| { |
| "results": { |
| "omnis-voluptatibus_lsat-rc_cot": { |
| "acc,none": 0.26394052044609667, |
| "acc_stderr,none": 0.026924155643902548, |
| "alias": "omnis-voluptatibus_lsat-rc_cot" |
| }, |
| "omnis-voluptatibus_lsat-lr_cot": { |
| "acc,none": 0.2196078431372549, |
| "acc_stderr,none": 0.018349383611423235, |
| "alias": "omnis-voluptatibus_lsat-lr_cot" |
| }, |
| "omnis-voluptatibus_lsat-ar_cot": { |
| "acc,none": 0.21739130434782608, |
| "acc_stderr,none": 0.027256850838819964, |
| "alias": "omnis-voluptatibus_lsat-ar_cot" |
| }, |
| "omnis-voluptatibus_logiqa_cot": { |
| "acc,none": 0.29073482428115016, |
| "acc_stderr,none": 0.018164056209177798, |
| "alias": "omnis-voluptatibus_logiqa_cot" |
| }, |
| "omnis-voluptatibus_logiqa2_cot": { |
| "acc,none": 0.29834605597964375, |
| "acc_stderr,none": 0.011543394639779811, |
| "alias": "omnis-voluptatibus_logiqa2_cot" |
| }, |
| "magnam-eius_lsat-rc_cot": { |
| "acc,none": 0.27137546468401486, |
| "acc_stderr,none": 0.027162503089239516, |
| "alias": "magnam-eius_lsat-rc_cot" |
| }, |
| "magnam-eius_lsat-lr_cot": { |
| "acc,none": 0.2196078431372549, |
| "acc_stderr,none": 0.018349383611423218, |
| "alias": "magnam-eius_lsat-lr_cot" |
| }, |
| "magnam-eius_lsat-ar_cot": { |
| "acc,none": 0.2, |
| "acc_stderr,none": 0.02643274401820356, |
| "alias": "magnam-eius_lsat-ar_cot" |
| }, |
| "magnam-eius_logiqa_cot": { |
| "acc,none": 0.29073482428115016, |
| "acc_stderr,none": 0.018164056209177805, |
| "alias": "magnam-eius_logiqa_cot" |
| }, |
| "magnam-eius_logiqa2_cot": { |
| "acc,none": 0.3142493638676845, |
| "acc_stderr,none": 0.011712031200512734, |
| "alias": "magnam-eius_logiqa2_cot" |
| }, |
| "libero-exercitationem_lsat-rc_cot": { |
| "acc,none": 0.26022304832713755, |
| "acc_stderr,none": 0.02680130130545777, |
| "alias": "libero-exercitationem_lsat-rc_cot" |
| }, |
| "libero-exercitationem_lsat-lr_cot": { |
| "acc,none": 0.20980392156862746, |
| "acc_stderr,none": 0.018047429112476115, |
| "alias": "libero-exercitationem_lsat-lr_cot" |
| }, |
| "libero-exercitationem_lsat-ar_cot": { |
| "acc,none": 0.24347826086956523, |
| "acc_stderr,none": 0.02836109930007507, |
| "alias": "libero-exercitationem_lsat-ar_cot" |
| }, |
| "libero-exercitationem_logiqa_cot": { |
| "acc,none": 0.3210862619808307, |
| "acc_stderr,none": 0.018675754307572432, |
| "alias": "libero-exercitationem_logiqa_cot" |
| }, |
| "libero-exercitationem_logiqa2_cot": { |
| "acc,none": 0.30916030534351147, |
| "acc_stderr,none": 0.0116598352236769, |
| "alias": "libero-exercitationem_logiqa2_cot" |
| }, |
| "illum-eaque_lsat-rc_cot": { |
| "acc,none": 0.26765799256505574, |
| "acc_stderr,none": 0.027044545314587293, |
| "alias": "illum-eaque_lsat-rc_cot" |
| }, |
| "illum-eaque_lsat-lr_cot": { |
| "acc,none": 0.23137254901960785, |
| "acc_stderr,none": 0.018691965462419545, |
| "alias": "illum-eaque_lsat-lr_cot" |
| }, |
| "illum-eaque_lsat-ar_cot": { |
| "acc,none": 0.2, |
| "acc_stderr,none": 0.026432744018203558, |
| "alias": "illum-eaque_lsat-ar_cot" |
| }, |
| "illum-eaque_logiqa_cot": { |
| "acc,none": 0.29233226837060705, |
| "acc_stderr,none": 0.018193366406024092, |
| "alias": "illum-eaque_logiqa_cot" |
| }, |
| "illum-eaque_logiqa2_cot": { |
| "acc,none": 0.294529262086514, |
| "acc_stderr,none": 0.011500471190116962, |
| "alias": "illum-eaque_logiqa2_cot" |
| }, |
| "amet-ullam_lsat-rc_cot": { |
| "acc,none": 0.2788104089219331, |
| "acc_stderr,none": 0.02739124797571039, |
| "alias": "amet-ullam_lsat-rc_cot" |
| }, |
| "amet-ullam_lsat-lr_cot": { |
| "acc,none": 0.20980392156862746, |
| "acc_stderr,none": 0.018047429112476105, |
| "alias": "amet-ullam_lsat-lr_cot" |
| }, |
| "amet-ullam_lsat-ar_cot": { |
| "acc,none": 0.2, |
| "acc_stderr,none": 0.02643274401820355, |
| "alias": "amet-ullam_lsat-ar_cot" |
| }, |
