GroundingDINO
GroundingDINO copied to clipboard
why clamp attn_weights with min-max
- https://github.com/IDEA-Research/GroundingDINO/blob/main/groundingdino/models/GroundingDINO/fuse_modules.py#L184
- Is it possible that this is the reason for the accuracy drop of TensorRT-FP16