open-llm-leaderboard-old/details_cerebras__Cerebras-GPT-111M
收藏数据集概述
数据集简介
该数据集是在评估模型 cerebras/Cerebras-GPT-111M 在 Open LLM Leaderboard 上的自动创建的。数据集包含 64 个配置,每个配置对应一个评估任务。
数据集结构
数据集由 2 次运行创建,每次运行可以在每个配置中找到特定的分割,分割名称使用运行的时间戳。"train" 分割始终指向最新的结果。
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。
数据加载示例
以下是加载特定运行详细信息的示例代码: python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_cerebras__Cerebras-GPT-111M", "harness_winogrande_5", split="train")
最新结果
以下是 2023-09-22T19:15:45.776483 运行的最新结果: python { "all": { "em": 0.0010486577181208054, "em_stderr": 0.00033145814652193176, "f1": 0.021427223154362497, "f1_stderr": 0.0008720566428263053, "acc": 0.23875295974743488, "acc_stderr": 0.00701912891202994 }, "harness|drop|3": { "em": 0.0010486577181208054, "em_stderr": 0.00033145814652193176, "f1": 0.021427223154362497, "f1_stderr": 0.0008720566428263053 }, "harness|gsm8k|5": { "acc": 0.0, "acc_stderr": 0.0 }, "harness|winogrande|5": { "acc": 0.47750591949486976, "acc_stderr": 0.01403825782405988 } }
配置详情
以下是数据集的配置详情:
配置列表
harness_arc_challenge_25harness_drop_3harness_gsm8k_5harness_hellaswag_10harness_hendrycksTest_5harness_hendrycksTest_abstract_algebra_5harness_hendrycksTest_anatomy_5harness_hendrycksTest_astronomy_5harness_hendrycksTest_business_ethics_5harness_hendrycksTest_clinical_knowledge_5harness_hendrycksTest_college_biology_5harness_hendrycksTest_college_chemistry_5harness_hendrycksTest_college_computer_science_5harness_hendrycksTest_college_mathematics_5harness_hendrycksTest_college_medicine_5harness_hendrycksTest_college_physics_5harness_hendrycksTest_computer_security_5harness_hendrycksTest_conceptual_physics_5harness_hendrycksTest_econometrics_5harness_hendrycksTest_electrical_engineering_5harness_hendrycksTest_elementary_mathematics_5
数据文件路径
每个配置包含多个数据文件路径,具体路径和分割名称如下:
harness_arc_challenge_25
- 分割:
2023_07_19T13_47_12.878137- 路径:
**/details_harness|arc:challenge|25_2023-07-19T13:47:12.878137.parquet
- 路径:
- 分割:
latest- 路径:
**/details_harness|arc:challenge|25_2023-07-19T13:47:12.878137.parquet
- 路径:
harness_drop_3
- 分割:
2023_09_22T19_15_45.776483- 路径:
**/details_harness|drop|3_2023-09-22T19-15-45.776483.parquet
- 路径:
- 分割:
latest- 路径:
**/details_harness|drop|3_2023-09-22T19-15-45.776483.parquet
- 路径:
harness_gsm8k_5
- 分割:
2023_09_22T19_15_45.776483- 路径:
**/details_harness|gsm8k|5_2023-09-22T19-15-45.776483.parquet
- 路径:
- 分割:
latest- 路径:
**/details_harness|gsm8k|5_2023-09-22T19-15-45.776483.parquet
- 路径:
harness_hellaswag_10
- 分割:
2023_07_19T13_47_12.878137- 路径:
**/details_harness|hellaswag|10_2023-07-19T13:47:12.878137.parquet
- 路径:
- 分割:
latest- 路径:
**/details_harness|hellaswag|10_2023-07-19T13:47:12.878137.parquet
- 路径:
harness_hendrycksTest_5
- 分割:
2023_07_19T13_47_12.878137- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-formal_logic|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-global_facts|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_biology|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_chemistry|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_computer_science|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_european_history|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_geography|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_mathematics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_physics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_psychology|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_statistics|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_us_history|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-high_school_world_history|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-human_aging|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-human_sexuality|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-international_law|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-jurisprudence|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-logical_fallacies|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-machine_learning|5_2023-07-19T13:47:12.878137.parquet - 路径:
**/details_harness|hendrycksTest-management|5_2023-07-19T13:47:12.878137.parquet - 路径:`**/details_harness|hendrycksTest-marketing|5_2023-07-19T13:47:1
- 路径:



