Dataset Viewer (First 5GB)
Auto-converted to Parquet Duplicate
task
stringclasses
1 value
env_id
stringclasses
1 value
env_kwargs
unknown
seed
int64
10M
1.31B
episode_seed
int64
10M
1.31B
extra_state
null
timestamp
stringdate
2026-04-25 05:12:22
2026-04-25 05:14:49
history
listlengths
5
20
stats
dict
init_state
dict
solve_kwargs
dict
version
stringclasses
1 value
episode
int64
0
130M
difficulty
stringclasses
1 value
dataset_pipeline
stringclasses
1 value
init_state_hash
stringlengths
64
64
solve_info
dict
generated_at
stringdate
2026-04-25 05:12:22
2026-04-25 05:14:49
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,000
null
2026-04-25T05:12:22.317197
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 8, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
0
easy
behavior_cloning
c3920c9d23cff2a403dad004b7119653bb504839d98397a4bdd2a83a3ebe05e0
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:22.317782
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,001
null
2026-04-25T05:12:22.481746
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 9, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
1
easy
behavior_cloning
7c1afe6233830aa570ed7658d3a47830a27fddcf775aef7daf9e93b94e88abbe
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:22.482223
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,002
null
2026-04-25T05:12:22.855139
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 14, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
2
easy
behavior_cloning
a9921b72d9e6c146b7d4be61070689d62ec7d19c6c4abbe6a8606657225dd0f0
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:22.855543
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,003
null
2026-04-25T05:12:23.019142
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 9, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
3
easy
behavior_cloning
350b45953941eb2c5b03f8f8c26d826ba0a7bcc83306a46ea2870f7668a86b61
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:23.019570
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,004
null
2026-04-25T05:12:23.456061
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 14, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
4
easy
behavior_cloning
320fc0147125b538063c6f5c66f712864ab8afbf0dfefbf247aa767b6d440ace
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:23.456981
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,005
null
2026-04-25T05:12:23.589431
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 8, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
5
easy
behavior_cloning
72d0f9accfd67883b614bfae2509aac2b1c8bf30fd92d512c0e6f1055874a1e4
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:23.590213
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,006
null
2026-04-25T05:12:23.674364
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 6, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
6
easy
behavior_cloning
43e51de06c0c5e941e416e52a1642fc5db2fa6c3c6b9a35307b47ee339d6f500
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:23.675880
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,007
null
2026-04-25T05:12:23.764038
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 7, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
7
easy
behavior_cloning
fdfe116ddb339f6f55306885339dbfebd206d5b80ff47a9bfb7cf670827e8817
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:23.764444
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,008
null
2026-04-25T05:12:24.061900
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 11, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
8
easy
behavior_cloning
27f95ce730b7624b2f2b050c61970adef305bdf67eb454f0926859352b08e201
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:24.062383
pacman_2d_easy_v0
pacman_2d/easy
{}
10,000,000
10,000,009
null
2026-04-25T05:12:24.266233
[{"step":0,"prompt":"You are Pacman in a deterministic 9x9 maze. Collect every food pellet, avoid th(...TRUNCATED)
{ "step": 9, "reward": 1, "terminated": true, "truncated": false }
{"backend":"custom_grid_pacman_python_assets","difficulty":"easy","grid_size":9,"food_count":4,"wall(...TRUNCATED)
{ "strategy": "receding_horizon", "horizon": 8 }
visgym_repo_solver_v1
9
easy
behavior_cloning
aa10ae97ebacd4d4c120b0a9b9df36e0a3424c889108b0bfa08604f12d3bb9aa
{"strategy_requested":"receding_horizon","planner_used":"shortest_safe","fallback_planner":"receding(...TRUNCATED)
2026-04-25T05:12:24.266698
End of preview. Expand in Data Studio

VisGym Pacman2D Deterministic Trajectories

This dataset contains behavior-cloning trajectories for the custom VisGym Pacman2D environment.

Each row is one successful oracle episode. The history entries include image_prev, image, and image_next; no synthetic thinking traces are stored. Images are JPEG base64 strings rendered from the greyblue9/pacman-python visual assets used by the environment.

Environment summary:

  • Grid size is constrained to 7x7 through 12x12.
  • Easy uses the 9x9 setting; hard uses the 11x11 setting.
  • Entities are Pacman, sparse food, walls, and exactly one deterministic ghost.
  • The ghost chases Pacman by BFS distance with fixed tie-break order up, down, left, right.
  • The movement action space is four directions; VisGym records can also end with ('stop', 'stop').

Splits:

  • data/pacman_2d_easy_v0/train/*.jsonl.gz
  • data/pacman_2d_easy_v0/test/*.jsonl.gz
  • data/pacman_2d_hard_v0/train/*.jsonl.gz
  • data/pacman_2d_hard_v0/test/*.jsonl.gz

Train/test initial states are checked to be disjoint by init_state_hash; see metadata/init_state_hash_audit.json.

Target Hub repo: https://huggingface.co/datasets/novastar111/pacman_v0

Downloads last month
717