[
{
"name": "ops14_csharp_bug",
"n_prompt": 713,
"n_gen": 128,
"target_preview": "<think>Let me think carefully about whether my fix is correct and complete.\n\nThe bug is: DOTNET_RESOURCE_LEAK in `InitReader()` method.\n- Leaked resource: `_pip"
},
{
"name": "ops17_term_multiturn",
"n_prompt": 1186,
"n_gen": 128,
"target_preview": "<think>Let me verify everything is in order:\n\n1. ✅ `/output/command_capture.txt` exists and contains the captured output in XYZ format:\n - Line 1: `3` (atom c"
},
{
"name": "ops19_term",
"n_prompt": 489,
"n_gen": 128,
"target_preview": "<think>Let me analyze the task:\n\n1. Read a fixed base date from a prepared file\n2. Print that base date\n3. Compute and output the calendar date exactly 90 days "
},
{
"name": "ops12_java_turn1",
"n_prompt": 244,
"n_gen": 128,
"target_preview": "<think>Let me verify the file is correctly in place before confirming task completion. The file was created at `/app/src/main/java/com/github/tobato/fastdfs/pro"
},
{
"name": "cj03_toolcall",
"n_prompt": 951,
"n_gen": 128,
"target_preview": "<think>\nThe working directory is /tmp/claude-1000/-home-sbuser-Repos--claude-worktrees-epic-hopper-8358b4/8af2c4cf-769c-4a73-9587-137fc7480293/scratchpad/bake/w"
},
{
"name": "cj20_tool_5turn_t3",
"n_prompt": 588,
"n_gen": 128,
"target_preview": "<think>\nSimple task. Modify `render_amount` to accept an optional keyword argument `symbol` (defaulting to `'$'`). Update the docstring. Add a test. Run pytest."
},
{
"name": "cj15_php_review",
"n_prompt": 1153,
"n_gen": 128,
"target_preview": "<think>\nWe need answer user's question. Need analyze files. User asks: In files above, which file defines top-level function format_directions, and what are its"
},
{
"name": "cj19_italian",
"n_prompt": 27,
"n_gen": 128,
"target_preview": "<think>\nWe need answer in Italian. User asks: \"Spiegami come funziona un modello GPT, considerando che di deep learning ne capisco poco\". Need explain simply. N"
},
{
"name": "cj06_dostoevsky",
"n_prompt": 26,
"n_gen": 128,
"target_preview": "<think>\nWe need to answer user: \"Can you summarize Dostoevsky's \"The House of the Dead\" novel?\" Need likely correct title: \"The House of the Dead Men\"? Actually"
},
{
"name": "cj07_json_extract_near",
"n_prompt": 163,
"n_gen": 128,
"target_preview": "<think>\nWe need answer user's request. Need produce final only JSON object no code fence. Need parse text: Invoice INV-7261 (EUR). Lines:\n1 x label printer at 3"
},
{
"name": "cj17_math",
"n_prompt": 43,
"n_gen": 128,
"target_preview": "<think>\nWe need answer math problem. Need solve step by step. Problem: For how many three-digit positive integers is sum of digits equal to 5? Three-digit posit"
},
{
"name": "ja05",
"n_prompt": 154,
"n_gen": 128,
"target_preview": "<think>\nまず、ユーザーが求めているのは、与えられた文章から仮定条件を含む重要なキーワードを抽出し、それらを使って検索に適した質問文と回答を作成することです。まず、文章の内容を理解する必要があります。研究テーマは微生物の機能解析で、食品製造に使われる微生物のルーツとして植物に注目し、未開拓の分野であること。大学の"
},
{
"name": "ja12_table",
"n_prompt": 370,
"n_gen": 128,
"target_preview": "<think>\nまず、与えられた文章から重要なキーワードを抽出する必要があります。文章の主な内容は関西国際空港のバリアフリー施設やサービスに関する情報です。例えば「車いす対応」「オストメイト」「ベビーベッド」といったトイレの設備や、「車いす対応エレベータ」「エスカレータ」などの移動に関するキーワードがあります。また、介"
},
{
"name": "ja22_pedal",
"n_prompt": 398,
"n_gen": 128,
"target_preview": "<think>\nまず、ユーザーの要求を正確に理解する必要があります。問題文では、BM25検索でマッチしにくくするために、原文の単語をなるべく使わず同義語や言い換え表現を使ったキーワードを5〜10個抽出し、それらを用いた質問文を作成する必要があります。さらに、その質問に対する回答を文章から抽出し、JSON形式で出力します"
},
{
"name": "ja30_climate",
"n_prompt": 568,
"n_gen": 128,
"target_preview": "<think>\nまず、文章から重要なキーワードを抽出する必要があります。反語表現は使われていないようですが、修辞的な表現を探します。GCF（グリーン気候基金）、COP15、COP16、仁川、理事会、緩和策、適応策、拠出表明額、トランプ政権、世界銀行などが目立ちます。特に主要拠出国や金額、設立の経緯、基金の使い道が重要そ"
},
{
"name": "persona0_t2",
"n_prompt": 458,
"n_gen": 128,
"target_preview": "<think>\nthe user wants exact systemd `.service` and `.timer` unit files. this is a format/register override: the unit file contents themselves must be clean, va"
},
{
"name": "persona3_t4",
"n_prompt": 1861,
"n_gen": 128,
"target_preview": "<think>\nlet's write out the exact 2D rotary block identity $R(\theta_{p + \\Delta}) = R(\theta_\\Delta) R(\theta_p)$ and then explain the difference between keys (wh"
},
{
"name": "cj07_answer_json",
"n_prompt": 697,
"n_gen": 104,
"target_preview": "\n\n{\"invoice\":\"INV-7261\",\"currency\":\"EUR\",\"lines\":[{\"item\":\"label printer\",\"qty\":1,\"unit_price\":325.71},{\"item\":\"standing desk\",\"qty\":4,\"unit_price\":356.91},{\"it"
},
{
"name": "cj20_t3_answer_tool",
"n_prompt": 629,
"n_gen": 128,
"target_preview": "\n\n<tool_call>\n<function=Write>\n<parameter=file_path>\n/workspace/proj00170/money.py\n</parameter>\n<parameter=content>\ndef render_amount(value, symbol=\"$\"):\n \"\""
},
{
"name": "ja05_answer",
"n_prompt": 632,
"n_gen": 22,
"target_preview": "\n微生物の生育可能限界の条件を40パターンほど設計して網羅的に検討を行いました。"
},
{
"name": "persona0_t2_answer",
"n_prompt": 531,
"n_gen": 128,
"target_preview": "\nclassic `.direnv` hoarding lol. here are the two units — `/etc/systemd/system/nix-prune-old-roots.service` removes `result*` and `.direnv` gcroots older than 1"
},
{
"name": "ops19_answer_json",
"n_prompt": 644,
"n_gen": 128,
"target_preview": "\n\n\n{\n \"analysis\": \"I'm at the /workspace directory. I need to explore what files exist, find a prepared file with a base date, and then compute the date 90 day"
}
]