DQS

Oh! Because Argoverse dataset's API must run on Ubuntu/MacOS and I coded on Windows, I used a very silly method to tackle the loading of dataset. I pre-processed the dataset...

Argoverse's test datasets and train/val datasets have different format, predicted agent in test datasets only have trajectory in first 2s, and others have first 5s, so we should divide the...

Generally, the input of the neural network is formed with batch, which means the network accepts several sets of data (maybe larger than 1) as the input, so the input...

Take easy to descript your question here, if necessary, you can contact me by email: [email protected]

Sorry, my model has some bugs, I will fix after some days, then I will give you the performance, thank you!

I found that this code maybe work? But I don't know how to use my `translation_fn` ```python from collections import namedtuple import os, time import jax import copy from jax...

Hi @fehiepsi, sorry for the ambiguity. This code is working but it uses the simple MH algorithm, I want to use some amazing algorithms in numpyro such as NUTS, HMCECS,...

Hi, I modified my code according to yours, but I didn't find the bug. My output of min_dep and max_dep are 0.117861, 0.789501 respectively, but your dataset shows 1.00787 and...