open-llm-leaderboard-old/details_NousResearch__Nous-Hermes-Llama2-13b
收藏数据集概述
数据集简介
该数据集是在对模型 NousResearch/Nous-Hermes-Llama2-13b 进行评估运行期间自动创建的,用于 Open LLM Leaderboard。
数据集结构
- 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 6 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。
- 额外的 "results" 配置存储所有运行的聚合结果,用于计算和显示 Open LLM Leaderboard 上的聚合指标。
数据加载示例
python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_NousResearch__Nous-Hermes-Llama2-13b", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-21T23:27:15.868927 运行的最新结果: python { "all": { "em": 0.22934144295302014, "em_stderr": 0.004305384313236111, "f1": 0.30605285234899415, "f1_stderr": 0.004296224150122663, "acc": 0.4276861222626263, "acc_stderr": 0.010194652064655127 }, "harness|drop|3": { "em": 0.22934144295302014, "em_stderr": 0.004305384313236111, "f1": 0.30605285234899415, "f1_stderr": 0.004296224150122663 }, "harness|gsm8k|5": { "acc": 0.10083396512509477, "acc_stderr": 0.008294031192126607 }, "harness|winogrande|5": { "acc": 0.7545382794001578, "acc_stderr": 0.012095272937183647 } }
配置详情
以下是数据集的配置详情:
配置名称:harness_arc_challenge_25
- 分割:2023_07_24T14_44_05.322938
- 路径:
**/details_harness|arc:challenge|25_2023-07-24T14:44:05.322938.parquet
- 路径:
- 分割:2023_07_25T11_02_46.466402
- 路径:
**/details_harness|arc:challenge|25_2023-07-25T11:02:46.466402.parquet
- 路径:
- 分割:2023_07_26T14_55_06.636628
- 路径:
**/details_harness|arc:challenge|25_2023-07-26T14:55:06.636628.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2023-07-26T14:55:06.636628.parquet
- 路径:
配置名称:harness_drop_3
- 分割:2023_10_21T20_14_41.722716
- 路径:
**/details_harness|drop|3_2023-10-21T20-14-41.722716.parquet
- 路径:
- 分割:2023_10_21T21_17_49.044019
- 路径:
**/details_harness|drop|3_2023-10-21T21-17-49.044019.parquet
- 路径:
- 分割:2023_10_21T23_27_15.868927
- 路径:
**/details_harness|drop|3_2023-10-21T23-27-15.868927.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|drop|3_2023-10-21T23-27-15.868927.parquet
- 路径:
配置名称:harness_gsm8k_5
- 分割:2023_10_21T20_14_41.722716
- 路径:
**/details_harness|gsm8k|5_2023-10-21T20-14-41.722716.parquet
- 路径:
- 分割:2023_10_21T21_17_49.044019
- 路径:
**/details_harness|gsm8k|5_2023-10-21T21-17-49.044019.parquet
- 路径:
- 分割:2023_10_21T23_27_15.868927
- 路径:
**/details_harness|gsm8k|5_2023-10-21T23-27-15.868927.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|gsm8k|5_2023-10-21T23-27-15.868927.parquet
- 路径:
配置名称:harness_hellaswag_10
- 分割:2023_07_24T14_44_05.322938
- 路径:
**/details_harness|hellaswag|10_2023-07-24T14:44:05.322938.parquet
- 路径:
- 分割:2023_07_25T11_02_46.466402
- 路径:
**/details_harness|hellaswag|10_2023-07-25T11:02:46.466402.parquet
- 路径:
- 分割:2023_07_26T14_55_06.636628
- 路径:
**/details_harness|hellaswag|10_2023-07-26T14:55:06.636628.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|hellaswag|10_2023-07-26T14:55:06.636628.parquet
- 路径:
配置名称:harness_hendrycksTest_5
- 分割:2023_07_24T14_44_05.322938
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-anatomy|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-astronomy|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-business_ethics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-college_biology|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-college_chemistry|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-college_computer_science|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-college_mathematics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-college_medicine|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-college_physics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-computer_security|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-econometrics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-formal_logic|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-global_facts|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_biology|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_chemistry|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_computer_science|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_european_history|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_geography|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_mathematics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_physics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_psychology|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_statistics|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_us_history|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-high_school_world_history|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-human_aging|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-human_sexuality|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-international_law|5_2023-07-24T14:44:05.322938.parquet**/details_harness|hendrycksTest-jurisprudence|5_2023-07-24T14:44:05.322938.parquet- `**/details_harness|hendrycks
- 路径:



