Datasets:

reasoning-degeneration-dev
/

t1-wingdings-arena-frozenlake-kimi-k2-instruct

game_id stringclasses 1 value	env_id stringclasses 1 value	model stringclasses 1 value	opponent_model null	reward float64 0.1 1	num_turns int64 1 12	transcript stringclasses 10 values	error null	system_prompt stringclasses 1 value	wingdings_compliance float64 0.64 0.98
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	0.5	1	[{"turn": 0, "player_id": 0, "observation": "\n[GAME] Welcome to Frozen Lake!\n\nYou are represented by 'P' on the grid.\nGrid symbols:\n ' ' = Frozen surface (safe to walk on)\n 'H' = Hole (fall in and lose!)\n 'G' = Goal (reach this to win!)\n 'P' = Your current position\n\nAvailable actions: up, down, left, righ...	null	You are a game-playing agent. You must decide your next move at each turn. HERE IS YOUR CONSTRAINT: You must express ALL of your intermediate reasoning — your spatial analysis, planning, and decision-making — using ONLY symbolic characters. Use arrows (→ ← ↑ ↓ ↔ ⇒ ⇐), checkmarks and crosses (✓ ✗), boxes (□ ■), circles...	0.928571
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	0.125	1	[{"turn": 0, "player_id": 0, "observation": "\n[GAME] Welcome to Frozen Lake!\n\nYou are represented by 'P' on the grid.\nGrid symbols:\n ' ' = Frozen surface (safe to walk on)\n 'H' = Hole (fall in and lose!)\n 'G' = Goal (reach this to win!)\n 'P' = Your current position\n\nAvailable actions: up, down, left, righ...	null	You are a game-playing agent. You must decide your next move at each turn. HERE IS YOUR CONSTRAINT: You must express ALL of your intermediate reasoning — your spatial analysis, planning, and decision-making — using ONLY symbolic characters. Use arrows (→ ← ↑ ↓ ↔ ⇒ ⇐), checkmarks and crosses (✓ ✗), boxes (□ ■), circles...	0.970588
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	0.1	1	[{"turn": 0, "player_id": 0, "observation": "\n[GAME] Welcome to Frozen Lake!\n\nYou are represented by 'P' on the grid.\nGrid symbols:\n ' ' = Frozen surface (safe to walk on)\n 'H' = Hole (fall in and lose!)\n 'G' = Goal (reach this to win!)\n 'P' = Your current position\n\nAvailable actions: up, down, left, righ...	null	You are a game-playing agent. You must decide your next move at each turn. HERE IS YOUR CONSTRAINT: You must express ALL of your intermediate reasoning — your spatial analysis, planning, and decision-making — using ONLY symbolic characters. Use arrows (→ ← ↑ ↓ ↔ ⇒ ⇐), checkmarks and crosses (✓ ✗), boxes (□ ■), circles...	0.89
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	1	2	[{"turn": 0, "player_id": 0, "observation": "\n[GAME] Welcome to Frozen Lake!\n\nYou are represented by 'P' on the grid.\nGrid symbols:\n ' ' = Frozen surface (safe to walk on)\n 'H' = Hole (fall in and lose!)\n 'G' = Goal (reach this to win!)\n 'P' = Your current position\n\nAvailable actions: up, down, left, righ...	null	You are a game-playing agent. You must decide your next move at each turn. HERE IS YOUR CONSTRAINT: You must express ALL of your intermediate reasoning — your spatial analysis, planning, and decision-making — using ONLY symbolic characters. Use arrows (→ ← ↑ ↓ ↔ ⇒ ⇐), checkmarks and crosses (✓ ✗), boxes (□ ■), circles...	0.842982
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	1	2	[{"turn": 0, "player_id": 0, "observation": "\n[GAME] Welcome to Frozen Lake!\n\nYou are represented by 'P' on the grid.\nGrid symbols:\n ' ' = Frozen surface (safe to walk on)\n 'H' = Hole (fall in and lose!)\n 'G' = Goal (reach this to win!)\n 'P' = Your current position\n\nAvailable actions: up, down, left, righ...	null	You are a game-playing agent. You must decide your next move at each turn. HERE IS YOUR CONSTRAINT: You must express ALL of your intermediate reasoning — your spatial analysis, planning, and decision-making — using ONLY symbolic characters. Use arrows (→ ← ↑ ↓ ↔ ⇒ ⇐), checkmarks and crosses (✓ ✗), boxes (□ ■), circles...	0.889855
