problem stringlengths 197 728 | reward_model dict | data_source stringclasses 1
value | extra_info dict | _rollout_source_idx int64 0 3.21k | expert_hints listlengths 6 6 | generated_hints listlengths 4 4 | generated_hint_responses listlengths 4 4 | generated_answers listlengths 4 4 | generated_responses listlengths 4 4 | solver_input_history_by_step listlengths 4 4 | hint_input_history_by_step listlengths 4 4 | solver_summary_update_history listlengths 4 4 | solver_summary_update_responses listlengths 4 4 | solver_summary_by_step listlengths 4 4 | final_solver_summary listlengths 4 4 | hint_summary_update_history listlengths 4 4 | hint_summary_update_responses listlengths 4 4 | hint_summary_by_step listlengths 4 4 | final_hint_summary listlengths 4 4 | revealed_hints_by_step listlengths 4 4 | actual_solver_revealed_hint_counts listlengths 4 4 | hint_sequence listlengths 4 4 | hint_similarity_scores listlengths 4 4 | hint_similarity_bonus_scores listlengths 4 4 | hint_similarity_raw_scores listlengths 4 4 | hint_similarity_outputs listlengths 4 4 | llm_judge_scores listlengths 4 4 | llm_judge_raw_scores listlengths 4 4 | llm_judge_outputs listlengths 4 4 | llm_judge_mean_score float64 0 1 | llm_judge_max_score float64 0 1 | llm_judge_step_mean_scores listlengths 5 5 | hint_similarity_mean_score float64 0.01 0.94 | hint_similarity_max_score float64 0.1 1 | hint_similarity_mean_bonus_score float64 0 0.23 | hint_similarity_step_mean_scores listlengths 4 4 | rollout_config dict | hint_similarity_inputs listlengths 4 4 | llm_judge_inputs listlengths 4 4 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
"+--------+--------+---------+-----------+\n| WANDA | RAN | ROOT | TOUCHDOWN |\n+--------+---(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"GIVE A TITLE TO\", \"words\": [\"CALL\", \"DUB\", \"LABE(...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 548","date":"2024-12-10T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 0 | ["Consider the category: GIVE A TITLE TO as a potential category for the words in the grid.","For th(...TRUNCATED) | [["ice cream**?\n**Night ice cream**?\n**Dawn ice cream**?\n**Dusk ice cream**?\n**Sunrise ice cream(...TRUNCATED) | [["<think>\nThe solver has repeated the exact same incorrect groupings from the previous turn: \"AVI(...TRUNCATED) | [[["AVIATION PHASES -> TAKEOFF, CRUISE, TAXI, TOUCHDOWN | VERBS MEANING TO NOMINATE -> CALL, DUB, LA(...TRUNCATED) | [[["<think>\nThe goal is to solve a Connections-style puzzle. I have a 4x4 grid of words. I need to (...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- Board state: 4x4 grid with words WANDA, RAN, ROOT, TOUCHDOWN, INLAND, FISH, TAKEOFF,(...TRUNCATED) | [["- Board state: 4x4 grid with words WANDA, RAN, ROOT, TOUCHDOWN, INLAND, FISH, TAKEOFF, FUMBLE, CA(...TRUNCATED) | ["- **Confirmed Groups**: 1) Words in famous movie titles (WANDA, RAN, INLAND, OLIVIA); 2) Phases of(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- Solver attempted 4 categories: AVIATION PHASES (Takeoff, Cruise, Taxi, Touchdown), N(...TRUNCATED) | [["- Solver attempted 4 categories: AVIATION PHASES (Takeoff, Cruise, Taxi, Touchdown), NOMINATE VER(...TRUNCATED) | ["- Solver submitted the exact correct solution for all four groups: Movie Title Words (WANDA, RAN, (...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["ice cream**?\n**Night ice cream**?\n**Dawn ice cream**?\n**Dusk ice cream**?\n**Sunrise ice cream(...TRUNCATED) | [
[
0,
0,
0,
0
],
[
0,
0,
0.7,
0.3
],
[
0.5,
0,
0,
0.9
],
[
0,
1,
0,
0
]
] | [
[
0,
0,
0,
0
],
[
0,
0,
0.175,
0.075
],
[
0.125,
0,
0,
0.225
],
[
0,
0.25,
0,
0
]
] | [
[
0,
0,
0,
0
],
[
0,
0,
7,
3
],
[
5,
0,
0,
9
],
[
0,
10,
