name stringlengths 3 18 | split stringclasses 1
value | game stringlengths 3 18 | source stringclasses 1
value | prompt stringlengths 4.65k 16.6k | system_prompt stringclasses 1
value | episodes int64 1k 1k | task_id int64 0 321 | family stringclasses 2
values | upstream_source_url stringlengths 120 135 | upstream_revision stringclasses 1
value | native_source_url stringlengths 111 138 ⌀ |
|---|---|---|---|---|---|---|---|---|---|---|---|
marusansi | train | marusansi | synth | Write from scratch the weights of a policy network that plays Marusansi as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 0 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/marusansi.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameMarusansi.c |
thunder | train | thunder | synth | Write from scratch the weights of a policy network that plays Thunder as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 1 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/thunder.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-11-games/blob/6fea4eefe2deaa42706051d96a04ec9e36104696/docs/thunder/main.js |
up_shot | train | up_shot | synth | Write from scratch the weights of a policy network that plays Up Shot as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 2 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/up_shot.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/upshot/main.js |
ladder_drop | train | ladder_drop | synth | Write from scratch the weights of a policy network that plays Ladder Drop as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edit... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 3 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/ladder_drop.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameLadderDrop.c |
scrambird | train | scrambird | synth | Write from scratch the weights of a policy network that plays Scrambird as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 4 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/scrambird.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/scrambird/main.js |
battle_zone | train | battle_zone | synth | Write from scratch the weights of a policy network that plays Battle Zone as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edit... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 5 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/battle_zone.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
b_blast | train | b_blast | synth | Write from scratch the weights of a policy network that plays B Blast as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 6 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/b_blast.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameBBlast.c |
zaxxon | train | zaxxon | synth | Write from scratch the weights of a policy network that plays Zaxxon as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 7 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/zaxxon.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
cate_p | train | cate_p | synth | Write from scratch the weights of a policy network that plays Cate P as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 8 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/cate_p.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/catep/main.js |
tt_fence | train | tt_fence | synth | Write from scratch the weights of a policy network that plays Tt Fence as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 9 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/tt_fence.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/ttfence/main.js |
cross_line | train | cross_line | synth | Write from scratch the weights of a policy network that plays Cross Line as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 10 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/cross_line.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/crossline/main.js |
geocent | train | geocent | synth | Write from scratch the weights of a policy network that plays Geocent as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 11 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/geocent.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/geocent/main.js |
skiing | train | skiing | synth | Write from scratch the weights of a policy network that plays Skiing as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 12 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/skiing.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
monkey_t | train | monkey_t | synth | Write from scratch the weights of a policy network that plays Monkey T as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 13 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/monkey_t.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameMonkeyT.c |
zip_fall | train | zip_fall | synth | Write from scratch the weights of a policy network that plays Zip Fall as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 14 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/zip_fall.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameZipFall.c |
hoard_spout | train | hoard_spout | synth | Write from scratch the weights of a policy network that plays Hoard Spout as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edit... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 15 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/hoard_spout.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameHoardSpout.c |
twin_jumpers | train | twin_jumpers | synth | Write from scratch the weights of a policy network that plays Twin Jumpers as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edi... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 16 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/twin_jumpers.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameTwinJumpers.c |
block_stacker | train | block_stacker | synth | Write from scratch the weights of a policy network that plays Block Stacker as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `ed... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 17 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/block_stacker.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameBlockStacker.c |
phoenix | train | phoenix | synth | Write from scratch the weights of a policy network that plays Phoenix as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 18 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/phoenix.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
graveler | train | graveler | synth | Write from scratch the weights of a policy network that plays Graveler as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 19 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/graveler.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/graveler/main.js |
c_nodes | train | c_nodes | synth | Write from scratch the weights of a policy network that plays C Nodes as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 20 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/c_nodes.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/cnodes/main.js |
teeter | train | teeter | synth | Write from scratch the weights of a policy network that plays Teeter as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 21 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/teeter.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/teeter/main.js |
b_math | train | b_math | synth | Write from scratch the weights of a policy network that plays B Math as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 22 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/b_math.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameBMath.c |
timber_test | train | timber_test | synth | Write from scratch the weights of a policy network that plays Timber Test as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edit... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 23 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/timber_test.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameTimberTest.c |
rb_drive | train | rb_drive | synth | Write from scratch the weights of a policy network that plays Rb Drive as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 24 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/rb_drive.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameRbDrive.c |
