Dataset Viewer
Auto-converted to Parquet Duplicate
task
stringclasses
1 value
answer
stringlengths
35
38
reward
int64
3
46
prompt
listlengths
1
1
completion
listlengths
7
93
snake/vanilla-8x8
apples_eaten=21; died=False; steps=126
21
[ { "role": "system", "content": "You are a Snake-planning assistant. On each turn, the user sends one apple-to-apple observation:\n\n<APPLE id=K>\n<OBS>\nsize:N\nhead:(r,c)\nbody:{(r0,c0),(r1,c1),...,(rL-1,cL-1)} # includes the head at index 0 and the tail at index L-1\nfood:(rf,cf)\n</OBS>\n</APPLE>\n\nYo...
[ { "role": "user", "content": "<APPLE id=1>\n<OBS>\nsize:8\nhead:(0,4)\nbody:{(0,4),(0,3),(0,2)}\nfood:(1,0)\n</OBS>\n</APPLE>" }, { "role": "assistant", "content": "<THINK>\npath coordinates: [(1,4), (1,3), (1,2), (1,1), (1,0)]\nmoves to reach the apple: [DOWN, LEFT, LEFT, LEFT, LEFT]\nbody squa...
snake/vanilla-8x8
apples_eaten=27; died=False; steps=164
27
[ { "role": "system", "content": "You are a Snake-planning assistant. On each turn, the user sends one apple-to-apple observation:\n\n<APPLE id=K>\n<OBS>\nsize:N\nhead:(r,c)\nbody:{(r0,c0),(r1,c1),...,(rL-1,cL-1)} # includes the head at index 0 and the tail at index L-1\nfood:(rf,cf)\n</OBS>\n</APPLE>\n\nYo...
[ { "role": "user", "content": "<APPLE id=1>\n<OBS>\nsize:8\nhead:(4,0)\nbody:{(4,0),(4,1),(4,2)}\nfood:(1,3)\n</OBS>\n</APPLE>" }, { "role": "assistant", "content": "<THINK>\npath coordinates: [(3,0), (2,0), (1,0), (1,1), (1,2), (1,3)]\nmoves to reach the apple: [UP, UP, UP, RIGHT, RIGHT, RIGHT]\...
snake/vanilla-8x8
apples_eaten=28; died=False; steps=200
28
[ { "role": "system", "content": "You are a Snake-planning assistant. On each turn, the user sends one apple-to-apple observation:\n\n<APPLE id=K>\n<OBS>\nsize:N\nhead:(r,c)\nbody:{(r0,c0),(r1,c1),...,(rL-1,cL-1)} # includes the head at index 0 and the tail at index L-1\nfood:(rf,cf)\n</OBS>\n</APPLE>\n\nYo...
[ { "role": "user", "content": "<APPLE id=1>\n<OBS>\nsize:8\nhead:(3,3)\nbody:{(3,3),(3,2),(3,1)}\nfood:(7,0)\n</OBS>\n</APPLE>" }, { "role": "assistant", "content": "<THINK>\npath coordinates: [(4,3), (4,2), (4,1), (4,0), (5,0), (6,0), (7,0)]\nmoves to reach the apple: [DOWN, LEFT, LEFT, LEFT, DO...
snake/vanilla-8x8
apples_eaten=27; died=False; steps=161
27
[ { "role": "system", "content": "You are a Snake-planning assistant. On each turn, the user sends one apple-to-apple observation:\n\n<APPLE id=K>\n<OBS>\nsize:N\nhead:(r,c)\nbody:{(r0,c0),(r1,c1),...,(rL-1,cL-1)} # includes the head at index 0 and the tail at index L-1\nfood:(rf,cf)\n</OBS>\n</APPLE>\n\nYo...
[ { "role": "user", "content": "<APPLE id=1>\n<OBS>\nsize:8\nhead:(0,7)\nbody:{(0,7),(0,6),(0,5)}\nfood:(2,2)\n</OBS>\n</APPLE>" }, { "role": "assistant", "content": "<THINK>\npath coordinates: [(1,7), (1,6), (1,5), (1,4), (1,3), (1,2), (2,2)]\nmoves to reach the apple: [DOWN, LEFT, LEFT, LEFT, LE...
snake/vanilla-8x8
apples_eaten=33; died=False; steps=237
33
[ { "role": "system", "content": "You are a Snake-planning assistant. On each turn, the user sends one apple-to-apple observation:\n\n<APPLE id=K>\n<OBS>\nsize:N\nhead:(r,c)\nbody:{(r0,c0),(r1,c1),...,(rL-1,cL-1)} # includes the head at index 0 and the tail at index L-1\nfood:(rf,cf)\n</OBS>\n</APPLE>\n\nYo...
[ { "role": "user", "content": "<APPLE id=1>\n<OBS>\nsize:8\nhead:(0,1)\nbody:{(0,1),(0,2),(0,3)}\nfood:(0,4)\n</OBS>\n</APPLE>" }, { "role": "assistant", "content": "<THINK>\npath coordinates: [(1,1), (1,2), (0,2), (0,3), (0,4)]\nmoves to reach the apple: [DOWN, RIGHT, UP, RIGHT, RIGHT]\nbody squ...
snake/vanilla-8x8
apples_eaten=14; died=False; steps=92
14
