Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

0.4.0 的超參數搜尋和Reopt訓練 #925

Open
4 tasks done
DrChungAlbert opened this issue Oct 11, 2023 · 0 comments
Open
4 tasks done

0.4.0 的超參數搜尋和Reopt訓練 #925

DrChungAlbert opened this issue Oct 11, 2023 · 0 comments
Assignees
Labels
question Further information is requested

Comments

@DrChungAlbert
Copy link

DrChungAlbert commented Oct 11, 2023

Before Asking

  • I have read the README carefully. 我已经仔细阅读了README上的操作指引。

  • I want to train my custom dataset, and I have read the tutorials for training your custom data carefully and organize my dataset correctly; (FYI: We recommand you to apply the config files of xx_finetune.py.) 我想训练自定义数据集,我已经仔细阅读了训练自定义数据的教程,以及按照正确的目录结构存放数据集。(FYI: 我们推荐使用xx_finetune.py等配置文件训练自定义数据集。)

  • I have pulled the latest code of main branch to run again and the problem still existed. 我已经拉取了主分支上最新的代码,重新运行之后,问题仍不能解决。

Search before asking

  • I have searched the YOLOv6 issues and found no similar questions.

Question

請問
我用0.4.0 版本為了進行量化感知訓練輸出浮點數模型(input-size 為640x384),以便在NPU平台利用該平台tools把浮點數模型轉換為int8模型,進行NPU推論時能夠維持浮點數模型的精度 (因為該NPU不適用TRT模型)
在這前提之下,我進行以下的超參數搜尋和ReOpt訓練:

超參數搜尋
1.Training of V6-RepOpt Step1
因為[pretrained scale] 為舊版無法使用
python tools/train.py --batch 32 --conf configs/repopt/yolov6n_hs.py --data data/coco.yaml --device 0
此處 yolov6n_hs.py 需要設定下面兩個參數嗎?
use_dfl=True,
reg_max=16

ReOpt訓練
2. Training of V6-RepOpt Step2
python tools/train.py --batch 32 --conf configs/repopt/yolov6n_opt.py --data data/coco.yaml --device 0
此處 yolov6n_opt.py 如何設定下面兩個參數?
pretrained=None, (這裡是否能提供0.4.0版本的 yolov6n_reopt.pt? 目前只有看到yolov6n_v2_reopt.pt,舊版本無法使用)
scales=' ../weight/v6n_v4_scale_640x384_last.pt', (Step1 訓練出來的)

Evaluation
python tools/eval.py --data data/coco.yaml --batch 32 --weights yolov6n_opt.pt ( Step2訓練出來的) --task val
是否有0.4.0版本 能正確計算的?

以上問題 還請大佬解答,如有錯誤之處,煩請告知 謝謝

Additional

另外請問是否量化感知訓練是否一定需要Reopt這流程產生的作為teacher_model才能進行?
如果我要訓練自定義Custom數據集的模型,是否能用COCO的Reopt模型來跑量化感知訓練QAT?
No response

@DrChungAlbert DrChungAlbert added the question Further information is requested label Oct 11, 2023
@DrChungAlbert DrChungAlbert changed the title 0.4.0 的超參數搜尋及量化感知訓練 0.4.0 的超參數搜尋訓練 Oct 11, 2023
@DrChungAlbert DrChungAlbert changed the title 0.4.0 的超參數搜尋訓練 0.4.0 的超參數搜尋和Reopt訓練 Oct 11, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
question Further information is requested
Projects
None yet
Development

No branches or pull requests

2 participants