Dataset Viewer
Auto-converted to Parquet Duplicate
task
stringclasses
1 value
env_id
stringclasses
1 value
env_kwargs
dict
seed
int64
2.03B
2.03B
episode_seed
int64
3.03B
3.03B
episode
int64
0
100k
init_state
dict
init_state_hash
stringlengths
64
64
version
stringclasses
1 value
messages
listlengths
2
2
chunk_size
stringclasses
1 value
n_steps
int64
13
37
n_actions
int64
13
37
n_imagined_frames
int64
0
0
max_steps
int64
48
48
cot_row_type
stringclasses
1 value
ablation
stringclasses
1 value
schema_version
stringclasses
1 value
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{ "difficulty": "hard", "grid_size": 12, "food_count": 12, "num_ghosts": 1, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "render_size": 256 }
2,026,080,200
3,026,080,263
0
{ "backend": "custom_grid_pacman_python_assets", "difficulty": "hard", "grid_size": 12, "food_count": 12, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "walls": [ [ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ...
3b57ff24d5d3d89643883804b3e95cc16c6e61addd497c2b7485331094a4a584
visgym_pacman_oracle_success_hard_v1
[ { "role": "user", "content": [ { "type": "text", "text": "You are Pacman in a deterministic 12x12 maze. Collect every food pellet, avoid the single red ghost, and stop only after all food is gone. The rendering uses pacman-python Pacman, ghost, pellet, and maze tile visuals, but the en...
inf
15
15
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{ "difficulty": "hard", "grid_size": 12, "food_count": 12, "num_ghosts": 1, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "render_size": 256 }
2,026,080,200
3,026,080,256
1
{ "backend": "custom_grid_pacman_python_assets", "difficulty": "hard", "grid_size": 12, "food_count": 12, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "walls": [ [ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ...
6cc6cfb1fedb305d80a1b12c2c23b2dedf95b1cdbfde2f43200fabd42bc2e7d5
visgym_pacman_oracle_success_hard_v1
[ { "role": "user", "content": [ { "type": "text", "text": "You are Pacman in a deterministic 12x12 maze. Collect every food pellet, avoid the single red ghost, and stop only after all food is gone. The rendering uses pacman-python Pacman, ghost, pellet, and maze tile visuals, but the en...
inf
17
17
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{ "difficulty": "hard", "grid_size": 12, "food_count": 12, "num_ghosts": 1, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "render_size": 256 }
2,026,080,200
3,026,080,203
2
{ "backend": "custom_grid_pacman_python_assets", "difficulty": "hard", "grid_size": 12, "food_count": 12, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "walls": [ [ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ...
9aef798b75bfe59d1b737b0d753c1635a9107355ae7f6e329242bedf6bffb6fb
visgym_pacman_oracle_success_hard_v1
[ { "role": "user", "content": [ { "type": "text", "text": "You are Pacman in a deterministic 12x12 maze. Collect every food pellet, avoid the single red ghost, and stop only after all food is gone. The rendering uses pacman-python Pacman, ghost, pellet, and maze tile visuals, but the en...
inf
19
19
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{ "difficulty": "hard", "grid_size": 12, "food_count": 12, "num_ghosts": 1, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "render_size": 256 }
2,026,080,200
3,026,080,227
3
{ "backend": "custom_grid_pacman_python_assets", "difficulty": "hard", "grid_size": 12, "food_count": 12, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "walls": [ [ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ...
03e628ac9fc691ba37117f8bf4be2bb9358b21ecfd7506f761cc8d6736ca9b33
visgym_pacman_oracle_success_hard_v1
[ { "role": "user", "content": [ { "type": "text", "text": "You are Pacman in a deterministic 12x12 maze. Collect every food pellet, avoid the single red ghost, and stop only after all food is gone. The rendering uses pacman-python Pacman, ghost, pellet, and maze tile visuals, but the en...
inf
30
30
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{ "difficulty": "hard", "grid_size": 12, "food_count": 12, "num_ghosts": 1, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "render_size": 256 }
2,026,080,200
3,026,080,265
4
