open-llm-leaderboard/details_BELLE-2__BELLE-Llama2-13B-chat-0.4M
收藏数据集概述
数据集名称
Evaluation run of BELLE-2/BELLE-Llama2-13B-chat-0.4M
数据集来源
该数据集是在模型 BELLE-2/BELLE-Llama2-13B-chat-0.4M 在 Open LLM Leaderboard 上的评估运行期间自动创建的。
数据集组成
- 该数据集包含 64 个配置,每个配置对应一个评估任务。
- 数据集从 2 次运行中创建,每次运行可以在每个配置中找到一个特定的拆分,拆分名称使用运行的时间戳。
- "train" 拆分始终指向最新的结果。
- 一个额外的配置 "results" 存储所有运行结果的聚合,用于计算和显示聚合指标在 Open LLM Leaderboard 上。
数据加载示例
python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_BELLE-2__BELLE-Llama2-13B-chat-0.4M", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-26T14:37:54.228887 运行的最新结果: python { "all": { "em": 0.28492030201342283, "em_stderr": 0.004622517599527834, "f1": 0.36695364932886015, "f1_stderr": 0.004514579216323901, "acc": 0.4496880334950361, "acc_stderr": 0.010877118313612513 }, "harness|drop|3": { "em": 0.28492030201342283, "em_stderr": 0.004622517599527834, "f1": 0.36695364932886015, "f1_stderr": 0.004514579216323901 }, "harness|gsm8k|5": { "acc": 0.14404852160727824, "acc_stderr": 0.009672110973065286 }, "harness|winogrande|5": { "acc": 0.755327545382794, "acc_stderr": 0.012082125654159738 } }
配置详情
以下是数据集的配置详情:
-
harness_arc_challenge_25
- 拆分:2023_10_01T13_36_40.123057
- 路径:
**/details_harness|arc:challenge|25_2023-10-01T13-36-40.123057.parquet
- 路径:
- 拆分:latest
- 路径:
**/details_harness|arc:challenge|25_2023-10-01T13-36-40.123057.parquet
- 路径:
- 拆分:2023_10_01T13_36_40.123057
-
harness_drop_3
- 拆分:2023_10_26T14_37_54.228887
- 路径:
**/details_harness|drop|3_2023-10-26T14-37-54.228887.parquet
- 路径:
- 拆分:latest
- 路径:
**/details_harness|drop|3_2023-10-26T14-37-54.228887.parquet
- 路径:
- 拆分:2023_10_26T14_37_54.228887
-
harness_gsm8k_5
- 拆分:2023_10_26T14_37_54.228887
- 路径:
**/details_harness|gsm8k|5_2023-10-26T14-37-54.228887.parquet
- 路径:
- 拆分:latest
- 路径:
**/details_harness|gsm8k|5_2023-10-26T14-37-54.228887.parquet
- 路径:
- 拆分:2023_10_26T14_37_54.228887
-
harness_hellaswag_10
- 拆分:2023_10_01T13_36_40.123057
- 路径:
**/details_harness|hellaswag|10_2023-10-01T13-36-40.123057.parquet
- 路径:
- 拆分:latest
- 路径:
**/details_harness|hellaswag|10_2023-10-01T13-36-40.123057.parquet
- 路径:
- 拆分:2023_10_01T13_36_40.123057
-
harness_hendrycksTest_5
- 拆分:2023_10_01T13_36_40.123057
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-formal_logic|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-global_facts|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_biology|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_chemistry|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_computer_science|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_european_history|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_geography|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_mathematics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_physics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_psychology|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_statistics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_us_history|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-high_school_world_history|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-human_aging|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-human_sexuality|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-international_law|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-jurisprudence|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-logical_fallacies|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-machine_learning|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-management|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-marketing|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-medical_genetics|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-miscellaneous|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-moral_disputes|5_2023-10-01T13-36-40.123057.parquet - 路径:
**/details_harness|hendrycksTest-moral_scenarios|5_2023-10-01T13-36-40.123057.parquet - 路径:`**/details_harness|hendrycksTest-nutrition|5_2023-10-01T13-36-40.12
- 路径:
- 拆分:2023_10_01T13_36_40.123057