montezumas_revenge | train | montezumas_revenge | synth | Write from scratch the weights of a policy network that plays Montezumas Revenge as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the fina... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 25 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/montezumas_revenge.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
kung_fu_master | train | kung_fu_master | synth | Write from scratch the weights of a policy network that plays Kung Fu Master as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `e... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 26 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/kung_fu_master.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
blaze_thrust | train | blaze_thrust | synth | Write from scratch the weights of a policy network that plays Blaze Thrust as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edi... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 27 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/blaze_thrust.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameBlazeThrust.c |
catching_wheel | train | catching_wheel | synth | Write from scratch the weights of a policy network that plays Catching Wheel as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `e... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 28 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/catching_wheel.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameCatchingWheel.c |
d_missile | train | d_missile | synth | Write from scratch the weights of a policy network that plays D Missile as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 29 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/d_missile.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/dmissile/main.js |
p_fit | train | p_fit | synth | Write from scratch the weights of a policy network that plays P Fit as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` cod... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 30 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/p_fit.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/pfit/main.js |
doshin | train | doshin | synth | Write from scratch the weights of a policy network that plays Doshin as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 31 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/doshin.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/doshin/main.js |
holes | train | holes | synth | Write from scratch the weights of a policy network that plays Holes as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` cod... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 32 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/holes.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/holes/main.js |
turbulent | train | turbulent | synth | Write from scratch the weights of a policy network that plays Turbulent as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 33 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/turbulent.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/turbulent/main.js |
reflector | train | reflector | synth | Write from scratch the weights of a policy network that plays Reflector as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 34 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/reflector.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/reflector/main.js |
grenadier | train | grenadier | synth | Write from scratch the weights of a policy network that plays Grenadier as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 35 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/grenadier.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/grenadier/main.js |
assault | train | assault | synth | Write from scratch the weights of a policy network that plays Assault as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 36 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/assault.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
kelp_climber | train | kelp_climber | synth | Write from scratch the weights of a policy network that plays Kelp Climber as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edi... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 37 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/kelp_climber.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameKelpClimber.c |
pileup | train | pileup | synth | Write from scratch the weights of a policy network that plays Pileup as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 38 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/pileup.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/pileup/main.js |
zoom_io | train | zoom_io | synth | Write from scratch the weights of a policy network that plays Zoom Io as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 39 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/zoom_io.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/zoomio/main.js |
balloon | train | balloon | synth | Write from scratch the weights of a policy network that plays Balloon as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 40 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/balloon.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameBalloon.c |
zartan | train | zartan | synth | Write from scratch the weights of a policy network that plays Zartan as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 41 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/zartan.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/zartan/main.js |
foot_laser | train | foot_laser | synth | Write from scratch the weights of a policy network that plays Foot Laser as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 42 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/foot_laser.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameFootLaser.c |
snaky | train | snaky | synth | Write from scratch the weights of a policy network that plays Snaky as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` cod... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 43 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/snaky.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/snaky/main.js |
froooog | train | froooog | synth | Write from scratch the weights of a policy network that plays Froooog as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 44 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/froooog.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameFroooog.c |
trojan_defense | train | trojan_defense | synth | Write from scratch the weights of a policy network that plays Trojan Defense as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `e... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 45 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/trojan_defense.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameTrojanDefense.c |
balance | train | balance | synth | Write from scratch the weights of a policy network that plays Balance as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 46 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/balance.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameBalance.c |
video_pinball | train | video_pinball | synth | Write from scratch the weights of a policy network that plays Video Pinball as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `ed... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 47 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/video_pinball.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
in_tow | train | in_tow | synth | Write from scratch the weights of a policy network that plays In Tow as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 48 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/in_tow.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameInTow.c |
divarr | train | divarr | synth | Write from scratch the weights of a policy network that plays Divarr as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 49 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/divarr.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/divarr/main.js |