{ "backend": "custom_grid_pacman_python_assets", "difficulty": "hard", "grid_size": 12, "food_count": 12, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "walls": [ [ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ...
1000e828e854ef352dd5a8625017e5ce15f50750a8ce09962f42510ff069827a
visgym_pacman_oracle_success_hard_v1
[ { "role": "user", "content": [ { "type": "text", "text": "You are Pacman in a deterministic 12x12 maze. Collect every food pellet, avoid the single red ghost, and stop only after all food is gone. The rendering uses pacman-python Pacman, ghost, pellet, and maze tile visuals, but the en...
inf
21
21
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{ "difficulty": "hard", "grid_size": 12, "food_count": 12, "num_ghosts": 1, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "render_size": 256 }
2,026,080,200
3,026,080,210
5
{ "backend": "custom_grid_pacman_python_assets", "difficulty": "hard", "grid_size": 12, "food_count": 12, "wall_density": 0.16, "max_steps": 48, "horizon": 8, "tile_size": 24, "walls": [ [ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ...
9e3e87cf202da0b1d10dfc5eb1b771ab534e3267080a911dd339f5c404721757
visgym_pacman_oracle_success_hard_v1
[ { "role": "user", "content": [ { "type": "text", "text": "You are Pacman in a deterministic 12x12 maze. Collect every food pellet, avoid the single red ghost, and stop only after all food is gone. The rendering uses pacman-python Pacman, ghost, pellet, and maze tile visuals, but the en...
inf
23
23
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{"difficulty":"hard","grid_size":12,"food_count":12,"num_ghosts":1,"wall_density":0.16,"max_steps":4(...TRUNCATED)
2,026,080,200
3,026,080,212
6
{"backend":"custom_grid_pacman_python_assets","difficulty":"hard","grid_size":12,"food_count":12,"wa(...TRUNCATED)
10f91d48585bbc10da8824538b41dcf9716f64f9c43d3e22bdeda4f5f1feb513
visgym_pacman_oracle_success_hard_v1
[{"role":"user","content":[{"type":"text","text":"You are Pacman in a deterministic 12x12 maze. Coll(...TRUNCATED)
inf
24
24
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{"difficulty":"hard","grid_size":12,"food_count":12,"num_ghosts":1,"wall_density":0.16,"max_steps":4(...TRUNCATED)
2,026,080,200
3,026,080,225
7
{"backend":"custom_grid_pacman_python_assets","difficulty":"hard","grid_size":12,"food_count":12,"wa(...TRUNCATED)
d316eed729dc2e0a494800808fbd32610e6bbc879d0bbb485886e83a2aa5bac1
visgym_pacman_oracle_success_hard_v1
[{"role":"user","content":[{"type":"text","text":"You are Pacman in a deterministic 12x12 maze. Coll(...TRUNCATED)
inf
15
15
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{"difficulty":"hard","grid_size":12,"food_count":12,"num_ghosts":1,"wall_density":0.16,"max_steps":4(...TRUNCATED)
2,026,080,200
3,026,080,221
8
{"backend":"custom_grid_pacman_python_assets","difficulty":"hard","grid_size":12,"food_count":12,"wa(...TRUNCATED)
cc771252ad63e4ad113e88f8d5753efea8ec4c229f4d37bf5f5f273ab977c7c0
visgym_pacman_oracle_success_hard_v1
[{"role":"user","content":[{"type":"text","text":"You are Pacman in a deterministic 12x12 maze. Coll(...TRUNCATED)
inf
28
28
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
pacman_2d_hard_oracle_success_v0
pacman_2d/hard
{"difficulty":"hard","grid_size":12,"food_count":12,"num_ghosts":1,"wall_density":0.16,"max_steps":4(...TRUNCATED)
2,026,080,200
3,026,080,222
9
{"backend":"custom_grid_pacman_python_assets","difficulty":"hard","grid_size":12,"food_count":12,"wa(...TRUNCATED)
aad0ba4ff08562d7536176f43dd431e603dce6af2a7aff222870e2b7e1ae1893
visgym_pacman_oracle_success_hard_v1
[{"role":"user","content":[{"type":"text","text":"You are Pacman in a deterministic 12x12 maze. Coll(...TRUNCATED)
inf
16
16
0
48
singleshot_openloop_baseline
single_frame_openloop_no_imagination
pacman_singleshot_openloop_v1
End of preview. Expand in Data Studio

pacman_hard_singleshot_train

BAGEL VLM-Gym world-model dataset (pacman / singleshot).

  • Single-shot open-loop train set (non-CoT; imagine the whole episode).
  • layout: Train-only. Gzipped-JSONL shards under training/; each row is one packed SFT sample with base64-JPEG frames inline.
  • images are base64-encoded JPEG frames stored inline in each JSONL row.

Pairs with the matching pacman checkpoint(s) under the companion model org; CoT and non-CoT variants share the same underlying rollouts.

Downloads last month
-