Bingyi Kang
Bingyi Kang
Hey @RaphaelMarinier, same issue observed, any update on the solutions? Hi @Antymon, have you solved this problem?
> > Hi @Antymon, have you solved this problem? > > Nope. Have you tried other envs?
> We tried scaling up our custom model on seed_rl codebase, but it seems the TPU integration is incomplete for tf2.3. > I could be wrong but there are a...
Sorry, there is a typo in the instruction. I actually use python 2.7 instead of python 3.6. Change the python version, this problem shall be fixed.
Hi, actually, the shape for prediction is (B, N, 13, 13, A, (5+1)). In RegionLoss the number of classes is actually 1 instead of N, as shown in the config...
Are you using the cfg files in the instruction?
Our code is implemented with PyTorch 0.3.1 and Python 2.7. You may want to try this version.
Please check your PyTorchb (0.3.1) and Python (2.7) versions.
Actually, the data split for coco is also released in folder "data". Just change the dataset config and number of classes, everything should be good.