jat-project/jat-dataset

---
annotations_creators:
- found
- machine-generated
license: apache-2.0
source_datasets:
- conceptual-captions
- ok-vqa
- oscar
task_categories:
- reinforcement-learning
- text-generation
- question-answering
pretty_name: JAT-dataset
configs:
- config_name: atari-alien
data_files:
- split: train
path: atari-alien/train-*
- split: test
path: atari-alien/test-*
- config_name: atari-amidar
data_files:
- split: train
path: atari-amidar/train-*
- split: test
path: atari-amidar/test-*
- config_name: atari-assault
data_files:
- split: train
path: atari-assault/train-*
- split: test
path: atari-assault/test-*
- config_name: atari-asterix
data_files:
- split: train
path: atari-asterix/train-*
- split: test
path: atari-asterix/test-*
- config_name: atari-asteroids
data_files:
- split: train
path: atari-asteroids/train-*
- split: test
path: atari-asteroids/test-*
- config_name: atari-atlantis
data_files:
- split: train
path: atari-atlantis/train-*
- split: test
path: atari-atlantis/test-*
- config_name: atari-bankheist
data_files:
- split: train
path: atari-bankheist/train-*
- split: test
path: atari-bankheist/test-*
- config_name: atari-battlezone
data_files:
- split: train
path: atari-battlezone/train-*
- split: test
path: atari-battlezone/test-*
- config_name: atari-beamrider
data_files:
- split: train
path: atari-beamrider/train-*
- split: test
path: atari-beamrider/test-*
- config_name: atari-berzerk
data_files:
- split: train
path: atari-berzerk/train-*
- split: test
path: atari-berzerk/test-*
- config_name: atari-bowling
data_files:
- split: train
path: atari-bowling/train-*
- split: test
path: atari-bowling/test-*
- config_name: atari-boxing
data_files:
- split: train
path: atari-boxing/train-*
- split: test
path: atari-boxing/test-*
- config_name: atari-breakout
data_files:
- split: train
path: atari-breakout/train-*
- split: test
path: atari-breakout/test-*
- config_name: atari-centipede
data_files:
- split: train
path: atari-centipede/train-*
- split: test
path: atari-centipede/test-*
- config_name: atari-choppercommand
data_files:
- split: train
path: atari-choppercommand/train-*
- split: test
path: atari-choppercommand/test-*
- config_name: atari-crazyclimber
data_files:
- split: train
path: atari-crazyclimber/train-*
- split: test
path: atari-crazyclimber/test-*
- config_name: atari-defender
data_files:
- split: train
path: atari-defender/train-*
- split: test
path: atari-defender/test-*
- config_name: atari-demonattack
data_files:
- split: train
path: atari-demonattack/train-*
- split: test
path: atari-demonattack/test-*
- config_name: atari-doubledunk
data_files:
- split: test
path: atari-doubledunk/test-*
- split: train
path: atari-doubledunk/train-*
- config_name: atari-enduro
data_files:
- split: train
path: atari-enduro/train-*
- split: test
path: atari-enduro/test-*
- config_name: atari-fishingderby
data_files:
- split: train
path: atari-fishingderby/train-*
- split: test
path: atari-fishingderby/test-*
- config_name: atari-freeway
data_files:
- split: train
path: atari-freeway/train-*
- split: test
path: atari-freeway/test-*
- config_name: atari-frostbite
data_files:
- split: train
path: atari-frostbite/train-*
- split: test
path: atari-frostbite/test-*
- config_name: atari-gopher
data_files:
- split: train
path: atari-gopher/train-*
- split: test
path: atari-gopher/test-*
- config_name: atari-gravitar
data_files:
- split: train
path: atari-gravitar/train-*
- split: test
path: atari-gravitar/test-*
- config_name: atari-hero
data_files:
- split: train
path: atari-hero/train-*
- split: test
path: atari-hero/test-*
- config_name: atari-icehockey
data_files:
- split: train
path: atari-icehockey/train-*
- split: test
path: atari-icehockey/test-*
- config_name: atari-jamesbond
data_files:
- split: train
path: atari-jamesbond/train-*
- split: test
path: atari-jamesbond/test-*
- config_name: atari-kangaroo
data_files:
- split: train
path: atari-kangaroo/train-*
- split: test
path: atari-kangaroo/test-*
- config_name: atari-krull
data_files:
- split: train
path: atari-krull/train-*
- split: test
path: atari-krull/test-*
- config_name: atari-kungfumaster
data_files:
- split: train
path: atari-kungfumaster/train-*
- split: test
path: atari-kungfumaster/test-*
- config_name: atari-montezumarevenge
data_files:
- split: train
path: atari-montezumarevenge/train-*
- split: test
path: atari-montezumarevenge/test-*
- config_name: atari-mspacman
data_files:
- split: train
path: atari-mspacman/train-*
- split: test
path: atari-mspacman/test-*
- config_name: atari-namethisgame
data_files:
- split: train
path: atari-namethisgame/train-*
- split: test
path: atari-namethisgame/test-*
- config_name: atari-phoenix
data_files:
- split: train
path: atari-phoenix/train-*
- split: test
path: atari-phoenix/test-*
- config_name: atari-pitfall
data_files:
- split: train
path: atari-pitfall/train-*
- split: test
path: atari-pitfall/test-*
- config_name: atari-pong
data_files:
- split: test
path: atari-pong/test-*
- split: train
path: atari-pong/train-*
- config_name: atari-privateeye
data_files:
- split: test
path: atari-privateeye/test-*
- split: train
path: atari-privateeye/train-*
- config_name: atari-qbert
data_files:
- split: test
path: atari-qbert/test-*
- split: train
path: atari-qbert/train-*
- config_name: atari-riverraid
data_files:
- split: test
path: atari-riverraid/test-*
- split: train
path: atari-riverraid/train-*
- config_name: atari-roadrunner
data_files:
- split: test
path: atari-roadrunner/test-*
- split: train
path: atari-roadrunner/train-*
- config_name: atari-robotank
data_files:
- split: test
path: atari-robotank/test-*
- split: train
path: atari-robotank/train-*
- config_name: atari-seaquest
data_files:
- split: test
path: atari-seaquest/test-*
- split: train
path: atari-seaquest/train-*
- config_name: atari-skiing
data_files:
- split: train
path: atari-skiing/train-*
- split: test
path: atari-skiing/test-*
- config_name: atari-solaris
data_files:
- split: train
path: atari-solaris/train-*
- split: test
path: atari-solaris/test-*
- config_name: atari-spaceinvaders
data_files:
- split: train
path: atari-spaceinvaders/train-*
- split: test
path: atari-spaceinvaders/test-*
- config_name: atari-stargunner
data_files:
- split: train
path: atari-stargunner/train-*
- split: test
path: atari-stargunner/test-*
- config_name: atari-surround
data_files:
- split: train
path: atari-surround/train-*
- split: test
path: atari-surround/test-*
- config_name: atari-tennis
data_files:
- split: train
path: atari-tennis/train-*
- split: test
path: atari-tennis/test-*
- config_name: atari-timepilot
data_files:
- split: train
path: atari-timepilot/train-*
- split: test
path: atari-timepilot/test-*
- config_name: atari-tutankham
data_files:
- split: train
path: atari-tutankham/train-*
- split: test
path: atari-tutankham/test-*
- config_name: atari-upndown
data_files:
- split: train
path: atari-upndown/train-*
- split: test
path: atari-upndown/test-*
- config_name: atari-venture
data_files:
- split: test
path: atari-venture/test-*
- split: train
path: atari-venture/train-*
- config_name: atari-videopinball
data_files:
- split: test
path: atari-videopinball/test-*
- split: train
path: atari-videopinball/train-*
- config_name: atari-wizardofwor
data_files:
- split: test
path: atari-wizardofwor/test-*
- split: train
path: atari-wizardofwor/train-*
- config_name: atari-yarsrevenge
data_files:
- split: test
path: atari-yarsrevenge/test-*
- split: train
path: atari-yarsrevenge/train-*
- config_name: atari-zaxxon
data_files:
- split: test
path: atari-zaxxon/test-*
- split: train
path: atari-zaxxon/train-*
- config_name: babyai-action-obj-door
data_files:
- split: train
path: babyai-action-obj-door/train-*
- split: test
path: babyai-action-obj-door/test-*
- config_name: babyai-blocked-unlock-pickup
data_files:
- split: test
path: babyai-blocked-unlock-pickup/test-*
- split: train
path: babyai-blocked-unlock-pickup/train-*
- config_name: babyai-boss-level
data_files:
- split: test
path: babyai-boss-level/test-*
- split: train
path: babyai-boss-level/train-*
- config_name: babyai-boss-level-no-unlock
data_files:
- split: test
path: babyai-boss-level-no-unlock/test-*
- split: train
path: babyai-boss-level-no-unlock/train-*
- config_name: babyai-find-obj-s5
data_files:
- split: train
path: babyai-find-obj-s5/train-*
- split: test
path: babyai-find-obj-s5/test-*
- config_name: babyai-go-to
data_files:
- split: train
path: babyai-go-to/train-*
- split: test
path: babyai-go-to/test-*
- config_name: babyai-go-to-door
data_files:
- split: train
path: babyai-go-to-door/train-*
- split: test
path: babyai-go-to-door/test-*
- config_name: babyai-go-to-imp-unlock
data_files:
- split: train
path: babyai-go-to-imp-unlock/train-*
- split: test
path: babyai-go-to-imp-unlock/test-*
- config_name: babyai-go-to-local
data_files:
- split: train
path: babyai-go-to-local/train-*
- split: test
path: babyai-go-to-local/test-*
- config_name: babyai-go-to-obj
data_files:
- split: train
path: babyai-go-to-obj/train-*
- split: test
path: babyai-go-to-obj/test-*
- config_name: babyai-go-to-obj-door
data_files:
- split: train
path: babyai-go-to-obj-door/train-*
- split: test
path: babyai-go-to-obj-door/test-*
- config_name: babyai-go-to-red-ball
data_files:
- split: train
path: babyai-go-to-red-ball/train-*
- split: test
path: babyai-go-to-red-ball/test-*
- config_name: babyai-go-to-red-ball-grey
data_files:
- split: train
path: babyai-go-to-red-ball-grey/train-*
- split: test
path: babyai-go-to-red-ball-grey/test-*
- config_name: babyai-go-to-red-ball-no-dists
data_files:
- split: train
path: babyai-go-to-red-ball-no-dists/train-*
- split: test
path: babyai-go-to-red-ball-no-dists/test-*
- config_name: babyai-go-to-red-blue-ball
data_files:
- split: train
path: babyai-go-to-red-blue-ball/train-*
- split: test
path: babyai-go-to-red-blue-ball/test-*
- config_name: babyai-go-to-seq
data_files:
- split: train
path: babyai-go-to-seq/train-*
- split: test
path: babyai-go-to-seq/test-*
- config_name: babyai-key-corridor
data_files:
- split: test
path: babyai-key-corridor/test-*
- split: train
path: babyai-key-corridor/train-*
- config_name: babyai-mini-boss-level
data_files:
- split: test
path: babyai-mini-boss-level/test-*
- split: train
path: babyai-mini-boss-level/train-*
- config_name: babyai-move-two-across-s8n9
data_files:
- split: test
path: babyai-move-two-across-s8n9/test-*
- split: train
path: babyai-move-two-across-s8n9/train-*
- config_name: babyai-one-room-s8
data_files:
- split: test
path: babyai-one-room-s8/test-*
- split: train
path: babyai-one-room-s8/train-*
- config_name: babyai-open
data_files:
- split: test
path: babyai-open/test-*
- split: train
path: babyai-open/train-*
- config_name: babyai-open-door
data_files:
- split: test
path: babyai-open-door/test-*
- split: train
path: babyai-open-door/train-*
- config_name: babyai-open-doors-order-n4