| "amet-ullam_logiqa_cot": { |
| "acc,none": 0.29233226837060705, |
| "acc_stderr,none": 0.018193366406024095, |
| "alias": "amet-ullam_logiqa_cot" |
| }, |
| "amet-ullam_logiqa2_cot": { |
| "acc,none": 0.2951653944020356, |
| "acc_stderr,none": 0.011507692175964774, |
| "alias": "amet-ullam_logiqa2_cot" |
| }, |
| "accusantium-inventore_lsat-rc_cot": { |
| "acc,none": 0.26765799256505574, |
| "acc_stderr,none": 0.027044545314587293, |
| "alias": "accusantium-inventore_lsat-rc_cot" |
| }, |
| "accusantium-inventore_lsat-lr_cot": { |
| "acc,none": 0.2, |
| "acc_stderr,none": 0.01772968828711749, |
| "alias": "accusantium-inventore_lsat-lr_cot" |
| }, |
| "accusantium-inventore_lsat-ar_cot": { |
| "acc,none": 0.20869565217391303, |
| "acc_stderr,none": 0.026854108265439658, |
| "alias": "accusantium-inventore_lsat-ar_cot" |
| }, |
| "accusantium-inventore_logiqa_cot": { |
| "acc,none": 0.3274760383386581, |
| "acc_stderr,none": 0.01877170136786437, |
| "alias": "accusantium-inventore_logiqa_cot" |
| }, |
| "accusantium-inventore_logiqa2_cot": { |
| "acc,none": 0.30470737913486007, |
| "acc_stderr,none": 0.01161280687039332, |
| "alias": "accusantium-inventore_logiqa2_cot" |
| } |
| }, |
| "configs": { |
| "accusantium-inventore_logiqa2_cot": { |
| "task": "accusantium-inventore_logiqa2_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "accusantium-inventore-logiqa2/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "accusantium-inventore_logiqa_cot": { |
| "task": "accusantium-inventore_logiqa_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "accusantium-inventore-logiqa/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "accusantium-inventore_lsat-ar_cot": { |
| "task": "accusantium-inventore_lsat-ar_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "accusantium-inventore-lsat-ar/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "accusantium-inventore_lsat-lr_cot": { |
| "task": "accusantium-inventore_lsat-lr_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "accusantium-inventore-lsat-lr/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "accusantium-inventore_lsat-rc_cot": { |
| "task": "accusantium-inventore_lsat-rc_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "accusantium-inventore-lsat-rc/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "amet-ullam_logiqa2_cot": { |
| "task": "amet-ullam_logiqa2_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "amet-ullam-logiqa2/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "amet-ullam_logiqa_cot": { |
| "task": "amet-ullam_logiqa_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "amet-ullam-logiqa/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "amet-ullam_lsat-ar_cot": { |
| "task": "amet-ullam_lsat-ar_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "amet-ullam-lsat-ar/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "amet-ullam_lsat-lr_cot": { |
| "task": "amet-ullam_lsat-lr_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "amet-ullam-lsat-lr/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "amet-ullam_lsat-rc_cot": { |
| "task": "amet-ullam_lsat-rc_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "amet-ullam-lsat-rc/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "illum-eaque_logiqa2_cot": { |
| "task": "illum-eaque_logiqa2_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "illum-eaque-logiqa2/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "illum-eaque_logiqa_cot": { |
| "task": "illum-eaque_logiqa_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "illum-eaque-logiqa/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "illum-eaque_lsat-ar_cot": { |
| "task": "illum-eaque_lsat-ar_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "illum-eaque-lsat-ar/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "illum-eaque_lsat-lr_cot": { |