0,
0
]
] | [["<think>\nWe need to evaluate candidate hint: A long list of phrases starting with \"ice cream?\" (...TRUNCATED) | [[[1.0],[1.0],[1.0],[1.0],[1.0]],[[1.0],[1.0],[0.0],[1.0],[0.75]],[[0.5],[0.0],[0.5],[0.5],[1.0]],[[(...TRUNCATED) | [[[4.0],[4.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[0.0],[4.0],[3.0]],[[2.0],[0.0],[2.0],[2.0],[4.0]],[[1(...TRUNCATED) | [[["<think>\nWe compare candidate answer:\n\n\"AVIATION PHASES -> TAKEOFF, CRUISE, TAXI, TOUCHDOWN |(...TRUNCATED) | 0.65 | 1 | [
0.6875,
0.5,
0.625,
0.625,
0.8125
] | 0.2125 | 1 | 0.053125 | [
0.125,
0.25,
0.175,
0.3
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
"+--------+---------+-----------+--------+\n| BIG | POPULAR | HANDSOME | HIPPO |\n+--------+---(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"OF-THE-MOMENT\", \"words\": [\"BIG\", \"HOT\", \"IN\", \(...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 228","date":"2024-01-25T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 1 | ["Consider the category: PICTOGRAPH as a potential category for the words in the grid.","For the cat(...TRUNCATED) | [["You have successfully identified all four groups, including the clever \"Body Parts\" category th(...TRUNCATED) | [["<hint>\nYou have successfully identified all four groups, including the clever \"Body Parts\" cat(...TRUNCATED) | [[["TAX AND PENALTY TERMS -> ASSESS, CHARGE, FINE, LEVY | WORDS WITH BODY PARTS -> HANDSOME, HIPPO, (...TRUNCATED) | [[["<think>\nThe goal is to solve a Connections puzzle. I have 16 words arranged in a 4x4 grid. I ne(...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- Board state: 16 words fixed (BIG, POPULAR, HANDSOME, HIPPO, CHARGE, GLYPH, LEGEND, A(...TRUNCATED) | [["- Board state: 16 words fixed (BIG, POPULAR, HANDSOME, HIPPO, CHARGE, GLYPH, LEGEND, ASSESS, HOT,(...TRUNCATED) | ["- Puzzle status: Fully solved and confirmed correct with no errors.\n- Group 1 (Taxes/Penalties): (...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- Solver successfully identified 4 correct categories: Tax/Penalty terms (ASSESS, CHAR(...TRUNCATED) | [["- Solver successfully identified 4 correct categories: Tax/Penalty terms (ASSESS, CHARGE, FINE, L(...TRUNCATED) | ["- Solver correctly identified all four categories: Taxes/Penalties, Visual Signs, Popular Synonyms(...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["You have successfully identified all four groups, including the clever \"Body Parts\" category th(...TRUNCATED) | [
[
1,
0,
1,
1
],
[
0.9,
0.7,
1,
1
],
[
1,
0,
0,
0
],
[
1,
1,
1,
1
]
] | [
[
0.25,
0,
0.25,
0.25
],
[
0.225,
0.175,
0.25,
0.25
],
[
0.25,
0,
0,
0
],
[
0.25,
0.25,
0.25,
0.25
]
] | [
[
10,
0,
10,
10
],
[
9,
7,
10,
10
],
[
10,
0,
0,
0
],
[
10,
10,
10,
10
]
] | [["<think>\nWe need to match candidate hint \"You have successfully identified all four groups, incl(...TRUNCATED) | [[[1.0],[1.0],[1.0],[1.0],[1.0]],[[0.0],[0.5],[1.0],[1.0],[1.0]],[[1.0],[1.0],[1.0],[1.0],[0.0]],[[1(...TRUNCATED) | [[[4.0],[4.0],[4.0],[4.0],[4.0]],[[0.0],[2.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[4.0],[4.0],[0.0]],[[4(...TRUNCATED) | [[["<think>\nWe need to analyze candidate's answer.\n\nPuzzle:\n\nGrid 4x4: words:\n\nRow1: BIG | PO(...TRUNCATED) | 0.875 | 1 | [
0.75,
0.875,
1,
1,
0.75
] | 0.725 | 1 | 0.18125 | [
0.975,
0.425,
0.75,
0.75
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