data_files:
- split: test
path: babyai-open-doors-order-n4/test-*
- split: train
path: babyai-open-doors-order-n4/train-*
- config_name: babyai-open-red-door
data_files:
- split: test
path: babyai-open-red-door/test-*
- split: train
path: babyai-open-red-door/train-*
- config_name: babyai-open-two-doors
data_files:
- split: test
path: babyai-open-two-doors/test-*
- split: train
path: babyai-open-two-doors/train-*
- config_name: babyai-pickup
data_files:
- split: test
path: babyai-pickup/test-*
- split: train
path: babyai-pickup/train-*
- config_name: babyai-pickup-above
data_files:
- split: test
path: babyai-pickup-above/test-*
- split: train
path: babyai-pickup-above/train-*
- config_name: babyai-pickup-dist
data_files:
- split: test
path: babyai-pickup-dist/test-*
- split: train
path: babyai-pickup-dist/train-*
- config_name: babyai-pickup-loc
data_files:
- split: test
path: babyai-pickup-loc/test-*
- split: train
path: babyai-pickup-loc/train-*
- config_name: babyai-put-next
data_files:
- split: train
path: babyai-put-next/train-*
- split: test
path: babyai-put-next/test-*
- config_name: babyai-put-next-local
data_files:
- split: train
path: babyai-put-next-local/train-*
- split: test
path: babyai-put-next-local/test-*
- config_name: babyai-synth
data_files:
- split: test
path: babyai-synth/test-*
- split: train
path: babyai-synth/train-*
- config_name: babyai-synth-loc
data_files:
- split: test
path: babyai-synth-loc/test-*
- split: train
path: babyai-synth-loc/train-*
- config_name: babyai-synth-seq
data_files:
- split: test
path: babyai-synth-seq/test-*
- split: train
path: babyai-synth-seq/train-*
- config_name: babyai-unblock-pickup
data_files:
- split: test
path: babyai-unblock-pickup/test-*
- split: train
path: babyai-unblock-pickup/train-*
- config_name: babyai-unlock
data_files:
- split: train
path: babyai-unlock/train-*
- split: test
path: babyai-unlock/test-*
- config_name: babyai-unlock-local
data_files:
- split: test
path: babyai-unlock-local/test-*
- split: train
path: babyai-unlock-local/train-*
- config_name: babyai-unlock-pickup
data_files:
- split: test
path: babyai-unlock-pickup/test-*
- split: train
path: babyai-unlock-pickup/train-*
- config_name: babyai-unlock-to-unlock
data_files:
- split: train
path: babyai-unlock-to-unlock/train-*
- split: test
path: babyai-unlock-to-unlock/test-*
- config_name: conceptual-captions
data_files:
- split: test
path: conceptual-captions/test-*
- split: train
path: conceptual-captions/train-*
- config_name: metaworld-assembly
data_files:
- split: train
path: metaworld-assembly/train-*
- split: test
path: metaworld-assembly/test-*
- config_name: metaworld-basketball
data_files:
- split: train
path: metaworld-basketball/train-*
- split: test
path: metaworld-basketball/test-*
- config_name: metaworld-bin-picking
data_files:
- split: train
path: metaworld-bin-picking/train-*
- split: test
path: metaworld-bin-picking/test-*
- config_name: metaworld-box-close
data_files:
- split: train
path: metaworld-box-close/train-*
- split: test
path: metaworld-box-close/test-*
- config_name: metaworld-button-press
data_files:
- split: train
path: metaworld-button-press/train-*
- split: test
path: metaworld-button-press/test-*
- config_name: metaworld-button-press-topdown
data_files:
- split: train
path: metaworld-button-press-topdown/train-*
- split: test
path: metaworld-button-press-topdown/test-*
- config_name: metaworld-button-press-topdown-wall
data_files:
- split: train
path: metaworld-button-press-topdown-wall/train-*
- split: test
path: metaworld-button-press-topdown-wall/test-*
- config_name: metaworld-button-press-wall
data_files:
- split: train
path: metaworld-button-press-wall/train-*
- split: test
path: metaworld-button-press-wall/test-*
- config_name: metaworld-coffee-button
data_files:
- split: train
path: metaworld-coffee-button/train-*
- split: test
path: metaworld-coffee-button/test-*
- config_name: metaworld-coffee-pull
data_files:
- split: train
path: metaworld-coffee-pull/train-*
- split: test
path: metaworld-coffee-pull/test-*
- config_name: metaworld-coffee-push
data_files:
- split: train
path: metaworld-coffee-push/train-*
- split: test
path: metaworld-coffee-push/test-*
- config_name: metaworld-dial-turn
data_files:
- split: train
path: metaworld-dial-turn/train-*
- split: test
path: metaworld-dial-turn/test-*
- config_name: metaworld-disassemble
data_files:
- split: train
path: metaworld-disassemble/train-*
- split: test
path: metaworld-disassemble/test-*
- config_name: metaworld-door-close
data_files:
- split: train
path: metaworld-door-close/train-*
- split: test
path: metaworld-door-close/test-*
- config_name: metaworld-door-lock
data_files:
- split: train
path: metaworld-door-lock/train-*
- split: test
path: metaworld-door-lock/test-*
- config_name: metaworld-door-open
data_files:
- split: train
path: metaworld-door-open/train-*
- split: test
path: metaworld-door-open/test-*
- config_name: metaworld-door-unlock
data_files:
- split: train
path: metaworld-door-unlock/train-*
- split: test
path: metaworld-door-unlock/test-*
- config_name: metaworld-drawer-close
data_files:
- split: train
path: metaworld-drawer-close/train-*
- split: test
path: metaworld-drawer-close/test-*
- config_name: metaworld-drawer-open
data_files:
- split: train
path: metaworld-drawer-open/train-*
- split: test
path: metaworld-drawer-open/test-*
- config_name: metaworld-faucet-close
data_files:
- split: train
path: metaworld-faucet-close/train-*
- split: test
path: metaworld-faucet-close/test-*
- config_name: metaworld-faucet-open
data_files:
- split: train
path: metaworld-faucet-open/train-*
- split: test
path: metaworld-faucet-open/test-*
- config_name: metaworld-hammer
data_files:
- split: train
path: metaworld-hammer/train-*
- split: test
path: metaworld-hammer/test-*
- config_name: metaworld-hand-insert
data_files:
- split: train
path: metaworld-hand-insert/train-*
- split: test
path: metaworld-hand-insert/test-*
- config_name: metaworld-handle-press
data_files:
- split: train
path: metaworld-handle-press/train-*
- split: test
path: metaworld-handle-press/test-*
- config_name: metaworld-handle-press-side
data_files:
- split: train
path: metaworld-handle-press-side/train-*
- split: test
path: metaworld-handle-press-side/test-*
- config_name: metaworld-handle-pull
data_files:
- split: train
path: metaworld-handle-pull/train-*
- split: test
path: metaworld-handle-pull/test-*
- config_name: metaworld-handle-pull-side
data_files:
- split: train
path: metaworld-handle-pull-side/train-*
- split: test
path: metaworld-handle-pull-side/test-*
- config_name: metaworld-lever-pull
data_files:
- split: train
path: metaworld-lever-pull/train-*
- split: test
path: metaworld-lever-pull/test-*
- config_name: metaworld-peg-insert-side
data_files:
- split: train
path: metaworld-peg-insert-side/train-*
- split: test
path: metaworld-peg-insert-side/test-*
- config_name: metaworld-peg-unplug-side
data_files:
- split: train
path: metaworld-peg-unplug-side/train-*
- split: test
path: metaworld-peg-unplug-side/test-*
- config_name: metaworld-pick-out-of-hole
data_files:
- split: train
path: metaworld-pick-out-of-hole/train-*
- split: test
path: metaworld-pick-out-of-hole/test-*
- config_name: metaworld-pick-place
data_files:
- split: train
path: metaworld-pick-place/train-*
- split: test
path: metaworld-pick-place/test-*
- config_name: metaworld-pick-place-wall
data_files:
- split: train
path: metaworld-pick-place-wall/train-*
- split: test
path: metaworld-pick-place-wall/test-*
- config_name: metaworld-plate-slide
data_files:
- split: train
path: metaworld-plate-slide/train-*
- split: test
path: metaworld-plate-slide/test-*
- config_name: metaworld-plate-slide-back
data_files:
- split: train
path: metaworld-plate-slide-back/train-*
- split: test
path: metaworld-plate-slide-back/test-*
- config_name: metaworld-plate-slide-back-side
data_files:
- split: train
path: metaworld-plate-slide-back-side/train-*
- split: test
path: metaworld-plate-slide-back-side/test-*
- config_name: metaworld-plate-slide-side
data_files:
- split: train
path: metaworld-plate-slide-side/train-*
- split: test
path: metaworld-plate-slide-side/test-*
- config_name: metaworld-push
data_files:
- split: train
path: metaworld-push/train-*
- split: test
path: metaworld-push/test-*
- config_name: metaworld-push-back
data_files:
- split: train
path: metaworld-push-back/train-*
- split: test
path: metaworld-push-back/test-*
- config_name: metaworld-push-wall
data_files:
- split: train
path: metaworld-push-wall/train-*
- split: test
path: metaworld-push-wall/test-*
- config_name: metaworld-reach
data_files:
- split: train
path: metaworld-reach/train-*
- split: test
path: metaworld-reach/test-*
- config_name: metaworld-reach-wall
data_files:
- split: train
path: metaworld-reach-wall/train-*
- split: test
path: metaworld-reach-wall/test-*
- config_name: metaworld-shelf-place
data_files:
- split: train
path: metaworld-shelf-place/train-*
- split: test
path: metaworld-shelf-place/test-*
- config_name: metaworld-soccer
data_files:
- split: train
path: metaworld-soccer/train-*
- split: test
path: metaworld-soccer/test-*
- config_name: metaworld-stick-pull
data_files:
- split: train
path: metaworld-stick-pull/train-*
- split: test
path: metaworld-stick-pull/test-*
- config_name: metaworld-stick-push
data_files:
- split: train
path: metaworld-stick-push/train-*
- split: test
path: metaworld-stick-push/test-*
- config_name: metaworld-sweep
data_files:
- split: train
path: metaworld-sweep/train-*
- split: test
path: metaworld-sweep/test-*
- config_name: metaworld-sweep-into
data_files:
- split: train
path: metaworld-sweep-into/train-*
- split: test
path: metaworld-sweep-into/test-*
- config_name: metaworld-window-close
data_files:
- split: train
path: metaworld-window-close/train-*
- split: test
path: metaworld-window-close/test-*
- config_name: metaworld-window-open
data_files:
- split: train
path: metaworld-window-open/train-*
- split: test
path: metaworld-window-open/test-*
- config_name: mujoco-ant
data_files:
- split: train
path: mujoco-ant/train-*
- split: test
path: mujoco-ant/test-*
- config_name: mujoco-doublependulum
data_files:
- split: train
path: mujoco-doublependulum/train-*
- split: test
path: mujoco-doublependulum/test-*
- config_name: mujoco-halfcheetah
data_files:
- split: train
path: mujoco-halfcheetah/train-*
- split: test
path: mujoco-halfcheetah/test-*
- config_name: mujoco-hopper
data_files:
- split: train
path: mujoco-hopper/train-*
- split: test
path: mujoco-hopper/test-*
- config_name: mujoco-humanoid
data_files:
- split: train
path: mujoco-humanoid/train-*
- split: test
path: mujoco-humanoid/test-*
- config_name: mujoco-pendulum
data_files:
- split: train
path: mujoco-pendulum/train-*
- split: test
path: mujoco-pendulum/test-*
- config_name: mujoco-pusher
data_files:
- split: train
path: mujoco-pusher/train-*
- split: test
path: mujoco-pusher/test-*