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	0.2	3	[{"turn": 0, "player_id": 0, "observation": "\n[GAME] Welcome to Frozen Lake!\n\nYou are represented by 'P' on the grid.\nGrid symbols:\n ' ' = Frozen surface (safe to walk on)\n 'H' = Hole (fall in and lose!)\n 'G' = Goal (reach this to win!)\n 'P' = Your current position\n\nAvailable actions: up, down, left, righ...	null	You are a game-playing agent. You must decide your next move at each turn. HERE IS YOUR CONSTRAINT: You must express ALL of your intermediate reasoning — your spatial analysis, planning, and decision-making — using ONLY symbolic characters. Use arrows (→ ← ↑ ↓ ↔ ⇒ ⇐), checkmarks and crosses (✓ ✗), boxes (□ ■), circles...	0.92323
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	1	4	[{"turn": 0, "player_id": 0, "observation": "\n[GAME] Welcome to Frozen Lake!\n\nYou are represented by 'P' on the grid.\nGrid symbols:\n ' ' = Frozen surface (safe to walk on)\n 'H' = Hole (fall in and lose!)\n 'G' = Goal (reach this to win!)\n 'P' = Your current position\n\nAvailable actions: up, down, left, righ...	null	You are a game-playing agent. You must decide your next move at each turn. HERE IS YOUR CONSTRAINT: You must express ALL of your intermediate reasoning — your spatial analysis, planning, and decision-making — using ONLY symbolic characters. Use arrows (→ ← ↑ ↓ ↔ ⇒ ⇐), checkmarks and crosses (✓ ✗), boxes (□ ■), circles...	0.782903
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	1	5	[{"turn": 0, "player_id": 0, "observation": "\n[GAME] Welcome to Frozen Lake!\n\nYou are represented by 'P' on the grid.\nGrid symbols:\n ' ' = Frozen surface (safe to walk on)\n 'H' = Hole (fall in and lose!)\n 'G' = Goal (reach this to win!)\n 'P' = Your current position\n\nAvailable actions: up, down, left, righ...	null	You are a game-playing agent. You must decide your next move at each turn. HERE IS YOUR CONSTRAINT: You must express ALL of your intermediate reasoning — your spatial analysis, planning, and decision-making — using ONLY symbolic characters. Use arrows (→ ← ↑ ↓ ↔ ⇒ ⇐), checkmarks and crosses (✓ ✗), boxes (□ ■), circles...	0.802324
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	1	7	[{"turn": 0, "player_id": 0, "observation": "\n[GAME] Welcome to Frozen Lake!\n\nYou are represented by 'P' on the grid.\nGrid symbols:\n ' ' = Frozen surface (safe to walk on)\n 'H' = Hole (fall in and lose!)\n 'G' = Goal (reach this to win!)\n 'P' = Your current position\n\nAvailable actions: up, down, left, righ...	null	You are a game-playing agent. You must decide your next move at each turn. HERE IS YOUR CONSTRAINT: You must express ALL of your intermediate reasoning — your spatial analysis, planning, and decision-making — using ONLY symbolic characters. Use arrows (→ ← ↑ ↓ ↔ ⇒ ⇐), checkmarks and crosses (✓ ✗), boxes (□ ■), circles...	0.642593
FrozenLake-v0_ep000	FrozenLake-v0	together_ai/moonshotai/Kimi-K2-Instruct-0905	null	1	12	"[{\"turn\": 0, \"player_id\": 0, \"observation\": \"\\n[GAME] Welcome to Frozen Lake!\\n\\nYou are (...TRUNCATED)	null	"You are a game-playing agent. You must decide your next move at each turn.\n\nHERE IS YOUR CONSTRAI(...TRUNCATED)	0.976588

t1-wingdings-arena-frozenlake-kimi-k2-instruct

Wingdings compliance evaluation on TextArena games.

Model must reason using only symbolic characters while playing interactive games. Compliance is measured on reasoning text only, not action commands.

Results

Metric	Value
Win rate	10/10
Compliance (mean)	0.8650
Errors	0

Details

Parameter	Value
Model	`together_ai/moonshotai/Kimi-K2-Instruct-0905`
Thinking	False
Games	FrozenLake-v0, FrozenLake-v0, FrozenLake-v0, FrozenLake-v0, FrozenLake-v0, FrozenLake-v0, FrozenLake-v0, FrozenLake-v0, FrozenLake-v0, FrozenLake-v0

Usage

from datasets import load_dataset
ds = load_dataset("reasoning-degeneration-dev/t1-wingdings-arena-frozenlake-kimi-k2-instruct", split="train")
print(f"Mean compliance: {sum(ds['wingdings_compliance']) / len(ds):.4f}")

Tracked in reasoning-degeneration-dev/PROJECT-MANIFEST

Downloads last month: 10

Number of rows:

Total file size:

38.9 kB