"+------------+--------+---------+--------+\n| DITCH | HOLE | JUDGE | BOXER |\n+----------(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"TYPES OF ACADEMIC COURSES\", \"words\": [\"DISCUSSION\",(...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 591","date":"2025-01-22T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 2 | ["Consider the category: RESULTS OF SOME DIGGING as a potential category for the words in the grid."(...TRUNCATED) | [["Your categorization is perfect; you have successfully solved the puzzle by identifying all four g(...TRUNCATED) | [["<think>\nThe solver's latest attempt perfectly matches the true solution provided in the compact (...TRUNCATED) | [[["EXCAVATIONS -> DITCH, HOLE, PIT, TRENCH | LEARNING SETTING -> DISCUSSION, LAB, LECTURE, SEMINAR (...TRUNCATED) | [[["<think>\nLet me look at the words in the grid and start brainstorming possible connections.\n\nT(...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- Board contains 16 words: DITCH, HOLE, JUDGE, BOXER, LAB, LAWYER, CHIN, PIT, DISCUSSI(...TRUNCATED) | [["- Board contains 16 words: DITCH, HOLE, JUDGE, BOXER, LAB, LAWYER, CHIN, PIT, DISCUSSION, TRENCH,(...TRUNCATED) | ["- Puzzle fully solved; all 16 words correctly assigned to four confirmed groups.\n- Group 1 (Excav(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- Solver successfully identified all four categories: EXCAVATIONS (DITCH, HOLE, PIT, T(...TRUNCATED) | [["- Solver successfully identified all four categories: EXCAVATIONS (DITCH, HOLE, PIT, TRENCH), LEA(...TRUNCATED) | ["- Solver successfully identified all four categories: EXCAVATIONS, LEARNING SETTINGS, BOXING TERMS(...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["Your categorization is perfect; you have successfully solved the puzzle by identifying all four g(...TRUNCATED) | [
[
0,
0,
0,
0
],
[
1,
0.3,
1,
1
],
[
1,
0,
0,
0
],
[
0.9,
0,
0.2,
0.2
]
] | [
[
0,
0,
0,
0
],
[
0.25,
0.075,
0.25,
0.25
],
[
0.25,
0,
0,
0
],
[
0.225,
0,
0.05,
0.05
]
] | [
[
0,
0,
0,
0
],
[
10,
3,
10,
10
],
[
10,
0,
0,
0
],
[
9,
0,
2,
2
]
] | [["<think>\nWe need to compare candidate hint with expert hints.\n\nExpert hints: There are three ca(...TRUNCATED) | [[[0.5],[0.5],[0.5],[0.5],[0.5]],[[1.0],[1.0],[1.0],[1.0],[1.0]],[[0.0],[0.0],[0.0],[0.0],[0.5]],[[0(...TRUNCATED) | [[[2.0],[2.0],[2.0],[2.0],[2.0]],[[4.0],[4.0],[4.0],[4.0],[4.0]],[[0.0],[0.0],[0.0],[0.0],[2.0]],[[3(...TRUNCATED) | [[["<think>\nWe need to check candidate answer groups.\n\nGround truth groups:\n\n1. Types of academ(...TRUNCATED) | 0.5375 | 1 | [
0.5625,
0.625,
0.5,
0.375,
0.625
] | 0.35 | 1 | 0.0875 | [
0.725,
0.075,
0.3,
0.3
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
"+---------+-----------+-----------+--------+\n| MURPHY | CASSEROLE | POT | CIRCLE |\n+------(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"LOOP\", \"words\": [\"BAND\", \"CIRCLE\", \"HOOP\", \"RI(...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 559","date":"2024-12-21T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 3 | ["Consider the category: COOKING VESSELS as a potential category for the words in the grid.","For th(...TRUNCATED) | [["Consider that \"MURPHY,\" \"BUNK,\" and \"CANOPY\" are types of something, but \"SLEIGH\" refers (...TRUNCATED) | [["<hint>\nConsider that \"MURPHY,\" \"BUNK,\" and \"CANOPY\" are types of something, but \"SLEIGH\"(...TRUNCATED) | [[["BEDS -> MURPHY, BUNK, CANOPY, SLEIGH | KITCHEN VESSELS -> CASSEROLE, CROCK, PAN, POT | CIRCULAR (...TRUNCATED) | [[["<think>\nThe goal is to solve a Connections-style puzzle. I have a 4x4 grid of words. I need to (...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- Board state: 4x4 grid with words MURPHY, CASSEROLE, POT, CIRCLE, GROUCH, PAN, STATUE(...TRUNCATED) | [["- Board state: 4x4 grid with words MURPHY, CASSEROLE, POT, CIRCLE, GROUCH, PAN, STATUETTE, O, BUN(...TRUNCATED) | ["- **Confirmed Groups**: BED TYPES (MURPHY, BUNK, CANOPY, SLEIGH), COOKING VESSELS (CASSEROLE, CROC(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- Solver attempted 4 categories: BEDS, KITCHEN VESSELS, CIRCULAR OBJECTS, OSCARS.\n- B(...TRUNCATED) | [["- Solver attempted 4 categories: BEDS, KITCHEN VESSELS, CIRCULAR OBJECTS, OSCARS.\n- BEDS (MURPHY(...TRUNCATED) | ["- **Status**: Puzzle is fully solved; solver correctly identified all four categories on the final(...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["Consider that \"MURPHY,\" \"BUNK,\" and \"CANOPY\" are types of something, but \"SLEIGH\" refers (...TRUNCATED) | [