- config_name: mujoco-reacher
data_files:
- split: train
path: mujoco-reacher/train-*
- split: test
path: mujoco-reacher/test-*
- config_name: mujoco-standup
data_files:
- split: train
path: mujoco-standup/train-*
- split: test
path: mujoco-standup/test-*
- config_name: mujoco-swimmer
data_files:
- split: train
path: mujoco-swimmer/train-*
- split: test
path: mujoco-swimmer/test-*
- config_name: mujoco-walker
data_files:
- split: train
path: mujoco-walker/train-*
- split: test
path: mujoco-walker/test-*
- config_name: ok-vqa
data_files:
- split: train
path: ok-vqa/train-*
- split: test
path: ok-vqa/test-*
- config_name: oscar
data_files:
- split: train
path: oscar/train-*
- split: test
path: oscar/test-*
- config_name: wikipedia
data_files:
- split: train
path: wikipedia/train-*
- split: test
path: wikipedia/test-*
tags:
- imitation-learning
- reinforcement-learning
- text-generation
- question-answering
- generalist-agent
dataset_info:
- config_name: atari-alien
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1340568536.0
num_examples: 97
- name: test
num_bytes: 140147997.0
num_examples: 11
download_size: 139482052
dataset_size: 1480716533.0
- config_name: atari-amidar
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 839195896.0
num_examples: 146
- name: test
num_bytes: 76328889.0
num_examples: 17
download_size: 849996308
dataset_size: 915524785.0
- config_name: atari-assault
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 798961431.0
num_examples: 53
- name: test
num_bytes: 70630737.0
num_examples: 6
download_size: 856465142
dataset_size: 869592168.0
- config_name: atari-asterix
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 981904668.0
num_examples: 470
- name: test
num_bytes: 94826831.0
num_examples: 53
download_size: 1025083959
dataset_size: 1076731499.0
- config_name: atari-asteroids
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 774344616.0
num_examples: 17
- name: test
num_bytes: 52617462.0
num_examples: 2
download_size: 815573512
dataset_size: 826962078.0
- config_name: atari-atlantis
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 915242786.0
num_examples: 44
- name: test
num_bytes: 68743372.0
num_examples: 5
download_size: 969604640
dataset_size: 983986158.0
- config_name: atari-bankheist
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1623230516.0
num_examples: 222
- name: test
num_bytes: 182769923.0
num_examples: 25
download_size: 1743163262
dataset_size: 1806000439.0
- config_name: atari-battlezone
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1406320758.0
num_examples: 97
- name: test
num_bytes: 167008797.0
num_examples: 11
download_size: 640049534
dataset_size: 1573329555.0
- config_name: atari-beamrider
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1028942918.0
num_examples: 46
- name: test
num_bytes: 165781602.0
num_examples: 6
download_size: 1190822803
dataset_size: 1194724520.0
- config_name: atari-berzerk
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 599497245.0
num_examples: 17
- name: test
num_bytes: 75010244.0
num_examples: 2
download_size: 652845047
dataset_size: 674507489.0
- config_name: atari-bowling
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 546770697.0
num_examples: 193
- name: test
num_bytes: 62611921.0
num_examples: 22
download_size: 534548773
dataset_size: 609382618.0
- config_name: atari-boxing
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1081525678.975
num_examples: 1025
- name: test
num_bytes: 119411032.0
num_examples: 114
download_size: 1196687855
dataset_size: 1200936710.975
- config_name: atari-breakout
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 449338850.0
num_examples: 32
- name: test
num_bytes: 57704753.0
num_examples: 4
download_size: 355232930
dataset_size: 507043603.0
- config_name: atari-centipede
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 740721041.0
num_examples: 460
- name: test
num_bytes: 85208346.0
num_examples: 52
download_size: 819207107
dataset_size: 825929387.0
- config_name: atari-choppercommand
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 989964507.0
num_examples: 144
- name: test
num_bytes: 147199310.0
num_examples: 16
download_size: 1131175930
dataset_size: 1137163817.0
- config_name: atari-crazyclimber
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1246068403.0
num_examples: 88
- name: test
num_bytes: 139541935.0
num_examples: 10
download_size: 1294452085
dataset_size: 1385610338.0
- config_name: atari-defender
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 631539225.0
num_examples: 16
- name: test
num_bytes: 78383287.0
num_examples: 2
download_size: 620482245
dataset_size: 709922512.0
- config_name: atari-demonattack
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 624524718.0
num_examples: 18
- name: test
num_bytes: 77648737.0
num_examples: 2
download_size: 692930877
dataset_size: 702173455.0
- config_name: atari-doubledunk
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 123241754.0
num_examples: 51
- name: train
num_bytes: 1109840257.0
num_examples: 456
download_size: 1208221748
dataset_size: 1233082011.0
- config_name: atari-enduro
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1341529954.0
num_examples: 16
- name: test
num_bytes: 170147714.0
num_examples: 2
download_size: 1506759932
dataset_size: 1511677668.0
- config_name: atari-fishingderby
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1515746411.0
num_examples: 275
- name: test
num_bytes: 179086977.0
num_examples: 31
download_size: 1692400820
dataset_size: 1694833388.0
- config_name: atari-freeway
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1109519748.0
num_examples: 219
- name: test
num_bytes: 126516219.0
num_examples: 25
download_size: 1232267662
dataset_size: 1236035967.0
- config_name: atari-frostbite
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1461470198.0
num_examples: 188
- name: test
num_bytes: 168294758.0
num_examples: 21
download_size: 1623699715
dataset_size: 1629764956.0
- config_name: atari-gopher
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 838220280.0
num_examples: 23
- name: test
num_bytes: 112043092.0
num_examples: 3
download_size: 942000464
dataset_size: 950263372.0
- config_name: atari-gravitar
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 795642642.0
num_examples: 750
- name: test
num_bytes: 88650726.0
num_examples: 84
download_size: 877506629
dataset_size: 884293368.0
- config_name: atari-hero
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1093415256.0
num_examples: 166
- name: test
num_bytes: 125418914.0
num_examples: 19
download_size: 1203346008
dataset_size: 1218834170.0
- config_name: atari-icehockey
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 764843072.0
num_examples: 118
- name: test
num_bytes: 87267657.0
num_examples: 14
download_size: 778055672
dataset_size: 852110729.0
- config_name: atari-jamesbond
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 735033584.0
num_examples: 54
- name: test
num_bytes: 168937080.0
num_examples: 7
download_size: 899088453
dataset_size: 903970664.0
- config_name: atari-kangaroo
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1040140729.0
num_examples: 495
- name: test
num_bytes: 112177810.0
num_examples: 56
download_size: 1148401746
dataset_size: 1152318539.0
- config_name: atari-krull
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 2283525995.0
num_examples: 318
- name: test
num_bytes: 253656157.0
num_examples: 36
download_size: 2526820904
dataset_size: 2537182152.0
- config_name: atari-kungfumaster
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1459405811.0
num_examples: 150
- name: test
num_bytes: 175710328.0
num_examples: 17
download_size: 1609871392
dataset_size: 1635116139.0
- config_name: atari-montezumarevenge
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1358041617.0
num_examples: 389
- name: test
num_bytes: 151969510.0
num_examples: 44
download_size: 1496389769
dataset_size: 1510011127.0
- config_name: atari-mspacman
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1450638504.0
num_examples: 179
- name: test
num_bytes: 158188150.0
num_examples: 20
download_size: 157083760
dataset_size: 1608826654.0
- config_name: atari-namethisgame
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1303134716.0
num_examples: 45
- name: test
num_bytes: 180906060.0
num_examples: 6
download_size: 1480907677
dataset_size: 1484040776.0
- config_name: atari-phoenix
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 710710054.0
num_examples: 17
- name: test
num_bytes: 90041382.0
num_examples: 2
download_size: 789132045
dataset_size: 800751436.0
- config_name: atari-pitfall
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1038921456.0
num_examples: 42
- name: test
num_bytes: 95477942.0
num_examples: 5
download_size: 563920504
dataset_size: 1134399398.0
- config_name: atari-pong
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 42460330.0
num_examples: 31
- name: train
num_bytes: 372438874.0
num_examples: 272
download_size: 340157509
dataset_size: 414899204.0
- config_name: atari-privateeye
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 188566614.0
num_examples: 19
- name: train
num_bytes: 1646331664.0
num_examples: 166
download_size: 999585816
dataset_size: 1834898278.0
- config_name: atari-qbert
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 212314952.0
num_examples: 12
- name: train
num_bytes: 1906885976.0
num_examples: 105
download_size: 2114236276
dataset_size: 2119200928.0
- config_name: atari-riverraid
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 138639529.0
num_examples: 31
- name: train
num_bytes: 1336041601.0
num_examples: 277
download_size: 1451357887
dataset_size: 1474681130.0
- config_name: atari-roadrunner
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 102119437.0
num_examples: 24
- name: train
num_bytes: 913351876.0
num_examples: 212
download_size: 1001454818
dataset_size: 1015471313.0
- config_name: atari-robotank
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 128435803.0
num_examples: 7
- name: train
num_bytes: 1292214032.0
num_examples: 63
download_size: 1388205947
dataset_size: 1420649835.0
- config_name: atari-seaquest
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 91834003.0
num_examples: 24
- name: train
num_bytes: 828174074.0
num_examples: 209
download_size: 908365754
dataset_size: 920008077.0
- config_name: atari-skiing
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1141286076.0
num_examples: 917
- name: test
num_bytes: 127551492.0
num_examples: 102
download_size: 1265105500
dataset_size: 1268837568.0
- config_name: atari-solaris
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 1146266482.0
num_examples: 34
- name: test
num_bytes: 122871787.0
num_examples: 4
download_size: 1257863864
dataset_size: 1269138269.0
- config_name: atari-spaceinvaders
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 888515140.0
num_examples: 30
- name: test
