jxbbb / TOD3Cap

[ECCV 2024] TOD3Cap: Towards 3D Dense Captioning in Outdoor Scenes
99 stars 5 forks source link

`phase` in `self.llama_adapter` #3

Open cocoshe opened 3 months ago

cocoshe commented 3 months ago

Thanks for your great work! And I failed to run the distributed training script, at least because of this:

https://github.com/jxbbb/TOD3Cap/blob/b215eeb25b2aea6993069bcc504d3d939f380305/projects/mmdet3d_plugin/bevformer/detectors/bevformer.py#L75

It need to be fixed in a special way? I don't know the details about the design of mmdet3d, can you give the suggestion to run the code?

cocoshe commented 3 months ago

And, the file of llama_ckpt_dir, llama_tokenzier_path is provided?

jxbbb commented 3 months ago

Thanks for your interest in TOD3Cap. We will fix it within this week.

cocoshe commented 3 months ago

Thanks for your interest in TOD3Cap. We will fix it within this week.

Thanks for your reply, Hoping for it~