hero | train | hero | synth | Write from scratch the weights of a policy network that plays Hero as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` code... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 50 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/hero.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
kite | train | kite | synth | Write from scratch the weights of a policy network that plays Kite as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` code... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 51 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/kite.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/kite/main.js |
inf_range | train | inf_range | synth | Write from scratch the weights of a policy network that plays Inf Range as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 52 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/inf_range.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/infrange/main.js |
counter_b | train | counter_b | synth | Write from scratch the weights of a policy network that plays Counter B as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 53 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/counter_b.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/counterb/main.js |
revolve_a | train | revolve_a | synth | Write from scratch the weights of a policy network that plays Revolve A as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 54 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/revolve_a.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/revolvea/main.js |
square_bar | train | square_bar | synth | Write from scratch the weights of a policy network that plays Square Bar as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 55 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/square_bar.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/squarebar/main.js |
gravity_bubble | train | gravity_bubble | synth | Write from scratch the weights of a policy network that plays Gravity Bubble as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `e... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 56 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/gravity_bubble.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameGravityBubble.c |
ball_tour | train | ball_tour | synth | Write from scratch the weights of a policy network that plays Ball Tour as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 57 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/ball_tour.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameBallTour.c |
feeding_frenzy | train | feeding_frenzy | synth | Write from scratch the weights of a policy network that plays Feeding Frenzy as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `e... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 58 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/feeding_frenzy.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameFeedingFrenzy.c |
kangaroo | train | kangaroo | synth | Write from scratch the weights of a policy network that plays Kangaroo as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 59 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/kangaroo.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
tilted | train | tilted | synth | Write from scratch the weights of a policy network that plays Tilted as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 60 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/tilted.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/tilted/main.js |
ant_lion | train | ant_lion | synth | Write from scratch the weights of a policy network that plays Ant Lion as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 61 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/ant_lion.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameAntLion.c |
tr_beam | train | tr_beam | synth | Write from scratch the weights of a policy network that plays Tr Beam as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 62 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/tr_beam.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/trbeam/main.js |
golfme | train | golfme | synth | Write from scratch the weights of a policy network that plays Golfme as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 63 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/golfme.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/golfme/main.js |
hexmin | train | hexmin | synth | Write from scratch the weights of a policy network that plays Hexmin as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 64 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/hexmin.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameHexmin.c |
catapult | train | catapult | synth | Write from scratch the weights of a policy network that plays Catapult as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 65 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/catapult.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/catapult/main.js |
t_lanes | train | t_lanes | synth | Write from scratch the weights of a policy network that plays T Lanes as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 66 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/t_lanes.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/tlanes/main.js |
invincible_man | train | invincible_man | synth | Write from scratch the weights of a policy network that plays Invincible Man as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `e... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 67 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/invincible_man.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameInvincibleMan.c |
rotating_tunnel | train | rotating_tunnel | synth | Write from scratch the weights of a policy network that plays Rotating Tunnel as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 68 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/rotating_tunnel.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameRotatingTunnel.c |
dark_sort | train | dark_sort | synth | Write from scratch the weights of a policy network that plays Dark Sort as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 69 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/dark_sort.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameDarkSort.c |
survivor | train | survivor | synth | Write from scratch the weights of a policy network that plays Survivor as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 70 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/survivor.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameSurvivor.c |
star_gunner | train | star_gunner | synth | Write from scratch the weights of a policy network that plays Star Gunner as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edit... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 71 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/star_gunner.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
sky_floor | train | sky_floor | synth | Write from scratch the weights of a policy network that plays Sky Floor as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 72 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/sky_floor.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/skyfloor/main.js |
yars_revenge | train | yars_revenge | synth | Write from scratch the weights of a policy network that plays Yars Revenge as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edi... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 73 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/yars_revenge.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
screen | train | screen | synth | Write from scratch the weights of a policy network that plays Screen as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 74 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/screen.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/screen/main.js |
pin_climb | train | pin_climb | synth | Write from scratch the weights of a policy network that plays Pin Climb as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 75 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/pin_climb.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gamePinClimb.c |