num_bytes: 183628032.0
num_examples: 4
download_size: 1044841686
dataset_size: 1072143172.0
- config_name: atari-stargunner
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 615092285.0
num_examples: 31
- name: test
num_bytes: 71315788.0
num_examples: 4
download_size: 677077474
dataset_size: 686408073.0
- config_name: atari-surround
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 526004197.0
num_examples: 144
- name: test
num_bytes: 67282927.0
num_examples: 17
download_size: 532120267
dataset_size: 593287124.0
- config_name: atari-tennis
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 709632525.0
num_examples: 49
- name: test
num_bytes: 76212648.0
num_examples: 6
download_size: 539956655
dataset_size: 785845173.0
- config_name: atari-timepilot
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 849962378.0
num_examples: 48
- name: test
num_bytes: 95939303.0
num_examples: 6
download_size: 919663541
dataset_size: 945901681.0
- config_name: atari-tutankham
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 833317180.0
num_examples: 27
- name: test
num_bytes: 137596199.0
num_examples: 4
download_size: 528781594
dataset_size: 970913379.0
- config_name: atari-upndown
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: train
num_bytes: 2963452811.0
num_examples: 16
- name: test
num_bytes: 371856958.0
num_examples: 2
download_size: 3320647022
dataset_size: 3335309769.0
- config_name: atari-venture
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 88888187.0
num_examples: 25
- name: train
num_bytes: 884080096.0
num_examples: 216
download_size: 869134091
dataset_size: 972968283.0
- config_name: atari-videopinball
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 50315326.0
num_examples: 3
- name: train
num_bytes: 1330483745.0
num_examples: 22
download_size: 1377534468
dataset_size: 1380799071.0
- config_name: atari-wizardofwor
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 121295756.0
num_examples: 14
- name: train
num_bytes: 1015986420.0
num_examples: 124
download_size: 1082615829
dataset_size: 1137282176.0
- config_name: atari-yarsrevenge
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 278195918.0
num_examples: 4
- name: train
num_bytes: 2348309471.0
num_examples: 31
download_size: 1988218999
dataset_size: 2626505389.0
- config_name: atari-zaxxon
features:
- name: image_observations
sequence: image
- name: rewards
sequence: float32
- name: discrete_actions
sequence: int64
splits:
- name: test
num_bytes: 117311384.0
num_examples: 8
- name: train
num_bytes: 982507552.0
num_examples: 64
download_size: 1093792295
dataset_size: 1099818936.0
- config_name: babyai-action-obj-door
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 730102581
num_examples: 95000
- name: test
num_bytes: 38820823
num_examples: 5000
download_size: 15937785
dataset_size: 768923404
- config_name: babyai-blocked-unlock-pickup
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 207846215
num_examples: 5000
- name: train
num_bytes: 3944315285
num_examples: 95000
download_size: 47671576
dataset_size: 4152161500
- config_name: babyai-boss-level
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 524421727
num_examples: 5000
- name: train
num_bytes: 10122220692
num_examples: 95000
download_size: 171013846
dataset_size: 10646642419
- config_name: babyai-boss-level-no-unlock
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 512206014
num_examples: 5000
- name: train
num_bytes: 9951813143
num_examples: 95000
download_size: 166637143
dataset_size: 10464019157
- config_name: babyai-find-obj-s5
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 3525778032
num_examples: 95000
- name: test
num_bytes: 183685740
num_examples: 5000
download_size: 49738428
dataset_size: 3709463772
- config_name: babyai-go-to
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 6152451450
num_examples: 95000
- name: test
num_bytes: 319842603
num_examples: 5000
download_size: 101378644
dataset_size: 6472294053
- config_name: babyai-go-to-door
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 615768109
num_examples: 95000
- name: test
num_bytes: 32599120
num_examples: 5000
download_size: 8940753
dataset_size: 648367229
- config_name: babyai-go-to-imp-unlock
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 13079777079.88
num_examples: 98000
- name: test
num_bytes: 266934226.12
num_examples: 2000
download_size: 222137618
dataset_size: 13346711306.0
- config_name: babyai-go-to-local
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 618625078
num_examples: 95000
- name: test
num_bytes: 32783633
num_examples: 5000
download_size: 14568281
dataset_size: 651408711
- config_name: babyai-go-to-obj
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 576503446
num_examples: 95000
- name: test
num_bytes: 30207684
num_examples: 5000
download_size: 8102560
dataset_size: 606711130
- config_name: babyai-go-to-obj-door
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 698247097
num_examples: 95000
- name: test
num_bytes: 36554007
num_examples: 5000
download_size: 18138758
dataset_size: 734801104
- config_name: babyai-go-to-red-ball
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 617255758
num_examples: 95000
- name: test
num_bytes: 32552614
num_examples: 5000
download_size: 14101801
dataset_size: 649808372
- config_name: babyai-go-to-red-ball-grey
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 685059164
num_examples: 95000
- name: test
num_bytes: 36316718
num_examples: 5000
download_size: 14234379
dataset_size: 721375882
- config_name: babyai-go-to-red-ball-no-dists
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 575338070
num_examples: 95000
- name: test
num_bytes: 30355826
num_examples: 5000
download_size: 7108473
dataset_size: 605693896
- config_name: babyai-go-to-red-blue-ball
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 684110113
num_examples: 95000
- name: test
num_bytes: 36050340
num_examples: 5000
download_size: 15617708
dataset_size: 720160453
- config_name: babyai-go-to-seq
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 8659717841
num_examples: 95000
- name: test
num_bytes: 457950086
num_examples: 5000
download_size: 142792284
dataset_size: 9117667927
- config_name: babyai-key-corridor
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 673861952
num_examples: 5000
- name: train
num_bytes: 12830544960
num_examples: 95000
download_size: 192785385
dataset_size: 13504406912
- config_name: babyai-mini-boss-level
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 165697671
num_examples: 5000
- name: train
num_bytes: 3160839261
num_examples: 95000
download_size: 49046590
dataset_size: 3326536932
- config_name: babyai-move-two-across-s8n9
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 263104296
num_examples: 5000
- name: train
num_bytes: 5010029188
num_examples: 95000
download_size: 67260892
dataset_size: 5273133484
- config_name: babyai-one-room-s8
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 35849856
num_examples: 5000
- name: train
num_bytes: 678323712
num_examples: 95000
download_size: 8726372
dataset_size: 714173568
- config_name: babyai-open
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 184341054
num_examples: 5000
- name: train
num_bytes: 3552284018
num_examples: 95000
download_size: 2850718
dataset_size: 3736625072
- config_name: babyai-open-door
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 44954852
num_examples: 5000
- name: train
num_bytes: 857776914
num_examples: 95000
download_size: 11397484
dataset_size: 902731766
- config_name: babyai-open-doors-order-n4
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 65109790
num_examples: 5000
- name: train
num_bytes: 1224959587
num_examples: 95000
download_size: 14918459
dataset_size: 1290069377
- config_name: babyai-open-red-door
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 28865701
num_examples: 5000
- name: train
num_bytes: 547345717
num_examples: 95000
download_size: 2723624
dataset_size: 576211418
- config_name: babyai-open-two-doors
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 85096451
num_examples: 5000
- name: train
num_bytes: 1614499890
num_examples: 95000
download_size: 12535076
dataset_size: 1699596341
- config_name: babyai-pickup
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 324751988
num_examples: 5000
- name: train
num_bytes: 6247776138
num_examples: 95000
download_size: 103094535
dataset_size: 6572528126
- config_name: babyai-pickup-above
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 181653115
num_examples: 5000
- name: train
num_bytes: 3399366642
num_examples: 95000
download_size: 47780316
dataset_size: 3581019757
- config_name: babyai-pickup-dist
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 29384140
num_examples: 5000
- name: train
num_bytes: 555920169
num_examples: 95000
download_size: 10606303
dataset_size: 585304309
- config_name: babyai-pickup-loc
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 36556968
num_examples: 5000
- name: train
num_bytes: 709012750
num_examples: 95000
download_size: 15292435
dataset_size: 745569718
- config_name: babyai-put-next
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 2139199682.62
num_examples: 98000
- name: test
num_bytes: 43657136.38
num_examples: 2000
download_size: 41550541
dataset_size: 2182856819.0
- config_name: babyai-put-next-local
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 1467122290.76
num_examples: 98000
- name: test
num_bytes: 29941271.24
num_examples: 2000
download_size: 31329711
dataset_size: 1497063562.0
- config_name: babyai-synth
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 307405687
num_examples: 5000
- name: train
num_bytes: 5948279603
num_examples: 95000
download_size: 100838075
dataset_size: 6255685290
- config_name: babyai-synth-loc
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 290016584
num_examples: 5000
- name: train
num_bytes: 5488393137
num_examples: 95000
download_size: 93570653
dataset_size: 5778409721
- config_name: babyai-synth-seq
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 489211184
num_examples: 5000
- name: train
num_bytes: 9238807765
num_examples: 95000
download_size: 140373267
dataset_size: 9728018949
- config_name: babyai-unblock-pickup
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 349148205
num_examples: 5000
- name: train
num_bytes: 6483599187
num_examples: 95000
download_size: 109831237
dataset_size: 6832747392
- config_name: babyai-unlock
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 10242834097.44
num_examples: 98000
- name: test
num_bytes: 209037430.56
num_examples: 2000
download_size: 189691513
dataset_size: 10451871528.0
- config_name: babyai-unlock-local
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 85036094
num_examples: 5000
- name: train
num_bytes: 1620777960
num_examples: 95000
download_size: 21461309
dataset_size: 1705814054
- config_name: babyai-unlock-pickup
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
