2

• 剪枝之后的模型也可以通过onnx转换和通过ncnn部署吗？

  #35 opened 4 years ago by zineos
  1

• 请问一下，使用预训练的yolov5s在coco上做基础训练，验证集的map非常接近于0，问题大概出在什么地方？

  #36 opened 4 years ago by August0424
  2

• 关于剪枝蒸馏的策略

  #37 opened 4 years ago by 15050188022
  2

• L1 稀疏化训练细节

  #33 opened 4 years ago by xuanyuyt
  0