open-llm-leaderboard/details_DUAL-GPO-2__phi-2-gpo-renew2-i0
收藏Hugging Face2024-04-24 更新2024-06-12 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_DUAL-GPO-2__phi-2-gpo-renew2-i0
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是在模型DUAL-GPO-2/phi-2-gpo-renew2-i0的评估运行期间自动创建的,用于在Open LLM Leaderboard上进行评估。数据集由63个配置组成,每个配置对应一个评估任务。数据集从1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,分割以运行的时间戳命名。train分割始终指向最新结果。此外,results配置存储了所有运行的聚合结果,并用于计算和显示Open LLM Leaderboard上的聚合指标。
该数据集是在模型DUAL-GPO-2/phi-2-gpo-renew2-i0的评估运行期间自动创建的,用于在Open LLM Leaderboard上进行评估。数据集由63个配置组成,每个配置对应一个评估任务。数据集从1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,分割以运行的时间戳命名。train分割始终指向最新结果。此外,results配置存储了所有运行的聚合结果,并用于计算和显示Open LLM Leaderboard上的聚合指标。
提供机构:
open-llm-leaderboard
原始信息汇总
数据集概述
数据集创建
- 创建背景:该数据集是在模型DUAL-GPO-2/phi-2-gpo-renew2-i0在Open LLM Leaderboard上的评估运行期间自动创建的。
数据集结构
- 配置数量:数据集包含63个配置,每个配置对应一个评估任务。
- 运行次数:数据集从1次运行中创建。每个运行在每个配置中作为一个特定的分割存在,分割名称使用运行的时间戳。"train"分割始终指向最新的结果。
- 结果配置: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。
数据加载示例
python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_DUAL-GPO-2__phi-2-gpo-renew2-i0", "harness_winogrande_5", split="train")
最新结果
- 最新结果时间戳:2024-04-24T09:47:54.342824
- 结果示例:
python
{
"all": {
"acc": 0.5866303999997428,
"acc_stderr": 0.03377278802848554,
"acc_norm": 0.5878804403886916,
"acc_norm_stderr": 0.034464731550254986,
"mc1": 0.32802937576499386,
"mc1_stderr": 0.016435632932815025,
"mc2": 0.4690868059219712,
"mc2_stderr": 0.015249195938194224
},
"harness|arc:challenge|25": {
"acc": 0.5998293515358362,
"acc_stderr": 0.014317197787809169,
"acc_norm": 0.6271331058020477,
"acc_norm_stderr": 0.014131176760131165
},
"harness|hellaswag|10": {
"acc": 0.574088826926907,
"acc_stderr": 0.004934698012050244,
"acc_norm": 0.7625970922127067,
"acc_norm_stderr": 0.004246216229989856
},
"harness|hendrycksTest-abstract_algebra|5": {
"acc": 0.28,
"acc_stderr": 0.045126085985421296,
"acc_norm": 0.28,
"acc_norm_stderr": 0.045126085985421296
},
更多任务结果...
}
配置详情
-
配置名称:harness_arc_challenge_25
- 数据文件:
- 分割:2024_04_24T09_47_54.342824
- 路径:
**/details_harness|arc:challenge|25_2024-04-24T09-47-54.342824.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2024-04-24T09-47-54.342824.parquet
- 路径:
- 分割:2024_04_24T09_47_54.342824
- 数据文件:
-
配置名称:harness_gsm8k_5
- 数据文件:
- 分割:2024_04_24T09_47_54.342824
- 路径:
**/details_harness|gsm8k|5_2024-04-24T09-47-54.342824.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|gsm8k|5_2024-04-24T09-47-54.342824.parquet
- 路径:
- 分割:2024_04_24T09_47_54.342824
- 数据文件:
-
配置名称:harness_hellaswag_10
- 数据文件:
- 分割:2024_04_24T09_47_54.342824
- 路径:
**/details_harness|hellaswag|10_2024-04-24T09-47-54.342824.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|hellaswag|10_2024-04-24T09-47-54.342824.parquet
- 路径:
- 分割:2024_04_24T09_47_54.342824
- 数据文件:
-
配置名称:harness_hendrycksTest_5
- 数据文件:
- 分割:2024_04_24T09_47_54.342824
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-anatomy|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-astronomy|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-business_ethics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-college_biology|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-college_chemistry|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-college_computer_science|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-college_mathematics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-college_medicine|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-college_physics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-computer_security|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-econometrics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-electrical_engineering|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-elementary_mathematics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-formal_logic|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-global_facts|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_biology|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_chemistry|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_computer_science|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_european_history|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_geography|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_government_and_politics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_macroeconomics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_mathematics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_microeconomics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_physics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_psychology|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_statistics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_us_history|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-high_school_world_history|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-human_aging|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-human_sexuality|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-international_law|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-jurisprudence|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-logical_fallacies|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-machine_learning|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-management|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-marketing|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-medical_genetics|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-miscellaneous|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-moral_disputes|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-moral_scenarios|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-nutrition|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-philosophy|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-prehistory|5_2024-04-24T09-47-54.342824.parquet**/details_harness|hendrycksTest-professional_accounting|5_2024-04-24T09-47-54.342824.parquet- `**/details_harness|hendrycksTest-professional_law|5_2024-04-24T09-47-54.3428
- 路径:
- 分割:2024_04_24T09_47_54.342824
- 数据文件:



