open-llm-leaderboard-old/details_KnutJaegersberg__Deita-1_8B
收藏数据集概述
数据集简介
该数据集是在评估模型 KnutJaegersberg/Deita-1_8B 在 Open LLM Leaderboard 上的自动创建的。数据集包含 63 个配置,每个配置对应一个评估任务。
数据集结构
数据集从 1 次运行中创建,每个运行可以在每个配置中作为一个特定的分割找到,分割名称使用运行的时间戳。"train" 分割总是指向最新的结果。
额外配置
一个额外的配置 "results" 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。
最新结果
以下是 2024-01-17T18:22:52.012956 运行的最新结果:
python { "all": { "acc": 0.4513695685616451, "acc_stderr": 0.03473174413713779, "acc_norm": 0.4572369573723266, "acc_norm_stderr": 0.035511285827617124, "mc1": 0.2521419828641371, "mc1_stderr": 0.015201522246299979, "mc2": 0.4002214148044727, "mc2_stderr": 0.014908452990717655 }, "harness|arc:challenge|25": { "acc": 0.32081911262798635, "acc_stderr": 0.013640943091946522, "acc_norm": 0.3651877133105802, "acc_norm_stderr": 0.014070265519268802 }, "harness|hellaswag|10": { "acc": 0.4574785899223262, "acc_stderr": 0.004971704917267752, "acc_norm": 0.6062537343158734, "acc_norm_stderr": 0.004875812021461993 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.3, "acc_stderr": 0.046056618647183814, "acc_norm": 0.3, "acc_norm_stderr": 0.046056618647183814 }, # 其他任务的结果省略... }
配置详情
数据集包含多个配置,每个配置对应不同的任务和数据文件。以下是部分配置的示例:
-
config_name: harness_arc_challenge_25
- 数据文件:
- 分割: 2024_01_17T18_22_52.012956
- 路径: **/details_harness|arc:challenge|25_2024-01-17T18-22-52.012956.parquet
- 分割: latest
- 路径: **/details_harness|arc:challenge|25_2024-01-17T18-22-52.012956.parquet
- 分割: 2024_01_17T18_22_52.012956
- 数据文件:
-
config_name: harness_gsm8k_5
- 数据文件:
- 分割: 2024_01_17T18_22_52.012956
- 路径: **/details_harness|gsm8k|5_2024-01-17T18-22-52.012956.parquet
- 分割: latest
- 路径: **/details_harness|gsm8k|5_2024-01-17T18-22-52.012956.parquet
- 分割: 2024_01_17T18_22_52.012956
- 数据文件:
-
config_name: harness_hellaswag_10
- 数据文件:
- 分割: 2024_01_17T18_22_52.012956
- 路径: **/details_harness|hellaswag|10_2024-01-17T18-22-52.012956.parquet
- 分割: latest
- 路径: **/details_harness|hellaswag|10_2024-01-17T18-22-52.012956.parquet
- 分割: 2024_01_17T18_22_52.012956
- 数据文件:
-
config_name: harness_hendrycksTest_5
- 数据文件:
- 分割: 2024_01_17T18_22_52.012956
- 路径:
- **/details_harness|hendrycksTest-abstract_algebra|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-anatomy|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-astronomy|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-business_ethics|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-clinical_knowledge|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-college_biology|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-college_chemistry|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-college_computer_science|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-college_mathematics|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-college_medicine|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-college_physics|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-computer_security|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-conceptual_physics|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-econometrics|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-electrical_engineering|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-elementary_mathematics|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-formal_logic|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-global_facts|5_2024-01-17T18-22-52.012956.parquet
- **/details_harness|hendrycksTest-high_school_biology|5_2024-01-17T18-22-52.012956.parquet
其他文件路径省略...
- 路径:
- 分割: 2024_01_17T18_22_52.012956
- 数据文件:
}



