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

Stage 1 Training on UCF-Crime #35

Open
borijang opened this issue Aug 12, 2022 · 6 comments
Open

Stage 1 Training on UCF-Crime #35

borijang opened this issue Aug 12, 2022 · 6 comments

Comments

@borijang
Copy link

Hi, thanks for the amazing work. I would like to train MIST on UCF-Crime from scratch. As I see, the training code for stage 1 is for SHT. What needs to be modified in order for it to work on UCF?

@TianmingXie
Copy link

I have successfully debug the stage1training. But during the stage2training, I was told lack of the logger path. Can you told me how can I produce the log file?

@seetlin
Copy link

seetlin commented Oct 28, 2022

I have successfully debug the stage1training. But during the stage2training, I was told lack of the logger path. Can you told me how can I produce the log file?

I don't know if it's a success,stage1 training accuracy rate is only 88 and not 90. Using the file generated by this first step to train the second stage, the cashback is a value from beginning to end.

@albertxwz
Copy link

I have successfully debug the stage1training. But during the stage2training, I was told lack of the logger path. Can you told me how can I produce the log file?

I don't know if it's a success,stage1 training accuracy rate is only 88 and not 90. Using the file generated by this first step to train the second stage, the cashback is a value from beginning to end.

In the paper, it seems that auc of stage 1 for SHT_I3D is 89.15%. However, I cannot reproduce it using the set of parameters in the source code and I just got about 87%.

@F3NGYUX1N
Copy link

In the paper, it seems that auc of stage 1 for SHT_I3D is 89.15%. However, I cannot reproduce it using the set of parameters in the source code and I just got about 87%.

Hello, the stage1 part of SHT_I3D has been trained, but the code lacks MIL.ph and cannot save the false label. May I ask how you conducted the stage2 part? Looking forward to your reply.

@albertxwz
Copy link

In the paper, it seems that auc of stage 1 for SHT_I3D is 89.15%. However, I cannot reproduce it using the set of parameters in the source code and I just got about 87%.

Hello, the stage1 part of SHT_I3D has been trained, but the code lacks MIL.ph and cannot save the false label. May I ask how you conducted the stage2 part? Looking forward to your reply.

Thank you for your reply. I conducted your code based on Training_Guidelines.md.

@zzyshowers
Copy link

请问大家都怎么调试阶段一的训练,我只出来set seed0?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

6 participants