length: 148
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: test
num_bytes: 120199548
num_examples: 5000
- name: train
num_bytes: 2279983679
num_examples: 95000
download_size: 26099013
dataset_size: 2400183227
- config_name: babyai-unlock-to-unlock
features:
- name: text_observations
sequence: string
- name: discrete_observations
sequence:
sequence: int64
- name: discrete_actions
sequence: int64
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 5179083910.0
num_examples: 98000
- name: test
num_bytes: 105695590.0
num_examples: 2000
download_size: 65725587
dataset_size: 5284779500.0
- config_name: conceptual-captions
features:
- name: images
dtype: image
- name: text
dtype: string
splits:
- name: test
num_bytes: 1564922274.875
num_examples: 12465
- name: train
num_bytes: 321742591779.0
num_examples: 2620472
download_size: 7559495686
dataset_size: 323307514053.875
- config_name: metaworld-assembly
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 31556512
dataset_size: 309971200
- config_name: metaworld-basketball
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 13457975
dataset_size: 309971200
- config_name: metaworld-bin-picking
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 148239551
dataset_size: 309971200
- config_name: metaworld-box-close
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 155046141
dataset_size: 309971200
- config_name: metaworld-button-press
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 92407404
dataset_size: 309971200
- config_name: metaworld-button-press-topdown
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 99643997
dataset_size: 309971200
- config_name: metaworld-button-press-topdown-wall
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 102330609
dataset_size: 309971200
- config_name: metaworld-button-press-wall
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 98686929
dataset_size: 309971200
- config_name: metaworld-coffee-button
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 98541376
dataset_size: 309971200
- config_name: metaworld-coffee-pull
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 141657803
dataset_size: 309971200
- config_name: metaworld-coffee-push
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 153493123
dataset_size: 309971200
- config_name: metaworld-dial-turn
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 90092180
dataset_size: 309971200
- config_name: metaworld-disassemble
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 55699141
dataset_size: 309971200
- config_name: metaworld-door-close
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 132047898
dataset_size: 309971200
- config_name: metaworld-door-lock
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 108135090
dataset_size: 309971200
- config_name: metaworld-door-open
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 123463142
dataset_size: 309971200
- config_name: metaworld-door-unlock
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 107047389
dataset_size: 309971200
- config_name: metaworld-drawer-close
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 86742866
dataset_size: 309971200
- config_name: metaworld-drawer-open
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 87426230
dataset_size: 309971200
- config_name: metaworld-faucet-close
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 75525957
dataset_size: 309971200
- config_name: metaworld-faucet-open
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 82798110
dataset_size: 309971200
- config_name: metaworld-hammer
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 156766229
dataset_size: 309971200
- config_name: metaworld-hand-insert
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 115425570
dataset_size: 309971200
- config_name: metaworld-handle-press
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 88721833
dataset_size: 309971200
- config_name: metaworld-handle-press-side
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 90271855
dataset_size: 309971200
- config_name: metaworld-handle-pull
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 106520317
dataset_size: 309971200
- config_name: metaworld-handle-pull-side
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 104725703
dataset_size: 309971200
- config_name: metaworld-lever-pull
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 147893313
dataset_size: 309971200
- config_name: metaworld-peg-insert-side
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 133765390
dataset_size: 309971200
- config_name: metaworld-peg-unplug-side
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 152488362
dataset_size: 309971200
- config_name: metaworld-pick-out-of-hole
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 15063825
dataset_size: 309971200
- config_name: metaworld-pick-place
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 156685126
dataset_size: 309971200
- config_name: metaworld-pick-place-wall
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 152697114
dataset_size: 309971200
- config_name: metaworld-plate-slide
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 91689118
dataset_size: 309971200
- config_name: metaworld-plate-slide-back
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 17682663
dataset_size: 309971200
- config_name: metaworld-plate-slide-back-side
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 16397415
dataset_size: 309971200
- config_name: metaworld-plate-slide-side
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 88672818
dataset_size: 309971200
- config_name: metaworld-push
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 146425498
dataset_size: 309971200
- config_name: metaworld-push-back
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 115758693
dataset_size: 309971200
- config_name: metaworld-push-wall
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 138978942
dataset_size: 309971200
- config_name: metaworld-reach
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 151264193
dataset_size: 309971200
- config_name: metaworld-reach-wall
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 153008204
dataset_size: 309971200
- config_name: metaworld-shelf-place
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 126421788
dataset_size: 309971200
- config_name: metaworld-soccer
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 139325515
dataset_size: 309971200
- config_name: metaworld-stick-pull
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 150611675
dataset_size: 309971200
- config_name: metaworld-stick-push
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 145549289
dataset_size: 309971200
- config_name: metaworld-sweep
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 144411349
dataset_size: 309971200
- config_name: metaworld-sweep-into
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 116977226
dataset_size: 309971200
- config_name: metaworld-window-close
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 82738762
dataset_size: 309971200
- config_name: metaworld-window-open
features:
- name: continuous_observations
sequence:
sequence: float32
length: 39
- name: continuous_actions
sequence:
sequence: float32
length: 4
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 281792000
num_examples: 16000
- name: test
num_bytes: 28179200
num_examples: 1600
download_size: 82547802
dataset_size: 309971200
- config_name: mujoco-ant
features:
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 1334666176
num_examples: 9000
- name: test
num_bytes: 149007264
num_examples: 1000
download_size: 1427489194
dataset_size: 1483673440
- config_name: mujoco-doublependulum
features:
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 539380200
num_examples: 9000
- name: test
num_bytes: 59838360
num_examples: 1000
download_size: 423057943
dataset_size: 599218560
- config_name: mujoco-halfcheetah
features:
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 936108000
num_examples: 9000
- name: test
num_bytes: 104012000
num_examples: 1000
download_size: 983767586
dataset_size: 1040120000
- config_name: mujoco-hopper
features:
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 277504480
num_examples: 9000
- name: test
num_bytes: 30493476
num_examples: 1000
download_size: 291016996
dataset_size: 307997956
- config_name: mujoco-humanoid
features:
- name: continuous_observations
sequence:
sequence: float32
- name: rewards
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
splits:
- name: train
num_bytes: 12855318192
num_examples: 9000
- name: test
num_bytes: 1436554272
num_examples: 1000
download_size: 10321727430
dataset_size: 14291872464
- config_name: mujoco-pendulum
features:
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 137118592
num_examples: 9000
- name: test
num_bytes: 15128704
num_examples: 1000
download_size: 107926228
dataset_size: 152247296
- config_name: mujoco-pusher
features:
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 118908000
num_examples: 9000
- name: test
num_bytes: 13212000
num_examples: 1000
download_size: 124763158
dataset_size: 132120000
- config_name: mujoco-reacher
features:
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 28908000
num_examples: 9000
- name: test
num_bytes: 3212000
num_examples: 1000
download_size: 34000959
dataset_size: 32120000
- config_name: mujoco-standup
features:
- name: rewards
sequence: float32
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
splits:
- name: train
num_bytes: 14256108000
num_examples: 9000
- name: test
num_bytes: 1584012000
num_examples: 1000
download_size: 1163281621
dataset_size: 15840120000
- config_name: mujoco-swimmer
features:
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 468108000
num_examples: 9000
- name: test
num_bytes: 52012000
num_examples: 1000
download_size: 459798751
dataset_size: 520120000
- config_name: mujoco-walker
features:
- name: continuous_observations
sequence:
sequence: float32
- name: continuous_actions
sequence:
sequence: float32
- name: rewards
sequence: float32
splits:
- name: train
num_bytes: 858590040
num_examples: 9000
- name: test
num_bytes: 95183024
num_examples: 1000
download_size: 892883623
dataset_size: 953773064
- config_name: ok-vqa
features:
- name: images
dtype: image
- name: text
dtype: string
splits:
- name: train
num_bytes: 149757863.0
num_examples: 9009
- name: test
num_bytes: 84544434.0
num_examples: 5046
download_size: 233832618
dataset_size: 234302297.0
- config_name: oscar
features:
- name: text
dtype: string
splits:
- name: train
num_bytes: 978937483730
num_examples: 232133013
- name: test
num_bytes: 59798696914
num_examples: 12329126
download_size: 0
dataset_size: 1038736180644
- config_name: wikipedia
features:
- name: text
dtype: string
splits:
- name: train
num_bytes: 19645170178.22369
num_examples: 6452211
- name: test
num_bytes: 19665840.77630859
num_examples: 6459
download_size: 11644655073
dataset_size: 19664836019.0
---

