Fixed critic pointer output device inconsistency

Legacy organized (forward and local_forward)

Local forward vectorized (the legacy remains as og-s)

context vector mask fixed; working on padding masks over samples in a…

Added network architecture

README.md organized

env settings and hyperparam settings

data of new model

averages

supports nan

README.md updated

supports state-action history and dynamic seed display

env linted

Data of the new model collected

New model trained backed up

train code for a new model; data-collect/view updated accordingly

directory name updated

Checkpoint supported (date automation not supported yet)

Data uploaded, collect/view updated

typos

model backed up

experiments updated

rm.md updated

bug fix: mask reverted; self-loop kept

critical bug fixed: att_score dtype and context_token value

Avoids all-zero errors from env_check in RLlib

Env with params updated

Add clarity in the RLlib level model (model_rllib.py)

model updated