[
1,
0.8,
0,
0
],
[
1,
0,
1,
0
],
[
1,
0.3,
0,
1
],
[
0,
0,
0,
0
]
] | [
[
0.25,
0.2,
0,
0
],
[
0.25,
0,
0.25,
0
],
[
0.25,
0.075,
0,
0.25
],
[
0,
0,
0,
0
]
] | [
[
10,
8,
0,
0
],
[
10,
0,
10,
0
],
[
10,
3,
0,
10
],
[
0,
0,
0,
0
]
] | [["<think>\nWe need to evaluate the candidate hint relative to expert hints. Expert hints are catego(...TRUNCATED) | [[[1.0],[0.0],[1.0],[1.0],[1.0]],[[1.0],[1.0],[1.0],[1.0],[1.0]],[[0.0],[1.0],[0.5],[0.0],[0.5]],[[0(...TRUNCATED) | [[[4.0],[0.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[4.0],[4.0],[4.0]],[[0.0],[4.0],[2.0],[0.0],[2.0]],[[0(...TRUNCATED) | [[["<think>\nWe need to evaluate candidate answer and ground truth. Candidate answer:\n\n\"BEDS -> M(...TRUNCATED) | 0.75 | 1 | [
0.5,
0.75,
0.875,
0.75,
0.875
] | 0.38125 | 1 | 0.095313 | [
0.75,
0.275,
0.25,
0.25
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
"+-----------+-------+------+--------+\n| BALL | DEAR | FLY | DASH |\n+-----------+-------+(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"MOVE QUICKLY\", \"words\": [\"BOLT\", \"DART\", \"DASH\"(...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 569","date":"2024-12-31T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 4 | ["Consider the category: NAME ___ as a potential category for the words in the grid.","For the categ(...TRUNCATED) | [["and Exploitation.\n- Multi-Agent and Cooperative.\n- Swarm Intelligence and Ant Colony.\n- Geneti(...TRUNCATED) | [["The solver is fixated on \"Brand Name\" and \"Game Name,\" but \"Name\" is not the unifying word;(...TRUNCATED) | [[["Dear / Bear / Tear / Wear.\nBrand / Land / Sand / Hand.\nNo.\n\nLet's try \"DEAR\", \"BRAND\", \(...TRUNCATED) | [[["<think>\nThe user wants me to solve a Connections-style puzzle.\nI need to group 16 words into 4(...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- **Confirmed Group 1 (Fast Movement):** {FLY, DASH, BOLT, DART} - Strong consensus on(...TRUNCATED) | [["- **Confirmed Group 1 (Fast Movement):** {FLY, DASH, BOLT, DART} - Strong consensus on \"Fast\" o(...TRUNCATED) | ["- **Status:** Puzzle fully solved with 4 confirmed groups.\n- **Fast Movement:** {BOLT, DASH, DART(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- **Solved/Confirmed Groups**: {FLY, DASH, BOLT, DART} (Fast movement/speed); {BALL, K(...TRUNCATED) | [["- **Solved/Confirmed Groups**: {FLY, DASH, BOLT, DART} (Fast movement/speed); {BALL, KICK, BLAST,(...TRUNCATED) | ["- **Solver Status**: Puzzle fully solved; all four categories correctly identified with perfect gr(...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["and Exploitation.\n- Multi-Agent and Cooperative.\n- Swarm Intelligence and Ant Colony.\n- Geneti(...TRUNCATED) | [
[
0,
1,
0,
1
],
[
1,
0,
0.1,
0
],
[
0,
0,
1,
0
],
[
0.1,
1,
1,
1
]
] | [
[
0,
0.25,
0,
0.25
],
[
0.25,
0,
0.025,
0
],
[
0,
0,
0.25,
0
],
[
0.025,
0.25,
0.25,
0.25
]
] | [
[
0,
10,
0,
10
],
[
10,
0,
1,
0
],
[
0,
0,
10,
0
],
[
1,
10,
10,
10
]