# JAT Dataset

## Dataset Description

The Jack of All Trades (JAT) dataset combines a wide range of individual datasets. It includes expert demonstrations by expert RL agents, image and caption pairs, textual data and more. The JAT dataset is part of the JAT project, which aims to build a multimodal generalist agent.

**Paper**: https://huggingface.co/papers/2402.09844

### Usage

```python
>>> from datasets import load_dataset
>>> dataset = load_dataset("jat-project/jat-dataset", "metaworld-assembly")
>>> first_episode = dataset["train"][0]
>>> first_episode.keys()
dict_keys(['continuous_observations', 'continuous_actions', 'rewards'])
>>> len(first_episode["rewards"])
500
>>> first_episode["continuous_actions"][0]
[6.459120273590088, 2.2422609329223633, -5.914587020874023, -19.799840927124023]
```

## Dataset Structure

### Data Instances

<details>
<summary>Click to expand the score information for each task</summary>

The following table presents a comparative analysis of scores across various domains and tasks. The scores highlight the performance difference between a random agent and the episodes recorded in our dataset.

| Task | Random Agent Score | Dataset Episode Score |
| ----------------------------------- | :-----------------: | :-------------------: |
| **Atari** | | |
| atari-alien | 205.50 ± 111.97 | 16912.50 ± 7087.42 |
| atari-amidar | 2.38 ± 2.50 | 2164.71 ± 1229.47 |
| atari-assault | 262.50 ± 89.61 | 15699.12 ± 9572.12 |
| atari-asterix | 213.50 ± 110.87 | 3699.62 ± 2421.30 |
| atari-asteroids | 856.40 ± 434.32 | 177011.05 ± 35334.20 |
| atari-atlantis | 17764.00 ± 6662.43 | 320679.59 ± 418247.37 |
| atari-bankheist | 13.40 ± 11.07 | 1322.43 ± 60.84 |
| atari-battlezone | 2170.00 ± 2121.58 | 295592.59 ± 161960.96 |
| atari-beamrider | 357.28 ± 143.97 | 29589.35 ± 16132.96 |
| atari-berzerk | 160.10 ± 118.87 | 57085.26 ± 13104.53 |
| atari-bowling | 23.81 ± 6.07 | 20.40 ± 7.29 |
| atari-boxing | 0.52 ± 4.37 | 97.97 ± 3.77 |
| atari-breakout | 1.24 ± 1.30 | 702.97 ± 203.62 |
| atari-centipede | 2150.06 ± 1113.28 | 11624.29 ± 4918.34 |
| atari-choppercommand | 875.00 ± 416.98 | 90990.62 ± 270876.93 |
| atari-crazyclimber | 7376.00 ± 2253.09 | 179296.94 ± 39862.06 |
| atari-defender | 3417.50 ± 1443.41 | 351958.33 ± 40466.82 |
| atari-demonattack | 165.55 ± 92.93 | 92195.25 ± 26174.79 |
| atari-doubledunk | -18.54 ± 3.07 | 20.94 ± 3.65 |
| atari-enduro | 0.00 ± 0.00 | 2292.22 ± 147.54 |
| atari-fishingderby | -93.90 ± 3.51 | 7.18 ± 25.06 |
| atari-freeway | 0.01 ± 0.10 | 33.88 ± 0.35 |
| atari-frostbite | 67.60 ± 37.61 | 13196.12 ± 4341.00 |
| atari-gopher | 319.40 ± 228.24 | 81676.15 ± 46329.48 |
| atari-gravitar | 188.50 ± 203.33 | 3986.57 ± 1729.05 |
| atari-hero | 475.25 ± 894.95 | 44677.35 ± 1754.42 |
| atari-icehockey | -9.83 ± 3.24 | 25.17 ± 5.79 |
| atari-jamesbond | 28.50 ± 45.42 | 27786.89 ± 33819.20 |
| atari-kangaroo | 52.00 ± 108.15 | 574.05 ± 636.94 |
| atari-krull | 1754.00 ± 583.56 | 11439.83 ± 1218.34 |
| atari-kungfumaster | 390.00 ± 359.03 | 32392.81 ± 10006.55 |
| atari-montezumarevenge | 0.00 ± 0.00 | 393.53 ± 50.45 |
| atari-mspacman | 246.40 ± 121.22 | 6896.08 ± 2031.99 |
| atari-namethisgame | 2447.40 ± 888.97 | 22991.18 ± 2473.15 |
| atari-phoenix | 776.80 ± 635.86 | 424583.16 ± 97649.17 |
| atari-pitfall | -259.75 ± 384.26 | -1.45 ± 4.50 |
| atari-pong | -20.22 ± 0.95 | 20.99 ± 0.18 |
| atari-privateeye | 41.65 ± 191.83 | 100.00 ± 0.00 |
| atari-qbert | 164.25 ± 151.79 | 42971.37 ± 85070.72 |
| atari-riverraid | 1474.40 ± 314.59 | 14800.94 ± 7924.56 |
| atari-roadrunner | 11.00 ± 42.18 | 77942.80 ± 6088.62 |
| atari-robotank | 1.87 ± 1.59 | 80.51 ± 13.28 |
| atari-seaquest | 73.20 ± 57.91 | 2597.34 ± 386.09 |
| atari-skiing | -16299.52 ± 1850.70 | -10738.06 ± 111.13 |
| atari-solaris | 2360.40 ± 1852.03 | 1353.68 ± 516.96 |
| atari-spaceinvaders | 137.20 ± 95.82 | 29425.29 ± 23623.89 |
| atari-stargunner | 652.00 ± 312.24 | 360588.57 ± 49207.71 |
| atari-surround | -9.99 ± 0.10 | 9.39 ± 0.85 |
| atari-tennis | -23.95 ± 0.22 | 11.11 ± 7.57 |
| atari-timepilot | 3396.00 ± 2128.85 | 69583.33 ± 29838.67 |
| atari-tutankham | 12.73 ± 17.40 | 291.16 ± 30.37 |
| atari-upndown | 358.90 ± 380.11 | 429418.33 ± 7187.43 |
| atari-venture | 0.00 ± 0.00 | 0.00 ± 0.00 |
| atari-videopinball | 23917.17 ± 19449.59 | 441507.92 ± 283264.62 |
| atari-wizardofwor | 620.00 ± 837.85 | 49333.33 ± 16157.08 |
| atari-yarsrevenge | 3503.91 ± 906.14 | 270262.86 ± 161815.96 |
| atari-zaxxon | 21.00 ± 102.27 | 73097.22 ± 14825.77 |
| **BabyAI** | | |
| babyai-action-obj-door | 0.37 ± 0.39 | 0.99 ± 0.01 |
| babyai-blocked-unlock-pickup | 0.00 ± 0.02 | 0.95 ± 0.01 |
| babyai-boss-level | 0.06 ± 0.21 | 0.94 ± 0.05 |
| babyai-boss-level-no-unlock | 0.06 ± 0.19 | 0.94 ± 0.05 |
| babyai-find-obj-s5 | 0.08 ± 0.23 | 0.95 ± 0.04 |
| babyai-go-to | 0.13 ± 0.29 | 0.92 ± 0.07 |
| babyai-go-to-door | 0.45 ± 0.38 | 0.99 ± 0.00 |
| babyai-go-to-imp-unlock | 0.08 ± 0.23 | 0.83 ± 0.13 |
| babyai-go-to-local | 0.16 ± 0.30 | 0.93 ± 0.04 |
| babyai-go-to-obj | 0.13 ± 0.27 | 0.93 ± 0.03 |
| babyai-go-to-obj-door | 0.53 ± 0.39 | 0.99 ± 0.01 |
| babyai-go-to-red-ball | 0.17 ± 0.30 | 0.93 ± 0.04 |
| babyai-go-to-red-ball-grey | 0.12 ± 0.27 | 0.92 ± 0.05 |
| babyai-go-to-red-ball-no-dists | 0.14 ± 0.28 | 0.93 ± 0.03 |
| babyai-go-to-red-blue-ball | 0.12 ± 0.27 | 0.92 ± 0.05 |
| babyai-go-to-seq | 0.08 ± 0.23 | 0.94 ± 0.05 |
| babyai-key-corridor | 0.00 ± 0.00 | 0.91 ± 0.01 |
| babyai-mini-boss-level | 0.07 ± 0.21 | 0.89 ± 0.10 |
| babyai-move-two-across-s8n9 | 0.00 ± 0.00 | 0.96 ± 0.01 |
| babyai-one-room-s8 | 0.08 ± 0.21 | 0.92 ± 0.03 |
| babyai-open | 0.10 ± 0.24 | 0.95 ± 0.05 |
| babyai-open-door | 0.23 ± 0.34 | 0.99 ± 0.00 |
| babyai-open-doors-order-n4 | 0.16 ± 0.30 | 0.99 ± 0.01 |
| babyai-open-red-door | 0.08 ± 0.21 | 0.92 ± 0.03 |
| babyai-open-two-doors | 0.08 ± 0.20 | 0.98 ± 0.00 |
| babyai-pickup | 0.08 ± 0.22 | 0.92 ± 0.07 |
| babyai-pickup-above | 0.02 ± 0.09 | 0.91 ± 0.07 |
| babyai-pickup-dist | 0.10 ± 0.24 | 0.86 ± 0.21 |
| babyai-pickup-loc | 0.08 ± 0.23 | 0.91 ± 0.04 |
| babyai-put-next | 0.00 ± 0.03 | 0.96 ± 0.01 |
| babyai-put-next-local | 0.00 ± 0.05 | 0.92 ± 0.03 |
| babyai-synth | 0.11 ± 0.26 | 0.93 ± 0.06 |
| babyai-synth-loc | 0.13 ± 0.29 | 0.94 ± 0.06 |
| babyai-synth-seq | 0.07 ± 0.20 | 0.95 ± 0.04 |
| babyai-unblock-pickup | 0.08 ± 0.22 | 0.91 ± 0.08 |
| babyai-unlock | 0.03 ± 0.15 | 0.87 ± 0.10 |
| babyai-unlock-local | 0.01 ± 0.09 | 0.98 ± 0.01 |
| babyai-unlock-pickup | 0.00 ± 0.00 | 0.75 ± 0.04 |
| babyai-unlock-to-unlock | 0.00 ± 0.00 | 0.96 ± 0.00 |
| **Meta-World** | | |
| metaworld-assembly | 45.30 ± 4.13 | 245.99 ± 3.50 |
| metaworld-basketball | 2.81 ± 1.24 | 627.99 ± 1.98 |
| metaworld-bin-picking | 1.89 ± 0.45 | 425.58 ± 101.86 |
| metaworld-box-close | 76.39 ± 17.91 | 512.49 ± 107.81 |
| metaworld-button-press | 31.73 ± 5.20 | 643.10 ± 12.85 |
| metaworld-button-press-topdown | 28.97 ± 10.37 | 490.18 ± 27.21 |
| metaworld-button-press-topdown-wall | 29.04 ± 10.52 | 497.19 ± 31.37 |
| metaworld-button-press-wall | 8.98 ± 3.99 | 675.41 ± 15.04 |
| metaworld-coffee-button | 31.72 ± 6.36 | 731.08 ± 29.34 |
| metaworld-coffee-pull | 4.09 ± 0.38 | 259.86 ± 88.48 |
| metaworld-coffee-push | 4.17 ± 0.76 | 496.78 ± 118.20 |
| metaworld-dial-turn | 29.64 ± 16.67 | 793.56 ± 80.06 |
| metaworld-disassemble | 40.31 ± 7.53 | 42.83 ± 6.30 |
| metaworld-door-close | 5.30 ± 1.33 | 529.75 ± 27.24 |
| metaworld-door-lock | 112.35 ± 28.63 | 811.52 ± 34.07 |
| metaworld-door-open | 56.37 ± 11.23 | 581.94 ± 19.67 |
| metaworld-door-unlock | 94.17 ± 15.56 | 802.88 ± 17.05 |
| metaworld-drawer-close | 116.73 ± 253.11 | 867.92 ± 4.48 |
| metaworld-drawer-open | 126.85 ± 25.22 | 492.99 ± 2.52 |
| metaworld-faucet-close | 253.12 ± 22.94 | 753.92 ± 13.42 |
| metaworld-faucet-open | 244.10 ± 23.25 | 705.76 ± 7.15 |
| metaworld-hammer | 95.33 ± 9.02 | 693.17 ± 34.62 |
| metaworld-hand-insert | 2.75 ± 3.53 | 740.53 ± 36.69 |
| metaworld-handle-press | 80.41 ± 110.19 | 855.91 ± 72.75 |
| metaworld-handle-press-side | 57.00 ± 39.47 | 861.12 ± 20.01 |
| metaworld-handle-pull | 10.34 ± 13.54 | 669.35 ± 24.81 |
| metaworld-handle-pull-side | 2.13 ± 2.76 | 384.65 ± 102.89 |
| metaworld-lever-pull | 60.31 ± 15.77 | 612.04 ± 38.85 |
| metaworld-peg-insert-side | 1.71 ± 0.36 | 315.23 ± 140.07 |
| metaworld-peg-unplug-side | 4.75 ± 2.83 | 456.12 ± 81.65 |
| metaworld-pick-out-of-hole | 1.51 ± 0.24 | 219.61 ± 88.85 |
| metaworld-pick-place | 1.61 ± 0.99 | 419.10 ± 98.19 |
| metaworld-pick-place-wall | 0.00 ± 0.01 | 450.57 ± 64.10 |
| metaworld-plate-slide | 74.64 ± 13.84 | 527.01 ± 155.34 |
| metaworld-plate-slide-back | 33.47 ± 11.22 | 718.22 ± 87.41 |
| metaworld-plate-slide-back-side | 34.34 ± 11.53 | 729.61 ± 69.15 |
| metaworld-plate-slide-side | 22.61 ± 17.36 | 662.81 ± 102.81 |
| metaworld-push | 5.51 ± 2.43 | 750.57 ± 43.98 |
| metaworld-push-back | 1.21 ± 0.16 | 85.05 ± 107.12 |
| metaworld-push-wall | 6.13 ± 3.17 | 748.87 ± 10.62 |
| metaworld-reach | 149.67 ± 44.70 | 681.37 ± 133.68 |
| metaworld-reach-wall | 143.26 ± 36.56 | 746.12 ± 104.19 |
| metaworld-shelf-place | 0.00 ± 0.01 | 241.34 ± 24.60 |
| metaworld-soccer | 5.66 ± 4.61 | 375.15 ± 140.24 |
| metaworld-stick-pull | 2.64 ± 1.41 | 523.55 ± 18.94 |
| metaworld-stick-push | 2.81 ± 1.04 | 627.95 ± 10.20 |
| metaworld-sweep | 11.23 ± 7.28 | 494.85 ± 43.29 |
| metaworld-sweep-into | 12.55 ± 10.72 | 799.21 ± 19.07 |
| metaworld-window-close | 57.46 ± 7.11 | 591.30 ± 38.63 |
| metaworld-window-open | 43.36 ± 2.09 | 590.82 ± 57.08 |
| **MuJoCo** | | |
| mujoco-ant | -59.95 ± 99.62 | 5846.42 ± 942.55 |
| mujoco-doublependulum | 57.46 ± 17.54 | 9338.69 ± 352.61 |
| mujoco-halfcheetah | -284.97 ± 79.83 | 7437.77 ± 173.30 |
| mujoco-hopper | 18.38 ± 17.09 | 1858.73 ± 534.07 |
| mujoco-humanoid | 122.02 ± 35.28 | 6281.02 ± 1795.84 |
| mujoco-pendulum | 6.07 ± 3.47 | 475.40 ± 178.96 |
| mujoco-pusher | -149.69 ± 7.41 | -25.21 ± 6.66 |
| mujoco-reacher | -43.00 ± 3.91 | -5.68 ± 2.53 |
| mujoco-standup | 33135.75 ± 2481.89 | 273574.16 ± 85253.26 |
| mujoco-swimmer | 0.80 ± 10.71 | 92.18 ± 4.44 |
| mujoco-walker | 2.68 ± 6.06 | 4631.22 ± 1059.01 |
</details>

