GoalFlow
GoalFlow copied to clipboard
YvanYin
Repo of "GoalFlow: Goal-Driven Flow Matching for Multimodal Trajectories Generation in End-to-End Autonomous Driving"
Hi, thank you for sharing GoalFlow! I'm training the perception model using `run_goalflow_training_perception.sh`, but my training loss keeps oscillating and doesn't converge even after 40 epochs. (Attached TensorBoard screenshot shows...
@YvanYin , When flow sampling, navi_dropout is used to get trajectory_nonavi and this woule be add with condition trajectory, my question is why **force_dropout** is not used to get uncondition...
Hi~ thanks for your work! I am trying to train your model. Following your readme, I first try to train the `perception` module. However, during training, I meet `nan` in...
Hi, Thanks for your excellent work. I tried to run the train script as follows with updated code: export OPENSCENE_DATA_ROOT=/data/navsim_workspace/dataset # 根据您的实际数据路径调整 export NAVSIM_EXP_ROOT=/data/navsim_workspace/exp # 根据您的实际实验路径调整 export NUPLAN_MAPS_ROOT=/data/navsim_workspace/dataset/maps export PYTHONPATH=/data/workspace/GoalFlow:/data/navsim_workspace/navsim:$PYTHONPATH...
Dear author, thank you for your work. I have a question about the method `normalize_xy_rotation`, why rotate `gt_trajs_` 10 times, for denoise training?
Hello, I don't quite understand the part about building the Goal Point Vocabulary while reading the paper. Could you explain it to me?
Thank you for your excellent work. I have a question about the dac_score_feature. In goalflow_model_navi.py, dac_score_featureis used with dac_scoresto calculate the loss, but I couldn't find how dac_score_featureis obtained. Could...
Could you provide an estimate for total training time on your 32 counts of 4090 GPU configuration? I am trying to estimate feasibility for training on two 48GB A6000 GPUs.
Hello, thank you for your excellent work. I'm really interested in these! I have a question: in your training code, has_navi=True. However, in the flow section, there is code like...