| "task": "illum-eaque_lsat-lr_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "illum-eaque-lsat-lr/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "illum-eaque_lsat-rc_cot": { |
| "task": "illum-eaque_lsat-rc_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "illum-eaque-lsat-rc/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "libero-exercitationem_logiqa2_cot": { |
| "task": "libero-exercitationem_logiqa2_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "libero-exercitationem-logiqa2/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "libero-exercitationem_logiqa_cot": { |
| "task": "libero-exercitationem_logiqa_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "libero-exercitationem-logiqa/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "libero-exercitationem_lsat-ar_cot": { |
| "task": "libero-exercitationem_lsat-ar_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "libero-exercitationem-lsat-ar/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "libero-exercitationem_lsat-lr_cot": { |
| "task": "libero-exercitationem_lsat-lr_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "libero-exercitationem-lsat-lr/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "libero-exercitationem_lsat-rc_cot": { |
| "task": "libero-exercitationem_lsat-rc_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "libero-exercitationem-lsat-rc/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "magnam-eius_logiqa2_cot": { |
| "task": "magnam-eius_logiqa2_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "magnam-eius-logiqa2/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "magnam-eius_logiqa_cot": { |
| "task": "magnam-eius_logiqa_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "magnam-eius-logiqa/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "magnam-eius_lsat-ar_cot": { |
| "task": "magnam-eius_lsat-ar_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "magnam-eius-lsat-ar/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "magnam-eius_lsat-lr_cot": { |
| "task": "magnam-eius_lsat-lr_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "magnam-eius-lsat-lr/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "magnam-eius_lsat-rc_cot": { |
| "task": "magnam-eius_lsat-rc_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "magnam-eius-lsat-rc/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "omnis-voluptatibus_logiqa2_cot": { |
| "task": "omnis-voluptatibus_logiqa2_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "omnis-voluptatibus-logiqa2/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "omnis-voluptatibus_logiqa_cot": { |
| "task": "omnis-voluptatibus_logiqa_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "omnis-voluptatibus-logiqa/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "omnis-voluptatibus_lsat-ar_cot": { |
| "task": "omnis-voluptatibus_lsat-ar_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "omnis-voluptatibus-lsat-ar/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "omnis-voluptatibus_lsat-lr_cot": { |
| "task": "omnis-voluptatibus_lsat-lr_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "omnis-voluptatibus-lsat-lr/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| }, |
| "omnis-voluptatibus_lsat-rc_cot": { |
| "task": "omnis-voluptatibus_lsat-rc_cot", |
| "group": "logikon-bench", |
| "dataset_path": "logikon/cot-eval-traces", |
| "dataset_kwargs": { |
| "data_files": { |
| "test": "omnis-voluptatibus-lsat-rc/test-00000-of-00001.parquet" |
| } |
| }, |
| "test_split": "test", |
