open-llm-leaderboard/details_vicgalle__Configurable-Llama-3-8B-v0.2
收藏数据集概述
数据集简介
该数据集是在对模型 vicgalle/Configurable-Llama-3-8B-v0.2 进行评估运行时自动创建的,用于 Open LLM Leaderboard 上的评估。
数据集组成
- 数据集包含 63 个配置,每个配置对应一个评估任务。
- 数据集由 1 次运行创建,每个运行在每个配置中都有一个特定的分割,分割名称使用运行的时间戳。
- 每个配置中的 "train" 分割始终指向最新的结果。
- 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 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_vicgalle__Configurable-Llama-3-8B-v0.2", "harness_winogrande_5", split="train")
最新结果
以下是 2024-04-19T03:19:20.040365 运行的最新结果:
python { "all": { "acc": 0.6694408893619185, "acc_stderr": 0.03176677993984921, "acc_norm": 0.6716394002581934, "acc_norm_stderr": 0.03240469776002056, "mc1": 0.37821297429620565, "mc1_stderr": 0.01697633590754687, "mc2": 0.5679248273257846, "mc2_stderr": 0.015218379312142727 }, "harness|arc:challenge|25": { "acc": 0.5836177474402731, "acc_stderr": 0.014405618279436169, "acc_norm": 0.6254266211604096, "acc_norm_stderr": 0.014144193471893454 }, "harness|hellaswag|10": { "acc": 0.5952997410874328, "acc_stderr": 0.004898308167211848, "acc_norm": 0.7977494523003386, "acc_norm_stderr": 0.004008571431483689 }, "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... }
配置详情
-
config_name: harness_arc_challenge_25
- 数据文件:
- 分割: 2024_04_19T03_19_20.040365
- 路径:
**/details_harness|arc:challenge|25_2024-04-19T03-19-20.040365.parquet
- 路径:
- 分割: latest
- 路径:
**/details_harness|arc:challenge|25_2024-04-19T03-19-20.040365.parquet
- 路径:
- 分割: 2024_04_19T03_19_20.040365
- 数据文件:
-
config_name: harness_gsm8k_5
- 数据文件:
- 分割: 2024_04_19T03_19_20.040365
- 路径:
**/details_harness|gsm8k|5_2024-04-19T03-19-20.040365.parquet
- 路径:
- 分割: latest
- 路径:
**/details_harness|gsm8k|5_2024-04-19T03-19-20.040365.parquet
- 路径:
- 分割: 2024_04_19T03_19_20.040365
- 数据文件:
-
config_name: harness_hellaswag_10
- 数据文件:
- 分割: 2024_04_19T03_19_20.040365
- 路径:
**/details_harness|hellaswag|10_2024-04-19T03-19-20.040365.parquet
- 路径:
- 分割: latest
- 路径:
**/details_harness|hellaswag|10_2024-04-19T03-19-20.040365.parquet
- 路径:
- 分割: 2024_04_19T03_19_20.040365
- 数据文件:
-
config_name: harness_hendrycksTest_5
- 数据文件:
- 分割: 2024_04_19T03_19_20.040365
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-anatomy|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-astronomy|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-business_ethics|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-college_biology|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-college_chemistry|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-college_computer_science|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-college_mathematics|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-college_medicine|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-college_physics|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-computer_security|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-econometrics|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-electrical_engineering|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-elementary_mathematics|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-formal_logic|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-global_facts|5_2024-04-19T03-19-20.040365.parquet**/details_harness|hendrycksTest-high_school_biology|5_2024-04-19T03-19-20.040365.parquet
其他任务的路径...
- 路径:
- 分割: 2024_04_19T03_19_20.040365
- 数据文件:
}