### Data Fields

- `text`: a `string` feature
- `images`: a `image` feature
- `image_observations` : a `Sequence(image)` feature
- `text_observations` : a `Sequence(string)` feature
- `discrete_observations`: a `Sequence(Sequence(int64))` feature
- `continuous_observations`: a `Sequence(Sequence(float32))` feature
- `continuous_actions`: a `Sequence(Sequence(float32))` feature
- `discrete_actions`: a `Sequence(int64)` feature
- `rewards`: a `Sequence(float32)` feature

### Data Splits

- `train`: `` examples
- `test`: `` examples

## Dataset Creation

This section describes how our dataset was created. We specifically detail how data for each domain and task were generated. The generation scripts are available in the [JAT repository](https://github.com/huggingface/jat). For RL tasks, we trained one agent per task using the [Sample Factory](https://www.samplefactory.dev). Then we used the trained agent to generate episodes.

### Atari

We used the 57 [ALE/Atari](https://github.com/Farama-Foundation/Arcade-Learning-Environment) games as our environment, configuring the following parameters for our experiments. We rendered the images in grayscale with an 84x84 pixel resolution. The agent interacted with the environment every 4 frames. Sticky actions were not used, and the raw reward (no clipping) was reported. Episodes were stored as complete, i.e. with no termination on life loss.

### BabyAI

We used BabyAI's implementation from [Minigrid](https://github.com/Farama-Foundation/Minigrid).
We reused the [bot agent](https://github.com/mila-iqia/babyai) provided with BabyAI's paper and adapted it to the new Minigrid API.
Using the bot, we generated 1.000.000 interractions for each of the 39 tasks of [Minigrid's BabyAI](https://minigrid.farama.org/environments/babyai/) and stored for each step:

- the mission: str
- the concatenation of the symbolic observation flattened and the direction: Array of integers of size (147,)
- the action: integer
- the reward: float

### Conceptual Captions

The [Conceptual Captions](https://github.com/google-research-datasets/conceptual-captions/tree/master) dataset, offered by Google LLC, comprises pairs of image links and their corresponding captions. Each image has been downloaded and, when required, resized to ensure the maximum dimension does not exceed 352 pixels.

### Meta-World

We used the 50 tasks from [Meta-World v2](https://github.com/Farama-Foundation/Metaworld). We constrained the episode to a duration of 100 timesteps, which is always sufficient to solve the task.

### MuJoCo

We used the 11 environments of Gymnasium MuJoCo.

### OK-VQA

The [OK-VQA](https://okvqa.allenai.org/index.html) dataset released by Kenneth Marino, Mohammad Rastegari, Ali Farhadi, Roozbeh Mottaghi was used.
The data were formatted to match Hugging Face dataset's requirements and images were resized such that the largest dimension is at most 352.

### OSCAR

We modified the "unshuffled_deduplicated_en" split of [OSCAR 2019](https://huggingface.co/datasets/oscar) dataset, initially put together by Pedro J. Ortiz, Benoît Sagot, and Laurent Romary and licensed under [CC BY 4.0](https://oscar-project.github.io/documentation/versions/oscar-2019/#license).
We cleaned and deduplicated the dataset using [the methods](https://github.com/bigscience-workshop/data-preparation/tree/main/preprocessing/training/01b_oscar_cleaning_and_filtering) and parameters used for the [ROOTS dataset](https://arxiv.org/abs/2303.03915) (Lurençon et al., 2023).

The dataset was splitted into 30 even shards each cleaned and deduplicated independently before being concatenated again.

### Wikipedia

We used the english version of the [Wikipedia dataset](https://huggingface.co/datasets/wikipedia).

## Considerations for Using the Data
### Known Issues

- Some BabyAI tasks are missing due to incompatibility with the training bot:
- `babyai-key-in-box`
- `babyai-go-to-imp-unlock`
- `babyai-unlock-to-unlock`
- `babyai-unlock`
- For some atari tasks, the episode is too long, causing an `OverflowError` when loading the dataset:
- `atari-enduro`
- For some tasks, although the score can be higher than the random agent, we can't consider the task as solved:
- `atari-bowling`
- `atari-privateeye`
- `atari-solaris`
- `atari-venture`
- `metaworld-bin-picking`
- `metaworld-disassemble`
- `metaworld-peg-insert-side`
- `metaworld-plate-slide`
- `metaworld-push-back`

### Future Developments

We plan to expand the dataset to include the following additional domains:

- [ ] DM Lab
- [ ] Sokoban
- [ ] Procgen
- [ ] DM Control Suite (w and w/o pixels)

## Additional Information

### Licensing Information

This dataset is release under the Apache 2.0 license.

