Here we provide the baseline model (MTR) and our models' checkpoint.

We provide three types of models based on how much training data they had used:

5%

MTR's performance (epoch_28):

       Waymo          mAP       minADE       minFDE     MissRate            
      VEHICLE      0.3288,      0.9097,      1.8636,      0.2156,            
   PEDESTRIAN      0.3192,      0.4202,      0.8957,      0.1134,            
      CYCLIST      0.2250,      0.9298,      1.9409,      0.2736,            
          Avg      0.2910,      0.7532,      1.5668,      0.2008,     

LLM-Augmented-MTR's performance (epoch_28):

       Waymo          mAP       minADE       minFDE     MissRate            
      VEHICLE      0.3367,      0.9216,      1.8960,      0.2236,            
   PEDESTRIAN      0.3613,      0.4295,      0.9059,      0.1103,            
      CYCLIST      0.2135,      0.9166,      1.9246,      0.2693,            
          Avg      0.3038,      0.7559,      1.5755,      0.2011,     

20%

MTR's performance (epoch_29):

       Waymo          mAP       minADE       minFDE     MissRate            
      VEHICLE      0.3912,      0.8239,      1.6778,      0.1800,            
   PEDESTRIAN      0.3608,      0.3829,      0.8091,      0.0935,            
      CYCLIST      0.2975,      0.8020,      1.6448,      0.2230,            
          Avg      0.3499,      0.6696,      1.3772,      0.1655,   

LLM-Augmented-MTR's performance (epoch_29):

       Waymo          mAP       minADE       minFDE     MissRate            
      VEHICLE      0.4028,      0.8090,      1.5943,      0.1711,            
   PEDESTRIAN      0.3621,      0.3880,      0.8107,      0.0958,            
      CYCLIST      0.2930,      0.8415,      1.7036,      0.2430,            
          Avg      0.3527,      0.6795,      1.3695,      0.1700,   

100%

MTR's performance (epoch_30):

       Waymo          mAP       minADE       minFDE     MissRate            
      VEHICLE      0.4464,      0.7545,      1.5161,      0.1523,            
   PEDESTRIAN      0.4149,      0.3456,      0.7252,      0.0750,            
      CYCLIST      0.3933,      0.6849,      1.3869,      0.1795,            
          Avg      0.4182,      0.5950,      1.2094,      0.1356,    

LLM-Augmented-MTR's performance (epoch_26):

       Waymo          mAP       minADE       minFDE     MissRate            
      VEHICLE      0.4578,      0.7570,      1.5308,      0.1523,            
   PEDESTRIAN      0.4794,      0.3535,      0.7376,      0.0765,            
      CYCLIST      0.3434,      0.7062,      1.4260,      0.1827,            
          Avg      0.4269,      0.6056,      1.2315,      0.1371,  
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support