bore_comet | train | bore_comet | synth | Write from scratch the weights of a policy network that plays Bore Comet as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 76 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/bore_comet.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameBoreComet.c |
levitation | train | levitation | synth | Write from scratch the weights of a policy network that plays Levitation as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 77 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/levitation.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameLevitation.c |
geoerase | train | geoerase | synth | Write from scratch the weights of a policy network that plays Geoerase as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 78 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/geoerase.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameGeoerase.c |
magnetic_pendulum | train | magnetic_pendulum | synth | Write from scratch the weights of a policy network that plays Magnetic Pendulum as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 79 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/magnetic_pendulum.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameMagneticPendulum.c |
boarding | train | boarding | synth | Write from scratch the weights of a policy network that plays Boarding as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 80 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/boarding.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/boarding/main.js |
pitfall | train | pitfall | synth | Write from scratch the weights of a policy network that plays Pitfall as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 81 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/pitfall.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
fish_grill | train | fish_grill | synth | Write from scratch the weights of a policy network that plays Fish Grill as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 82 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/fish_grill.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameFishGrill.c |
knight_vs_pawns | train | knight_vs_pawns | synth | Write from scratch the weights of a policy network that plays Knight Vs Pawns as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 83 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/knight_vs_pawns.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameKnightVsPawns.c |
shiny | train | shiny | synth | Write from scratch the weights of a policy network that plays Shiny as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` cod... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 84 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/shiny.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameShiny.c |
wavy_bird | train | wavy_bird | synth | Write from scratch the weights of a policy network that plays Wavy Bird as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 85 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/wavy_bird.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameWavyBird.c |
descents | train | descents | synth | Write from scratch the weights of a policy network that plays Descents as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 86 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/descents.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/descents/main.js |
throw_m | train | throw_m | synth | Write from scratch the weights of a policy network that plays Throw M as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 87 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/throw_m.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/throwm/main.js |
wind_power | train | wind_power | synth | Write from scratch the weights of a policy network that plays Wind Power as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 88 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/wind_power.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameWindPower.c |
tutankham | train | tutankham | synth | Write from scratch the weights of a policy network that plays Tutankham as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits`... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 89 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/tutankham.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
find_a_star | train | find_a_star | synth | Write from scratch the weights of a policy network that plays Find A Star as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edit... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 90 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/find_a_star.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/findastar/main.js |
defender | train | defender | synth | Write from scratch the weights of a policy network that plays Defender as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` ... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 91 | atari_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/defender.py | 4038830175c4201ca10e004ed4441985f2acddf2 | null |
floater | train | floater | synth | Write from scratch the weights of a policy network that plays Floater as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 92 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/floater.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/floater/main.js |
color_roll | train | color_roll | synth | Write from scratch the weights of a policy network that plays Color Roll as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 93 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/color_roll.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameColorRoll.c |
aerial_bar | train | aerial_bar | synth | Write from scratch the weights of a policy network that plays Aerial Bar as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 94 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/aerial_bar.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameAeralBar.c |
pillars_3d | train | pillars_3d | synth | Write from scratch the weights of a policy network that plays Pillars 3D as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 95 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/pillars_3d.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gamePillars3d.c |
periscope_ping | train | periscope_ping | synth | Write from scratch the weights of a policy network that plays Periscope Ping as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `e... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 96 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/periscope_ping.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gamePeriscopePing.c |
inkinh | train | inkinh | synth | Write from scratch the weights of a policy network that plays Inkinh as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` co... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 97 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/inkinh.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/joyrider3774/crisp-game-lib-portable-sdl/blob/04bc2c4cfc2853628eda25a46190bb7ac6fe216a/src/games/gameInkinh.c |
smily_angry | train | smily_angry | synth | Write from scratch the weights of a policy network that plays Smily Angry as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edit... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 98 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/smily_angry.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/smilyangry/main.js |
m_rider | train | m_rider | synth | Write from scratch the weights of a policy network that plays M Rider as well as possible.
There is no training, data, loss, gradients, fine-tuning, code execution, calculator, or environment feedback. This is one response. Briefly describe the strategy and hidden units, then provide the weights in the final `edits` c... | You directly program neural-network weights. Follow the task constraints exactly. | 1,000 | 99 | crisp_synthetic | https://github.com/Trajectorylabs/minatar-synth/blob/4038830175c4201ca10e004ed4441985f2acddf2/minatar_synth/games/m_rider.py | 4038830175c4201ca10e004ed4441985f2acddf2 | https://github.com/abagames/crisp-game-lib-games/blob/cfb39d2f988feb5918eb83697145b5a35adabf58/docs/mrider/main.js |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.