Training time

[shllgtca]

Hi!

Congratulations on your paper! I'm working on reproducing your results to propose a continuation of your work. The paper mentions that you used 8 NVIDIA T4 GPUs, but could you let me know how long you ran the experiments for? I only have access to 2 or possibly 4 GPUs, and I'd like to understand if reproducing the results is feasible and how much time it might take.

[shllgtca]

Additionally, I'm trying to run the experiments with the readme commands, and it seems to need the datasets shard files for instance:

train_shards: "/home/ubuntu/movi_e/train/shard-{000000..000679}.tar"
train_size: 9749
val_shards: "/home/ubuntu/movi_e/val/shard-{000000..000017}.tar"
val_size: 250
test_shards: "/home/ubuntu/movi_e/val/shard-{000000..000017}.tar"
test_size: 250

[FBehrad]

Additionally, I'm trying to run the experiments with the readme commands, and it seems to need the datasets shard files for instance:

train_shards: "/home/ubuntu/movi_e/train/shard-{000000..000679}.tar" train_size: 9749 val_shards: "/home/ubuntu/movi_e/val/shard-{000000..000017}.tar" val_size: 250 test_shards: "/home/ubuntu/movi_e/val/shard-{000000..000017}.tar" test_size: 250

Hi,
Could you figure it out? I have the same problem but for COCO :)