### Citation Information

```bibtex
@article{gallouedec2024jack,
title = {{Jack of All Trades, Master of Some: a Multi-Purpose Transformer Agent}},
author = {Gallouédec, Quentin and Beeching, Edward and Romac, Clément and Dellandréa, Emmanuel},
journal = {arXiv preprint arXiv:2402.09844},
year = {2024},
url = {https://arxiv.org/abs/2402.09844}
}
```

## Acknowledgment

We would like to extend our sincere gratitude to:

- [Shengyi Costa Huang](https://huggingface.co/vwxyzjn) for his invaluable assistance with the pretrained models used in this research

原始信息汇总

数据集概述

基本信息

• 数据集名称: JAT-dataset
• 许可证: Apache 2.0
• 标注创建者: 发现和机器生成
• 源数据集: conceptual-captions, ok-vqa, oscar
• 任务类别: 强化学习, 文本生成, 问答

配置详情

数据集包含多个配置，每个配置对应不同的游戏或任务，并分为训练集和测试集。以下是部分配置的详细信息：

Atari 游戏配置

• atari-alien
  • 训练集: atari-alien/train-*
  • 测试集: atari-alien/test-*
• atari-amidar
  • 训练集: atari-amidar/train-*
  • 测试集: atari-amidar/test-*
• atari-assault
  • 训练集: atari-assault/train-*
  • 测试集: atari-assault/test-*
• atari-asterix
  • 训练集: atari-asterix/train-*
  • 测试集: atari-asterix/test-*
• atari-asteroids
  • 训练集: atari-asteroids/train-*
  • 测试集: atari-asteroids/test-*
• atari-atlantis
  • 训练集: atari-atlantis/train-*
  • 测试集: atari-atlantis/test-*
• atari-bankheist
  • 训练集: atari-bankheist/train-*
  • 测试集: atari-bankheist/test-*
• atari-battlezone
  • 训练集: atari-battlezone/train-*
  • 测试集: atari-battlezone/test-*
• atari-beamrider
  • 训练集: atari-beamrider/train-*
  • 测试集: atari-beamrider/test-*
• atari-berzerk
  • 训练集: atari-berzerk/train-*
  • 测试集: atari-berzerk/test-*
• atari-bowling
  • 训练集: atari-bowling/train-*
  • 测试集: atari-bowling/test-*
• atari-boxing
  • 训练集: atari-boxing/train-*
  • 测试集: atari-boxing/test-*
• atari-breakout
  • 训练集: atari-breakout/train-*
  • 测试集: atari-breakout/test-*
• atari-centipede
  • 训练集: atari-centipede/train-*
  • 测试集: atari-centipede/test-*
• atari-choppercommand
  • 训练集: atari-choppercommand/train-*
  • 测试集: atari-choppercommand/test-*
• atari-crazyclimber
  • 训练集: atari-crazyclimber/train-*
  • 测试集: atari-crazyclimber/test-*
• atari-defender
  • 训练集: atari-defender/train-*
  • 测试集: atari-defender/test-*
• atari-demonattack
  • 训练集: atari-demonattack/train-*
  • 测试集: atari-demonattack/test-*
• atari-doubledunk
  • 训练集: atari-doubledunk/train-*
  • 测试集: atari-doubledunk/test-*
• atari-enduro
  • 训练集: atari-enduro/train-*
  • 测试集: atari-enduro/test-*
• atari-fishingderby
  • 训练集: atari-fishingderby/train-*
  • 测试集: atari-fishingderby/test-*
• atari-freeway
  • 训练集: atari-freeway/train-*
  • 测试集: atari-freeway/test-*
• atari-frostbite
  • 训练集: atari-frostbite/train-*
  • 测试集: atari-frostbite/test-*
• atari-gopher
  • 训练集: atari-gopher/train-*
  • 测试集: atari-gopher/test-*
• atari-gravitar
  • 训练集: atari-gravitar/train-*
  • 测试集: atari-gravitar/test-*
• atari-hero
  • 训练集: atari-hero/train-*
  • 测试集: atari-hero/test-*
• atari-icehockey
  • 训练集: atari-icehockey/train-*
  • 测试集: atari-icehockey/test-*
• atari-jamesbond
  • 训练集: atari-jamesbond/train-*
  • 测试集: atari-jamesbond/test-*
• atari-kangaroo
  • 训练集: atari-kangaroo/train-*
  • 测试集: atari-kangaroo/test-*
• atari-krull
  • 训练集: atari-krull/train-*
  • 测试集: atari-krull/test-*
• atari-kungfumaster
  • 训练集: atari-kungfumaster/train-*
  • 测试集: atari-kungfumaster/test-*
• atari-montezumarevenge
  • 训练集: atari-montezumarevenge/train-*
  • 测试集: atari-montezumarevenge/test-*
• atari-mspacman
  • 训练集: atari-mspacman/train-*
  • 测试集: atari-mspacman/test-*
• atari-namethisgame
  • 训练集: atari-namethisgame/train-*
  • 测试集: atari-namethisgame/test-*
• atari-phoenix
  • 训练集: atari-phoenix/train-*
  • 测试集: atari-phoenix/test-*
• atari-pitfall
  • 训练集: atari-pitfall/train-*
  • 测试集: atari-pitfall/test-*
• atari-pong
  • 训练集: atari-pong/train-*
  • 测试集: atari-pong/test-*
• atari-privateeye
  • 训练集: atari-privateeye/train-*
  • 测试集: atari-privateeye/test-*
• atari-qbert
  • 训练集: atari-qbert/train-*
  • 测试集: atari-qbert/test-*
• atari-riverraid
  • 训练集: atari-riverraid/train-*
  • 测试集: atari-riverraid/test-*
• atari-roadrunner
  • 训练集: atari-roadrunner/train-*
  • 测试集: atari-roadrunner/test-*
• atari-robotank
  • 训练集: atari-robotank/train-*
  • 测试集: atari-robotank/test-*
• atari-seaquest
  • 训练集: atari-seaquest/train-*
  • 测试集: atari-seaquest/test-*
• atari-skiing
  • 训练集: atari-skiing/train-*
  • 测试集: atari-skiing/test-*
• atari-solaris
  • 训练集: atari-solaris/train-*
  • 测试集: atari-solaris/test-*
• atari-spaceinvaders
  • 训练集: atari-spaceinvaders/train-*
  • 测试集: atari-spaceinvaders/test-*
• atari-stargunner
  • 训练集: atari-stargunner/train-*
  • 测试集: atari-stargunner/test-*
• atari-surround
  • 训练集: atari-surround/train-*
  • 测试集: atari-surround/test-*
• atari-tennis
  • 训练集: atari-tennis/train-*
  • 测试集: atari-tennis/test-*
• atari-timepilot
  • 训练集: atari-timepilot/train-*
  • 测试集: atari-timepilot/test-*
• atari-tutankham
  • 训练集: atari-tutankham/train-*
  • 测试集: atari-tutankham/test-*
• atari-upndown
  • 训练集: atari-upndown/train-*
  • 测试集: atari-upndown/test-*
• atari-venture
  • 训练集: atari-venture/train-*
  • 测试集: atari-venture/test-*
• atari-videopinball
  • 训练集: atari-videopinball/train-*
  • 测试集: atari-videopinball/test-*
• atari-wizardofwor
  • 训练集: atari-wizardofwor/train-*
  • 测试集: atari-wizardofwor/test-*
• atari-yarsrevenge
  • 训练集: atari-yarsrevenge/train-*
  • 测试集: atari-yarsrevenge/test-*
• atari-zaxxon
  • 训练集: atari-zaxxon/train-*
  • 测试集: atari-zaxxon/test-*

BabyAI 配置

• babyai-action-obj-door
  • 训练集: babyai-action-obj-door/train-*
  • 测试集: babyai-action-obj-door/test-*
• babyai-blocked-unlock-pickup
  • 训练集: babyai-blocked-unlock-pickup/train-*
  • 测试集: babyai-blocked-unlock-pickup/test-*
• babyai-boss-level
  • 训练集: babyai-boss-level/train-*
  • 测试集: babyai-boss-level/test-*
• babyai-boss-level-no-unlock
  • 训练集: babyai-boss-level-no-unlock/train-*
  • 测试集: babyai-boss-level-no-unlock/test-*
• babyai-find-obj-s5
  • 训练集: babyai-find-obj-s5/train-*
  • 测试集: babyai-find-obj-s5/test-*
• babyai-go-to
  • 训练集: babyai-go-to/train-*
  • 测试集: babyai-go-to/test-*
• babyai-go-to-door
  • 训练集: babyai-go-to-door/train-*
  • 测试集: babyai-go-to-door/test-*
• babyai-go-to-imp-unlock
  • 训练集: babyai-go-to-imp-unlock/train-*
  • 测试集: babyai-go-to-imp-unlock/test-*
• babyai-go-to-local
  • 训练集: babyai-go-to-local/train-*
  • 测试集: babyai-go-to-local/test-*
• babyai-go-to-obj
  • 训练集: babyai-go-to-obj/train-*
  • 测试集: babyai-go-to-obj/test-*
• babyai-go-to-obj-door
  • 训练集: babyai-go-to-obj-door/train-*
  • 测试集: babyai-go-to-obj-door/test-*
• babyai-go-to-red-ball
  • 训练集: babyai-go-to-red-ball/train-*
  • 测试集: babyai-go-to-red-ball/test-*
• babyai-go-to-red-ball-grey
  • 训练集: babyai-go-to-red-ball-grey/train-*
  • 测试集: babyai-go-to-red-ball-grey/test-*
• babyai-go-to-red-ball-no-dists
  • 训练集: babyai-go-to-red-ball-no-dists/train-*
  • 测试集: babyai-go-to-red-ball-no-dists/test-*
• babyai-go-to-red-blue-ball
  • 训练集: babyai-go-to-red-blue-ball/train-*
  • 测试集: babyai-go-to-red-blue-ball/test-*
• babyai-go-to-seq
  • 训练集: babyai-go-to-seq/train-*
  • 测试集: babyai-go-to-seq/test-*
• babyai-key-corridor
  • 训练集: babyai-key-corridor/train-*
  • 测试集: babyai-key-corridor/test-*
• babyai-mini-boss-level
  • 训练集: babyai-mini-boss-level/train-*
  • 测试集: babyai-mini-boss-level/test-*
• babyai-move-two-across-s8n9
  • 训练集: babyai-move-two-across-s8n9/train-*
  • 测试集: babyai-move-two-across-s8n9/test-*
• babyai-one-room-s8
  • 训练集: babyai-one-room-s8/train-*
  • 测试集: babyai-one-room-s8/test-*
• babyai-open
  • 训练集: babyai-open/train-*
  • 测试集: babyai-open/test-*
• babyai-open-door
  • 训练集: babyai-open-door/train-*
  • 测试集: babyai-open-door/test-*
• babyai-open-doors-order-n4
  • 训练集: babyai-open-doors-order-n4/train-*
  • 测试集: babyai-open-doors-order-n4/test-*
• babyai-open-red-door
  • 训练集: babyai-open-red-door/train-*
  • 测试集: babyai-open-red-door/test-*
• babyai-open-two-doors
  • 训练集: babyai-open-two-doors/train-*
  • 测试集: babyai-open-two-doors/test-*
• babyai-pickup
  • 训练集: babyai-pickup/train-*
  • 测试集: babyai-pickup/test-*
• babyai-pickup-above
  • 训练集: babyai-pickup-above/train-*
  • 测试集: babyai-pickup-above/test-*
• babyai-pickup-dist
  • 训练集: babyai-pickup-dist/train-*
  • 测试集: babyai-pickup-dist/test-*
• babyai-pickup-loc
  • 训练集: babyai-pickup-loc/train-*
  • 测试集: babyai-pickup-loc/test-*
• babyai-put-next
  • 训练集: babyai-put-next/train-*
  • 测试集: `babyai-put