open-llm-leaderboard-old/details_microsoft__Orca-2-13b
收藏数据集概述
数据集摘要
该数据集是在评估模型microsoft/Orca-2-13b在Open LLM Leaderboard上的运行过程中自动创建的。数据集包含64个配置,每个配置对应一个评估任务。数据集从2次运行中创建,每次运行可以在每个配置中找到特定的分割,分割名称使用运行的时间戳。"train"分割始终指向最新的结果。
数据集结构
数据集由以下配置组成:
- harness_arc_challenge_25
- harness_drop_3
- harness_gsm8k_5
- harness_hellaswag_10
- harness_hendrycksTest_5
每个配置包含多个分割,每个分割对应不同的运行时间戳,例如:
2023_11_23T09_00_59.7743772023_12_30T00_44_18.166149latest
数据加载示例
python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_microsoft__Orca-2-13b", "harness_winogrande_5", split="train")
最新结果
最新结果来自2023-12-30T00:44:18.166149的运行,包含多个任务的评估结果,例如: python { "all": { "acc": 0.601679092820444, "acc_stderr": 0.03296876808787226, "acc_norm": 0.6064308784221981, "acc_norm_stderr": 0.03364034807631641, "mc1": 0.3990208078335373, "mc1_stderr": 0.017142825728496767, "mc2": 0.5642038222037025, "mc2_stderr": 0.01593463688746652 }, "harness|arc:challenge|25": { "acc": 0.5742320819112628, "acc_stderr": 0.014449464278868802, "acc_norm": 0.6092150170648464, "acc_norm_stderr": 0.014258563880513778 }, ... }
数据文件路径
每个配置包含多个数据文件路径,例如:
-
harness_arc_challenge_25
**/details_harness|arc:challenge|25_2023-11-23T09-00-59.774377.parquet**/details_harness|arc:challenge|25_2023-12-30T00-44-18.166149.parquet**/details_harness|arc:challenge|25_2023-12-30T00-44-18.166149.parquet
-
harness_drop_3
**/details_harness|drop|3_2023-11-23T09-00-59.774377.parquet**/details_harness|drop|3_2023-11-23T09-00-59.774377.parquet
-
harness_gsm8k_5
**/details_harness|gsm8k|5_2023-11-23T09-00-59.774377.parquet**/details_harness|gsm8k|5_2023-12-30T00-44-18.166149.parquet**/details_harness|gsm8k|5_2023-12-30T00-44-18.166149.parquet
-
harness_hellaswag_10
**/details_harness|hellaswag|10_2023-11-23T09-00-59.774377.parquet**/details_harness|hellaswag|10_2023-12-30T00-44-18.166149.parquet**/details_harness|hellaswag|10_2023-12-30T00-44-18.166149.parquet
-
harness_hendrycksTest_5
**/details_harness|hendrycksTest-abstract_algebra|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-anatomy|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-astronomy|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-business_ethics|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-college_biology|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-college_chemistry|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-college_computer_science|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-college_mathematics|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-college_medicine|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-college_physics|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-computer_security|5_2023-11-23T09-00-59.774377.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2023-11-23T09-00-59.774377.parquet... }