| "doc_to_text": "def doc_to_text_cot(doc) -> str:\n \"\"\"\n Answer the following question about the given passage. [Base your answer on the reasoning below.]\n \n Passage: <passage>\n \n Question: <question>\n A. <choice1>\n B. <choice2>\n C. <choice3>\n D. <choice4>\n [E. <choice5>]\n \n [Reasoning: <reasoning>]\n \n Answer:\n \"\"\"\n k = len(doc[\"options\"])\n choices = [\"a\", \"b\", \"c\", \"d\", \"e\"][:k]\n prompt = \"Answer the following question about the given passage. Base your answer on the reasoning below.\\n\\n\"\n prompt = \"Passage: \" + doc[\"passage\"] + \"\\n\\n\"\n prompt += \"Question: \" + doc[\"question\"] + \"\\n\"\n for choice, option in zip(choices, doc[\"options\"]):\n prompt += f\"{choice.upper()}. {option}\\n\"\n prompt += \"\\n\"\n prompt += \"Reasoning: \" + doc[\"reasoning_trace\"] + \"\\n\\n\" \n prompt += \"Answer:\"\n return prompt\n", |
| "doc_to_target": "{{answer}}", |
| "doc_to_choice": "{{options}}", |
| "description": "", |
| "target_delimiter": " ", |
| "fewshot_delimiter": "\n\n", |
| "num_fewshot": 0, |
| "metric_list": [ |
| { |
| "metric": "acc", |
| "aggregation": "mean", |
| "higher_is_better": true |
| } |
| ], |
| "output_type": "multiple_choice", |
| "repeats": 1, |
| "should_decontaminate": false, |
| "metadata": { |
| "version": 0.0 |
| } |
| } |
| }, |
| "versions": { |
| "accusantium-inventore_logiqa2_cot": 0.0, |
| "accusantium-inventore_logiqa_cot": 0.0, |
| "accusantium-inventore_lsat-ar_cot": 0.0, |
| "accusantium-inventore_lsat-lr_cot": 0.0, |
| "accusantium-inventore_lsat-rc_cot": 0.0, |
| "amet-ullam_logiqa2_cot": 0.0, |
| "amet-ullam_logiqa_cot": 0.0, |
| "amet-ullam_lsat-ar_cot": 0.0, |
| "amet-ullam_lsat-lr_cot": 0.0, |
| "amet-ullam_lsat-rc_cot": 0.0, |
| "illum-eaque_logiqa2_cot": 0.0, |
| "illum-eaque_logiqa_cot": 0.0, |
| "illum-eaque_lsat-ar_cot": 0.0, |
| "illum-eaque_lsat-lr_cot": 0.0, |
| "illum-eaque_lsat-rc_cot": 0.0, |
| "libero-exercitationem_logiqa2_cot": 0.0, |
| "libero-exercitationem_logiqa_cot": 0.0, |
| "libero-exercitationem_lsat-ar_cot": 0.0, |
| "libero-exercitationem_lsat-lr_cot": 0.0, |
| "libero-exercitationem_lsat-rc_cot": 0.0, |
| "magnam-eius_logiqa2_cot": 0.0, |
| "magnam-eius_logiqa_cot": 0.0, |
| "magnam-eius_lsat-ar_cot": 0.0, |
| "magnam-eius_lsat-lr_cot": 0.0, |
| "magnam-eius_lsat-rc_cot": 0.0, |
| "omnis-voluptatibus_logiqa2_cot": 0.0, |
| "omnis-voluptatibus_logiqa_cot": 0.0, |
| "omnis-voluptatibus_lsat-ar_cot": 0.0, |
| "omnis-voluptatibus_lsat-lr_cot": 0.0, |
| "omnis-voluptatibus_lsat-rc_cot": 0.0 |
| }, |
| "n-shot": { |
| "accusantium-inventore_logiqa2_cot": 0, |
| "accusantium-inventore_logiqa_cot": 0, |
| "accusantium-inventore_lsat-ar_cot": 0, |
| "accusantium-inventore_lsat-lr_cot": 0, |
| "accusantium-inventore_lsat-rc_cot": 0, |
| "amet-ullam_logiqa2_cot": 0, |
| "amet-ullam_logiqa_cot": 0, |
| "amet-ullam_lsat-ar_cot": 0, |
| "amet-ullam_lsat-lr_cot": 0, |
| "amet-ullam_lsat-rc_cot": 0, |
| "illum-eaque_logiqa2_cot": 0, |
| "illum-eaque_logiqa_cot": 0, |
| "illum-eaque_lsat-ar_cot": 0, |
| "illum-eaque_lsat-lr_cot": 0, |
| "illum-eaque_lsat-rc_cot": 0, |
| "libero-exercitationem_logiqa2_cot": 0, |
| "libero-exercitationem_logiqa_cot": 0, |
| "libero-exercitationem_lsat-ar_cot": 0, |
| "libero-exercitationem_lsat-lr_cot": 0, |
| "libero-exercitationem_lsat-rc_cot": 0, |
| "magnam-eius_logiqa2_cot": 0, |
| "magnam-eius_logiqa_cot": 0, |
| "magnam-eius_lsat-ar_cot": 0, |
| "magnam-eius_lsat-lr_cot": 0, |
| "magnam-eius_lsat-rc_cot": 0, |
| "omnis-voluptatibus_logiqa2_cot": 0, |
| "omnis-voluptatibus_logiqa_cot": 0, |
| "omnis-voluptatibus_lsat-ar_cot": 0, |
| "omnis-voluptatibus_lsat-lr_cot": 0, |
| "omnis-voluptatibus_lsat-rc_cot": 0 |
| }, |
| "config": { |
| "model": "vllm", |
| "model_args": "pretrained=01-ai/Yi-6B,revision=main,dtype=auto,tensor_parallel_size=1,gpu_memory_utilization=0.9,trust_remote_code=true,max_length=4096", |
| "batch_size": "auto", |
| "batch_sizes": [], |
| "device": null, |
| "use_cache": null, |
| "limit": null, |
| "bootstrap_iters": 100000, |
| "gen_kwargs": null |
| }, |
| "git_hash": "5044cf9" |
| } |