[ { "role": "system", "content": "You are a Snake-planning assistant. On each turn, the user sends one apple-to-apple observation:\n\n<APPLE id=K>\n<OBS>\nsize:N\nhead:(r,c)\nbody:{(r0,c0),(r1,c1),...,(rL-1,cL-1)} # includes the head at index 0 and the tail at index L-1\nfood:(rf,cf)\n</OBS>\n</APPLE>\n\nYo...
[ { "role": "user", "content": "<APPLE id=1>\n<OBS>\nsize:8\nhead:(3,6)\nbody:{(3,6),(3,5),(3,4)}\nfood:(4,3)\n</OBS>\n</APPLE>" }, { "role": "assistant", "content": "<THINK>\npath coordinates: [(4,6), (4,5), (4,4), (4,3)]\nmoves to reach the apple: [DOWN, LEFT, LEFT, LEFT]\nbody squares on this p...
snake/vanilla-8x8
apples_eaten=26; died=False; steps=178
26
[ { "role": "system", "content": "You are a Snake-planning assistant. On each turn, the user sends one apple-to-apple observation:\n\n<APPLE id=K>\n<OBS>\nsize:N\nhead:(r,c)\nbody:{(r0,c0),(r1,c1),...,(rL-1,cL-1)} # includes the head at index 0 and the tail at index L-1\nfood:(rf,cf)\n</OBS>\n</APPLE>\n\nYo...
[ { "role": "user", "content": "<APPLE id=1>\n<OBS>\nsize:8\nhead:(4,0)\nbody:{(4,0),(4,1),(4,2)}\nfood:(7,3)\n</OBS>\n</APPLE>" }, { "role": "assistant", "content": "<THINK>\npath coordinates: [(5,0), (5,1), (5,2), (5,3), (6,3), (7,3)]\nmoves to reach the apple: [DOWN, RIGHT, RIGHT, RIGHT, DOWN, ...
snake/vanilla-8x8
apples_eaten=23; died=False; steps=149
23
[ { "role": "system", "content": "You are a Snake-planning assistant. On each turn, the user sends one apple-to-apple observation:\n\n<APPLE id=K>\n<OBS>\nsize:N\nhead:(r,c)\nbody:{(r0,c0),(r1,c1),...,(rL-1,cL-1)} # includes the head at index 0 and the tail at index L-1\nfood:(rf,cf)\n</OBS>\n</APPLE>\n\nYo...
[ { "role": "user", "content": "<APPLE id=1>\n<OBS>\nsize:8\nhead:(6,4)\nbody:{(6,4),(6,3),(6,2)}\nfood:(6,5)\n</OBS>\n</APPLE>" }, { "role": "assistant", "content": "<THINK>\npath coordinates: [(6,5)]\nmoves to reach the apple: [RIGHT]\nbody squares on this path: []\narrival steps n (to each body...
snake/vanilla-8x8
apples_eaten=29; died=False; steps=230
29
[{"role":"system","content":"You are a Snake-planning assistant. On each turn, the user sends one ap(...TRUNCATED)
[{"role":"user","content":"<APPLE id=1>\n<OBS>\nsize:8\nhead:(5,5)\nbody:{(5,5),(5,4),(5,3)}\nfood:((...TRUNCATED)
snake/vanilla-8x8
apples_eaten=27; died=False; steps=154
27
[{"role":"system","content":"You are a Snake-planning assistant. On each turn, the user sends one ap(...TRUNCATED)
[{"role":"user","content":"<APPLE id=1>\n<OBS>\nsize:8\nhead:(6,2)\nbody:{(6,2),(6,1),(6,0)}\nfood:((...TRUNCATED)
End of preview. Expand in Data Studio
YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

Snake SFT Dataset

A dataset of 2,000 Snake gameplay episodes with step-by-step reasoning and planning.

Dataset Statistics

  • Episodes: 2,000
  • Training examples: 48,429
  • Board size: 8x8
  • Average apples per episode: 24.2
  • Episodes with 25+ apples: 984 (49.2%)

Data Format

Each line contains a JSON object with:

  • task: Game identifier
  • answer: Episode summary (apples eaten, death status, steps)
  • reward: Number of apples eaten
  • prompt: System prompt for the AI
  • completion: Alternating user/assistant messages showing the gameplay

User Messages

Contain <APPLE id=X> observations with:

  • Snake head position
  • Snake body coordinates
  • Apple location
  • Board size

Assistant Messages

Contain <THINK> reasoning blocks with:

  • Path coordinates to the apple
  • Move sequence
  • Tail safety calculations
  • Safety verification

Followed by <ACTIONS> with the actual moves to execute.

Usage

This dataset is designed for fine-tuning language models to play Snake with sophisticated planning and reasoning capabilities.

License

Apache 2.0

Downloads last month
4