] | [["<think>\nWe need to see what is the candidate hint. The candidate hint list is a long list, refer(...TRUNCATED) | [[[0.0],[1.0],[1.0],[1.0],[1.0]],[[0.0],[1.0],[1.0],[1.0],[1.0]],[[0.0],[1.0],[1.0],[1.0],[1.0]],[[1(...TRUNCATED) | [[[0.0],[4.0],[4.0],[4.0],[4.0]],[[0.0],[4.0],[4.0],[4.0],[4.0]],[[0.0],[4.0],[4.0],[4.0],[4.0]],[[4(...TRUNCATED) | [[["<think>\nWe need to parse candidate answer. The conversation: Candidate answer includes a lot of(...TRUNCATED) | 0.825 | 1 | [
0.25,
0.875,
1,
1,
1
] | 0.45 | 1 | 0.1125 | [
0.275,
0.5,
0.525,
0.5
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
"+-------+-----------+-------+---------+\n| SNAG | SACK | COLOR | ATTEMPT |\n+-------+--------(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"HAIR SALON OFFERINGS\", \"words\": [\"BLOWOUT\", \"COLOR(...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 452","date":"2024-09-05T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 5 | ["Consider the category: MEASURED IN FOOTBALL STATS as a potential category for the words in the gri(...TRUNCATED) | [["Maybe \"ARRANGE\"? No.\nMaybe \"SET\"? No.\nMaybe \"PLACE\"? No.\nMaybe \"POSITION\"? No.\nMaybe (...TRUNCATED) | [["<think>\nThe solver has submitted the exact same grouping as their previous attempt, which is con(...TRUNCATED) | [[["HARSH CRITICISM -> PAN, SLAM, BASH, BLAST | FOOTBALL TERMS -> SACK, SCORE, YARD, RECEPTION | SAL(...TRUNCATED) | [[["<think>\nI'm looking at a 4x4 grid of 16 words and need to sort them into four categories of fou(...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- Current hypothesis: 4 groups identified as \"HARSH CRITICISM\" (PAN, SLAM, BASH, BLA(...TRUNCATED) | [["- Current hypothesis: 4 groups identified as \"HARSH CRITICISM\" (PAN, SLAM, BASH, BLAST), \"FOOT(...TRUNCATED) | ["- **Status**: Puzzle fully solved; no unresolved groups or errors remain.\n- **Confirmed Groups**:(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- Solver attempted 4 categories: HARSH CRITICISM (PAN, SLAM, BASH, BLAST), FOOTBALL TE(...TRUNCATED) | [["- Solver attempted 4 categories: HARSH CRITICISM (PAN, SLAM, BASH, BLAST), FOOTBALL TERMS (SACK, (...TRUNCATED) | ["- **Status**: Puzzle fully solved; all four categories correctly identified in the latest attempt.(...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["Maybe \"ARRANGE\"? No.\nMaybe \"SET\"? No.\nMaybe \"PLACE\"? No.\nMaybe \"POSITION\"? No.\nMaybe (...TRUNCATED) | [
[
0,
0.6,
1,
1
],
[
1,
0.9,
1,
0
],
[
1,
0,
0.1,
0
],
[
1,
1,
0,
1
]
] | [
[
0,
0.15,
0.25,
0.25
],
[
0.25,
0.225,
0.25,
0
],
[
0.25,
0,
0.025,
0
],
[
0.25,
0.25,
0,
0.25
]
] | [
[
0,
6,
10,
10
],
[
10,
9,
10,
0
],
[
10,
0,
1,
0
],
[
10,
10,
0,
10
]
] | [["<think>\nWe have expert hints: categories: MEASURED IN FOOTBALL STATS, HAIR SALON OFFERINGS, OBTA(...TRUNCATED) | [[[0.75],[1.0],[1.0],[1.0],[1.0]],[[1.0],[1.0],[1.0],[1.0],[1.0]],[[1.0],[1.0],[1.0],[1.0],[1.0]],[[(...TRUNCATED) | [[[3.0],[4.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[4.0],[4.0],[4.0]],[[4(...TRUNCATED) | [[["<think>\nWe need to evaluate candidate answer groups. Candidate output:\n\n\"Harsh Criticism -> (...TRUNCATED) | 0.9875 | 1 | [
0.9375,
1,
1,
1,
1
] | 0.6 | 1 | 0.15 | [
0.75,
0.625,
0.525,
0.5
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
"+--------+---------+-------+-------+\n| FLAT | SPEAKER | MIC | WHIP |\n+--------+---------+---(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"GET EXCITED, WITH \\\"UP\\\"\", \"words\": [\"AMP\", \"F(...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 464","date":"2024-09-17T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 6 | ["Consider the category: LEGISLATIVE ROLES as a potential category for the words in the grid.","For (...TRUNCATED) | [["HYPER? HYPER ACTIVE, HYPER REALITY.