MCZhi / GameFormer

[ICCV 2023 Oral] Game-theoretic modeling and learning of Transformer-based interactive prediction and planning
https://mczhi.github.io/GameFormer/
240 stars 32 forks source link

nuplan training #12

Open 942411526 opened 3 weeks ago

942411526 commented 3 weeks ago

Dear authors, I hope this email finds you well. I am reaching out to seek some clarification regarding the training process described in your paper

I am currently attempting to reproduce your results using the NuPlan dataset, which consists of three parts: Train, Val, and Test. I would like to confirm whether the training data used in your research included both the Train and Val sets, or if only the Train set was utilized for model training.

I initially trained my model using only the Train set, but the results show a noticeable discrepancy from those reported in your paper. I would greatly appreciate your guidance on whether including the Val set as part of the training data is recommended, or if there are other aspects of the training process that I should consider.