olmoe-dclm-metrics / humaneval-OLMoE-7B-A1B-step1223842-tokens5100B.json
dwadden's picture
Upload 8 files
3389d64 verified
{
"name": "/net/nfs.cirrascale/allennlp/davidw/proj/OLMoE/script/humaneval",
"uuid": "3a11e770-7075-42f3-81f6-c14a6db4953d",
"model": "/net/nfs.cirrascale/allennlp/davidw/checkpoints/moe-release/OLMoE-7B-A1B/step1223842-tokens5100B",
"creation_date": "2024_08_06-02_21_03",
"eval_metrics": {
"icl": {
"human_eval": 0.012195121496915817,
"human_eval_cpp": 0.1304347813129425,
"human_eval_js": 0.0,
"human_eval_return_simple": 0.8648648858070374,
"human_eval_return_complex": 0.14960630238056183,
"human_eval_25": 0.04268292710185051,
"human_eval_50": 0.09146341681480408,
"human_eval_75": 0.16463415324687958
}
},
"missing tasks": "['mmlu_zeroshot', 'hellaswag_zeroshot', 'jeopardy', 'triviaqa_sm_sub', 'gsm8k_cot', 'agi_eval_sat_math_cot', 'aqua_cot', 'svamp_cot', 'bigbench_qa_wikidata', 'arc_easy', 'arc_challenge', 'mmlu_fewshot', 'bigbench_misconceptions', 'copa', 'siqa', 'commonsense_qa', 'piqa', 'openbook_qa', 'bigbench_novel_concepts', 'bigbench_strange_stories', 'bigbench_strategy_qa', 'lambada_openai', 'hellaswag', 'winograd', 'winogrande', 'bigbench_conlang_translation', 'bigbench_language_identification', 'bigbench_conceptual_combinations', 'bigbench_elementary_math_qa', 'bigbench_dyck_languages', 'agi_eval_lsat_ar', 'bigbench_cs_algorithms', 'bigbench_logical_deduction', 'bigbench_operators', 'bigbench_repeat_copy_logic', 'simple_arithmetic_nospaces', 'simple_arithmetic_withspaces', 'math_qa', 'logi_qa', 'pubmed_qa_labeled', 'squad', 'agi_eval_lsat_rc', 'agi_eval_lsat_lr', 'coqa', 'bigbench_understanding_fables', 'boolq', 'agi_eval_sat_en', 'winogender_mc_female', 'winogender_mc_male', 'enterprise_pii_classification', 'bbq', 'gpqa_main', 'gpqa_diamond']",
"aggregated_task_categories_centered": {
"commonsense reasoning": NaN,
"language understanding": NaN,
"reading comprehension": NaN,
"safety": NaN,
"symbolic problem solving": NaN,
"world knowledge": NaN
},
"aggregated_centered_results": NaN,
"aggregated_results": NaN,
"rw_small": NaN,
"rw_small_centered": NaN,
"95%_CI_above": NaN,
"95%_CI_above_centered": NaN,
"99%_CI_above": NaN,
"99%_CI_above_centered": NaN,
"low_variance_datasets": NaN,
"low_variance_datasets_centered": NaN
}