\n What about **HYPER** as a prefix for **FIRE**? HYPERFIRE(...TRUNCATED) | [["<think>\nThe solver is currently stuck in a deep loop of guessing synonyms for abstract concepts (...TRUNCATED) | [[["PURPOSE\"? No.\nMaybe \"NEXT PURPOSE\"? No.\nMaybe \"THIS MEANING\"? No.\nMaybe \"NEXT MEANING\"(...TRUNCATED) | [[["<think>\nThe first thing I notice about this puzzle is the set of 16 words provided in the grid:(...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- Board: 4x4 grid with words FLAT, SPEAKER, MIC, WHIP, AMP, DUG, PEAT, SLIDE, PSYCH, C(...TRUNCATED) | [["- Board: 4x4 grid with words FLAT, SPEAKER, MIC, WHIP, AMP, DUG, PEAT, SLIDE, PSYCH, CHAIR, MULE,(...TRUNCATED) | ["- **Status**: Puzzle fully solved and confirmed; all four groups are correct with no remaining amb(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- Solver is stuck in a \"synonym/definition\" loop, testing abstract concepts like \"P(...TRUNCATED) | [["<summary>\n- Solver is stuck in a \"synonym/definition\" loop, testing abstract concepts like \"P(...TRUNCATED) | ["- Solver successfully submitted all four correct categories on the first attempt.\n- Groups identi(...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["HYPER? HYPER ACTIVE, HYPER REALITY.\n What about **HYPER** as a prefix for **FIRE**? HYPERFIRE(...TRUNCATED) | [
[
0.8,
0,
0,
0.3
],
[
0,
0,
1,
0
],
[
1,
0,
0,
0.2
],
[
0,
0,
1,
1
]
] | [
[
0.2,
0,
0,
0.075
],
[
0,
0,
0.25,
0
],
[
0.25,
0,
0,
0.05
],
[
0,
0,
0.25,
0.25
]
] | [
[
8,
0,
0,
3
],
[
0,
0,
10,
0
],
[
10,
0,
0,
2
],
[
0,
0,
10,
10
]
] | [["<think>\nWe need to evaluate the candidate hint relative to expert hints. The expert hints:\n\n1.(...TRUNCATED) | [[[0.0],[1.0],[1.0],[1.0],[1.0]],[[1.0],[1.0],[1.0],[1.0],[1.0]],[[0.0],[0.25],[0.0],[0.0],[0.0]],[[(...TRUNCATED) | [[[0.0],[4.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[4.0],[4.0],[4.0]],[[0.0],[1.0],[0.0],[0.0],[0.0]],[[4(...TRUNCATED) | [[["<think>\nThe candidate answer is a long nonsense attempt; seems to not provide any groups. There(...TRUNCATED) | 0.7125 | 1 | [
0.5,
0.8125,
0.75,
0.75,
0.75
] | 0.33125 | 1 | 0.082813 | [
0.45,
0,
0.5,
0.375
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
"+-------+-------+---------+-------+\n| PANT | SHORT | TOM | BUCK |\n+-------+-------+--------(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"THINGS TO PAY\", \"words\": [\"BILL\", \"CHECK\", \"INVO(...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 217","date":"2024-01-14T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 7 | ["Consider the category: THIEVE as a potential category for the words in the grid.","For the categor(...TRUNCATED) | [["<board>\n+-------+-------+---------+-------+\n| PANT | SHORT | TOM | BUCK |\n+-------+-----(...TRUNCATED) | [["<board>\n+-------+-------+---------+-------+\n| PANT | SHORT | TOM | BUCK |\n+-------+-----(...TRUNCATED) | [[["BULL (Group 4).\n \n Counts per row:\n Row 1: 1,1,1,1.\n Row 2: 1,1,1,1.\n R(...TRUNCATED) | [[["<think>\nThe goal is to solve a Connections-style puzzle. I have 16 words arranged in a 4x4 grid(...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- **Confirmed Groups**: \n 1. TYPES OF PANTS: JEAN, PANT, SHORT, TIGHT\n 2. WORDS FO(...TRUNCATED) | [["- **Confirmed Groups**: \n 1. TYPES OF PANTS: JEAN, PANT, SHORT, TIGHT\n 2. WORDS FOR BILL: BIL(...TRUNCATED) | ["- **Status**: Puzzle fully solved and verified; no further action required.\n- **Confirmed Groups*(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- Solver successfully identified all four categories: Legwear (PANT, SHORT, JEAN, TIGH(...TRUNCATED) | [["- Solver successfully identified all four categories: Legwear (PANT, SHORT, JEAN, TIGHT), Bills ((...TRUNCATED) | ["- Solver correctly identified all four categories: Legwear (Pant, Jean, Short, Tight), Bills (Bill(...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["<board>\n+-------+-------+---------+-------+\n| PANT | SHORT | TOM | BUCK |\n+-------+-----(...TRUNCATED) | [
