open-llm-leaderboard-old/details_ConvexAI__Metabird-7B
收藏数据集概述
数据集简介
该数据集是在对模型 ConvexAI/Metabird-7B 进行评估运行期间自动创建的,用于 Open LLM Leaderboard。
数据集结构
- 配置数量:63个配置,每个配置对应一个评估任务。
- 创建来源:数据集来自1次运行,每个运行结果作为一个特定的分割(split)存储,分割名称使用运行的时间戳。
- 最新结果:"train" 分割始终指向最新的结果。
- 聚合结果:一个额外的配置 "results" 存储所有运行的聚合结果,用于计算和显示在 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_ConvexAI__Metabird-7B", "harness_winogrande_5", split="train")
最新结果
以下是 2024-01-20T17:15:55.821392 运行的最新结果:
python { "all": { "acc": 0.6562196818092919, "acc_stderr": 0.031888120668163725, "acc_norm": 0.6572328630765639, "acc_norm_stderr": 0.03253749439366192, "mc1": 0.40758873929008566, "mc1_stderr": 0.01720194923455311, "mc2": 0.579354979250579, "mc2_stderr": 0.015314953137607924 }, "harness|arc:challenge|25": { "acc": 0.6680887372013652, "acc_stderr": 0.013760988200880541, "acc_norm": 0.6953924914675768, "acc_norm_stderr": 0.01344952210993249 }, "harness|hellaswag|10": { "acc": 0.6965743875721968, "acc_stderr": 0.0045879786255824785, "acc_norm": 0.8754232224656443, "acc_norm_stderr": 0.0032956349076664645 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.33, "acc_stderr": 0.047258156262526045, "acc_norm": 0.33, "acc_norm_stderr": 0.047258156262526045 }, ... }
配置详情
-
配置名称:harness_arc_challenge_25
- 数据文件:
- 分割:2024_01_20T17_15_55.821392
- 路径:
**/details_harness|arc:challenge|25_2024-01-20T17-15-55.821392.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2024-01-20T17-15-55.821392.parquet
- 路径:
- 分割:2024_01_20T17_15_55.821392
- 数据文件:
-
配置名称:harness_gsm8k_5
- 数据文件:
- 分割:2024_01_20T17_15_55.821392
- 路径:
**/details_harness|gsm8k|5_2024-01-20T17-15-55.821392.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|gsm8k|5_2024-01-20T17-15-55.821392.parquet
- 路径:
- 分割:2024_01_20T17_15_55.821392
- 数据文件:
-
配置名称:harness_hellaswag_10
- 数据文件:
- 分割:2024_01_20T17_15_55.821392
- 路径:
**/details_harness|hellaswag|10_2024-01-20T17-15-55.821392.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|hellaswag|10_2024-01-20T17-15-55.821392.parquet
- 路径:
- 分割:2024_01_20T17_15_55.821392
- 数据文件:
-
配置名称:harness_hendrycksTest_5
- 数据文件:
- 分割:2024_01_20T17_15_55.821392
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-anatomy|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-astronomy|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-business_ethics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_biology|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_chemistry|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_computer_science|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_mathematics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_medicine|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_physics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-computer_security|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-econometrics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-electrical_engineering|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-elementary_mathematics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-formal_logic|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-global_facts|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-high_school_biology|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-high_school_chemistry|5_2024-01-20T17-15-55.821392.parquet- ...
- 路径:
- 分割:latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-anatomy|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-astronomy|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-business_ethics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_biology|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_chemistry|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_computer_science|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_mathematics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_medicine|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-college_physics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-computer_security|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-econometrics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-electrical_engineering|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-elementary_mathematics|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-formal_logic|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-global_facts|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-high_school_biology|5_2024-01-20T17-15-55.821392.parquet**/details_harness|hendrycksTest-high_school_chemistry|5_2024-01-20T17-15-55.821392.parquet- ... }
- 路径:
- 分割:2024_01_20T17_15_55.821392
- 数据文件:



