NonsansWD
NonsansWD
I used the flexible training script as it was suggested within the readme file and now i have this logdir with npz files and stuff. Is there a way i...
@Alescontrela Initializer expected to generate shape (4, 86) but got shape (16, 86) instead for parameter "d_0" in "/model/BroadcastPositionBiases_0". (https://flax.readthedocs.io/en/latest/api_reference/flax.errors.html#flax.errors.ScopeParamShapeError) this happened when running videogpt training in transformer.py at line...
 As you can see in the image I provided, the losses are really sky rocketing to up to 10^7 so I wanted to ask if this is fine or...
@Alescontrela You have provided the videogpt_reward_model.py in the notebooks folder which shows the evaluation of the reward model. I would like to do my own kind of evaluation based on...