[
0.2,
1,
0,
1
],
[
0.6,
1,
1,
1
],
[
1,
1,
1,
0.4
],
[
0.9,
0.9,
0,
0
]
] | [
[
0.05,
0.25,
0,
0.25
],
[
0.15,
0.25,
0.25,
0.25
],
[
0.25,
0.25,
0.25,
0.1
],
[
0.225,
0.225,
0,
0
]
] | [
[
2,
10,
0,
10
],
[
6,
10,
10,
10
],
[
10,
10,
10,
4
],
[
9,
9,
0,
0
]
] | [["<think>\nWe have to judge the candidate hint versus the expert hints. The candidate hint is basic(...TRUNCATED) | [[[1.0],[1.0],[1.0],[1.0],[1.0]],[[1.0],[1.0],[1.0],[1.0],[1.0]],[[1.0],[1.0],[0.0],[1.0],[1.0]],[[1(...TRUNCATED) | [[[4.0],[4.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[0.0],[4.0],[4.0]],[[4(...TRUNCATED) | [[["<think>\nWe need to evaluate the candidate answer. The puzzle: ground-truth groups:\n\n1. THINGS(...TRUNCATED) | 0.95 | 1 | [
1,
1,
0.75,
1,
1
] | 0.6875 | 1 | 0.171875 | [
0.675,
0.975,
0.5,
0.6
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
"+---------+----------+---------+---------+\n| GHOST | CHERRIES | MOUSE | SLIPPER |\n+---------+(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"GOLDILOCKS\", \"words\": [\"BEAR\", \"BED\", \"CHAIR\", (...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 873","date":"2025-10-31T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 8 | ["Consider the category: POPEYE as a potential category for the words in the grid.","For the categor(...TRUNCATED) | [["The word \"BEAR\" is likely part of a specific story rather than a general category of animals; c(...TRUNCATED) | [["<think>\nThe solver is currently stuck in a loop of random associations, specifically fixating on(...TRUNCATED) | [[["HP treat).\n... What is the 4th?\nCAP (Sorting Hat is a hat, but maybe referred to as a \"cap\" (...TRUNCATED) | [[["<think>\nThe user wants me to solve a Connections-style puzzle.\nI have a 4x4 grid of words:\n1.(...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- **Puzzle Board**: 4x4 grid containing GHOST, CHERRIES, MOUSE, SLIPPER, CAP, WAND, PI(...TRUNCATED) | [["- **Puzzle Board**: 4x4 grid containing GHOST, CHERRIES, MOUSE, SLIPPER, CAP, WAND, PIPE, PUMPKIN(...TRUNCATED) | ["- **Confirmed Groups**: GOLDILOCKS (BEAR, BED, CHAIR, PORRIDGE) and CINDERELLA (SLIPPER, PUMPKIN, (...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- Solver is actively testing **Harry Potter** (GHOST, CHERRIES, MOUSE, SLIPPER, CAP, W(...TRUNCATED) | [["Solver is exploring **Types of Bear** (BEAR) but the list of associations is unfocused and yields(...TRUNCATED) | ["<summary>\n- **Solver's Latest Attempt**: Submitted 4 groups (Goldilocks, Cinderella, Video Games,(...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["The word \"BEAR\" is likely part of a specific story rather than a general category of animals; c(...TRUNCATED) | [
[
1,
0,
0.3,
0
],
[
0,
0,
0,
0
],
[
0,
0,
0,
0
],
[
0,
0,
1,
0.1
]
] | [
[
0.25,
0,
0.075,
0
],
[
0,
0,
0,
0
],
[
0,
0,
0,
0
],
[
0,
0,
0.25,
0.025
]
] | [
[
10,
0,
3,
0
],
[
0,
0,
0,
0
],
[
0,
0,
0,
0
],
[
0,
0,
10,
1
]
] | [["<think>\nWe're comparing candidate hint: \"The word 'BEAR' is likely part of a specific story rat(...TRUNCATED) | [[[0.0],[0.25],[0.25],[0.5],[0.5]],[[0.5],[0.5],[0.0],[0.5],[0.5]],[[1.0],[0.0],[0.0],[1.0],[0.0]],[(...TRUNCATED) | [[[0.0],[1.0],[1.0],[2.0],[2.0]],[[2.0],[2.0],[0.0],[2.0],[2.0]],[[4.0],[0.0],[0.0],[4.0],[0.0]],[[2(...TRUNCATED) | [[["<think>\nWe need to analyze candidate answer: the answer content is huge but mostly scratch work(...TRUNCATED) | 0.4 | 1 | [
0.5,
0.3125,
0.1875,
0.625,
0.375
] | 0.15 | 1 | 0.0375 | [
0.25,
