open-llm-leaderboard/details_EleutherAI__gpt-j-6b
收藏Hugging Face2023-12-03 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_EleutherAI__gpt-j-6b
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是在评估模型EleutherAI/gpt-j-6b时自动创建的,包含122个配置,每个配置对应一个评估任务。数据集由8次运行生成,每次运行的结果作为特定的分割存储在配置中。此外,数据集还包含一个名为"results"的配置,用于存储所有运行的聚合结果。
提供机构:
open-llm-leaderboard
原始信息汇总
数据集概述
数据集简介
该数据集是在对模型 EleutherAI/gpt-j-6b 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:122个配置,每个配置对应一个评估任务。
- 运行次数:数据集由8次运行创建,每次运行在每个配置中作为一个特定的分片存在,分片名称使用运行的时间戳。
- 最新结果:"train" 分片始终指向最新的结果。
- 汇总结果:一个额外的配置 "results" 存储所有运行的汇总结果,用于计算和显示 Open LLM Leaderboard 上的汇总指标。
数据加载示例
python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_EleutherAI__gpt-j-6b", "harness_gsm8k_5", split="train")
最新结果
以下是 2023-12-03T16:52:26.173919 运行的最新结果: python { "all": { "acc": 0.029567854435178165, "acc_stderr": 0.004665893134220772 }, "harness|gsm8k|5": { "acc": 0.029567854435178165, "acc_stderr": 0.004665893134220772 } }
配置详情
以下是部分配置及其数据文件路径:
配置:harness_arc_challenge_25
- 分片:2023_07_18T11_28_34.858547
- 路径:
**/details_harness|arc:challenge|25_2023-07-18T11:28:34.858547.parquet
- 路径:
- 分片:2023_07_19T10_47_17.854530
- 路径:
**/details_harness|arc:challenge|25_2023-07-19T10:47:17.854530.parquet
- 路径:
- 分片:2023_08_29T19_41_28.653242
- 路径:
**/details_harness|arc:challenge|25_2023-08-29T19:41:28.653242.parquet
- 路径:
- 分片:2023_09_21T23_10_49.133869
- 路径:
**/details_harness|arc:challenge|25_2023-09-21T23-10-49.133869.parquet
- 路径:
- 分片:latest
- 路径:
**/details_harness|arc:challenge|25_2023-09-21T23-10-49.133869.parquet
- 路径:
配置:harness_drop_3
- 分片:2023_09_08T17_46_12.907701
- 路径:
**/details_harness|drop|3_2023-09-08T17-46-12.907701.parquet
- 路径:
- 分片:latest
- 路径:
**/details_harness|drop|3_2023-09-08T17-46-12.907701.parquet
- 路径:
配置:harness_gsm8k_5
- 分片:2023_09_08T17_46_12.907701
- 路径:
**/details_harness|gsm8k|5_2023-09-08T17-46-12.907701.parquet
- 路径:
- 分片:2023_12_03T16_51_11.923338
- 路径:
**/details_harness|gsm8k|5_2023-12-03T16-51-11.923338.parquet
- 路径:
- 分片:2023_12_03T16_52_26.173919
- 路径:
**/details_harness|gsm8k|5_2023-12-03T16-52-26.173919.parquet
- 路径:
- 分片:latest
- 路径:
**/details_harness|gsm8k|5_2023-12-03T16-52-26.173919.parquet
- 路径:
配置:harness_hellaswag_10
- 分片:2023_07_18T11_28_34.858547
- 路径:
**/details_harness|hellaswag|10_2023-07-18T11:28:34.858547.parquet
- 路径:
- 分片:2023_07_19T10_47_17.854530
- 路径:
**/details_harness|hellaswag|10_2023-07-19T10:47:17.854530.parquet
- 路径:
- 分片:2023_08_29T19_41_28.653242
- 路径:
**/details_harness|hellaswag|10_2023-08-29T19:41:28.653242.parquet
- 路径:
- 分片:2023_09_21T23_10_49.133869
- 路径:
**/details_harness|hellaswag|10_2023-09-21T23-10-49.133869.parquet
- 路径:
- 分片:latest
- 路径:
**/details_harness|hellaswag|10_2023-09-21T23-10-49.133869.parquet
- 路径:
配置:harness_hendrycksTest_5
- 分片:2023_07_18T11_28_34.858547
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-formal_logic|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-global_facts|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_biology|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_chemistry|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_computer_science|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_european_history|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_geography|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_mathematics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_physics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_psychology|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_statistics|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_us_history|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-high_school_world_history|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-human_aging|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-human_sexuality|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-international_law|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-jurisprudence|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-logical_fallacies|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-machine_learning|5_2023-07-18T11:28:34.858547.parquet - 路径:
**/details_harness|hendrycksTest-management|5_2023-07-18T11:28:34.858547.parquet - 路径:`**/details_harness|hendrycksTest-marketing|5_2023-07-
- 路径:



