Yanwei

Hi, I use Batch 8 for KITTI (without using road map) and do not perform experiments on Waymo dataset. And I do not test other batch size for experiments.

Sorry for the late reply. I didn't find the response updated here. I checked the code and did not find the "road map" in the config. Can you point out...

Hi, this is only used to put the sampled GT instance into the current scenes, like that in https://github.com/dvlab-research/VFF/blob/54ff6ef470d0bfbe0dda0c2598ba31e12dfcf7a8/pcdet/datasets/augmentor/database_sampler.py#L205

Hi, maybe you can refer to this [repo](https://github.com/dvlab-research/UVTR). I implement the cross-modality DB sampler for nuScenes dataset.

Hi, I use the V100 with 32G memory for training.

Hi, it seems some variables are close to inf in the assignment. Could you provide the specific config you are using? Maybe you should lower the learning rate.

Hi, may I ask which model you are using? It seems the 2B and 7B models may not have great instruction-following ability. I try the same prompt but got the...

Hi, this could be a bug and should be fixed. Please try it again. I try the demo with input "Hello", it replied like this: ``` Hello! How can I...

Hi, the word in Appendix A means that we do not perform an extra PaddleOCR detector for evaluation. For the TextVQA, we keep the OCR Token with that in LLaVA....

Hi @AllentDan , thanks for your great work! It's really cool. We are happy to see it is supported by more repos.