0,
0.325,
0.025
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
"+----------+-------------+-----------+------------+\n| STONER | FOREST | SURFER | MOVIE E(...TRUNCATED) | {"ground_truth":"[{\"answerDescription\": \"CALIFORNIA-BASED CHARACTER TROPES\", \"words\": [\"MOVIE(...TRUNCATED) | NYT-Connections-Processed | {"contest":"NYT Connections 929","date":"2025-12-26T00:00:00","generated_articles":null,"generated_p(...TRUNCATED) | 9 | ["Consider the category: CALIFORNIA-BASED CHARACTER TROPES as a potential category for the words in (...TRUNCATED) | [["Since you have successfully identified all four categories with this attempt, there is no need fo(...TRUNCATED) | [["<hint>\nSince you have successfully identified all four categories with this attempt, there is no(...TRUNCATED) | [[["BLACK ____ -> FOREST, FRIDAY, PANTHER, WIDOW | CLASSIC ROCK BANDS -> AMERICA, CHICAGO, FOREIGNER(...TRUNCATED) | [[["<think>\nThe goal is to solve a Connections puzzle. I have 16 words arranged in a 4x4 grid. I ne(...TRUNCATED) | [[[{"role":"user","content":"You are solving a New York Times Connections-style puzzle.\n\nThe block(...TRUNCATED) | [[[{"role":"system","content":"You are an expert hint generator for New York Times Connections-style(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for a solver that receives sequentia(...TRUNCATED) | [["<summary>\n- Board hypothesis: 4 groups identified (BLACK ____, CLASSIC ROCK BANDS, TYPES OF COME(...TRUNCATED) | [["- Board hypothesis: 4 groups identified (BLACK ____, CLASSIC ROCK BANDS, TYPES OF COMEDY, SOCIAL (...TRUNCATED) | ["- Puzzle status: Fully solved; all four groups correctly identified on the first attempt.\n- Group(...TRUNCATED) | [[[{"role":"user","content":"You are maintaining compact memory for an expert coach that gives seque(...TRUNCATED) | [["<summary>\n- Solver successfully identified all four categories on first try: \"Black ____\" (For(...TRUNCATED) | [["- Solver successfully identified all four categories on first try: \"Black ____\" (Forest, Friday(...TRUNCATED) | ["- Solver successfully identified all four categories (Classic Rock Bands, Black ____, Types of Com(...TRUNCATED) | [
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
],
[
[],
[],
[],
[]
]
] | [
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
],
[
0,
0,
0,
0,
0
]
] | [["Since you have successfully identified all four categories with this attempt, there is no need fo(...TRUNCATED) | [
[
0,
0,
1,
0
],
[
1,
0,
1,
0
],
[
0.8,
0.2,
1,
0.7
],
[
0.8,
0.9,
1,
1
]
] | [
[
0,
0,
0.25,
0
],
[
0.25,
0,
0.25,
0
],
[
0.2,
0.05,
0.25,
0.175
],
[
0.2,
0.225,
0.25,
0.25
]
] | [
[
0,
0,
10,
0
],
[
10,
0,
10,
0
],
[
8,
2,
10,
7
],
[
8,
9,
10,
10
]
] | [["<think>\nWe have candidate hint: \"Since you have successfully identified all four categories wit(...TRUNCATED) | [[[1.0],[1.0],[1.0],[1.0],[1.0]],[[1.0],[1.0],[1.0],[1.0],[1.0]],[[1.0],[1.0],[1.0],[1.0],[1.0]],[[1(...TRUNCATED) | [[[4.0],[4.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[4.0],[4.0],[4.0]],[[4.0],[4.0],[4.0],[4.0],[4.0]],[[4(...TRUNCATED) | [[["<think>\nWe need to evaluate the candidate answer. Provided candidate:\n\n> BLACK ____ -> FOREST(...TRUNCATED) | 0.975 | 1 | [
1,
0.875,
1,
1,
1
] | 0.5875 | 1 | 0.146875 | [
0.65,
0.275,
1,
0.425
] | {"max_steps":4,"num_rollouts":4,"num_solutions":1,"batch_solver_solutions":true,"response_serializat(...TRUNCATED) | [["You are judging how similar a candidate Connections hint is to a set of expert hints.\n\nExpert h(...TRUNCATED) | [[["You are judging whether a candidate answer correctly solves a Connections-style word puzzle.\n\n(...TRUNCATED) |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 98