open-llm-leaderboard/details_CHIH-HUNG__llama-2-13b-FINETUNE3_3.3w-r8-gate_up_down
收藏数据集概述
数据集来源
该数据集是在评估模型 CHIH-HUNG/llama-2-13b-FINETUNE3_3.3w-r8-gate_up_down 在 Open LLM Leaderboard 上的运行过程中自动创建的。
数据集结构
- 配置数量:64个配置,每个配置对应一个评估任务。
- 数据来源:数据集由2次运行结果组成,每次运行的结果作为一个特定的分割(split)存储,分割名称使用运行的时间戳。
- 最新结果:"train" 分割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,用于计算和显示在 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_CHIH-HUNG__llama-2-13b-FINETUNE3_3.3w-r8-gate_up_down", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-28T04:59:45.337848 运行 的最新结果: python { "all": { "em": 0.21780620805369127, "em_stderr": 0.004226997205914972, "f1": 0.26442428691275155, "f1_stderr": 0.004202345101166552, "acc": 0.4478647632531162, "acc_stderr": 0.010273220250493906 }, "harness|drop|3": { "em": 0.21780620805369127, "em_stderr": 0.004226997205914972, "f1": 0.26442428691275155, "f1_stderr": 0.004202345101166552 }, "harness|gsm8k|5": { "acc": 0.11751326762699014, "acc_stderr": 0.00887033125649 }, "harness|winogrande|5": { "acc": 0.7782162588792423, "acc_stderr": 0.011676109244497811 } }
配置详情
以下是数据集的配置详情:
-
harness_arc_challenge_25
- 分割:2023_10_01T14_35_45.846264
- 路径:
**/details_harness|arc:challenge|25_2023-10-01T14-35-45.846264.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2023-10-01T14-35-45.846264.parquet
- 路径:
- 分割:2023_10_01T14_35_45.846264
-
harness_drop_3
- 分割:2023_10_28T04_59_45.337848
- 路径:
**/details_harness|drop|3_2023-10-28T04-59-45.337848.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|drop|3_2023-10-28T04-59-45.337848.parquet
- 路径:
- 分割:2023_10_28T04_59_45.337848
-
harness_gsm8k_5
- 分割:2023_10_28T04_59_45.337848
- 路径:
**/details_harness|gsm8k|5_2023-10-28T04-59-45.337848.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|gsm8k|5_2023-10-28T04-59-45.337848.parquet
- 路径:
- 分割:2023_10_28T04_59_45.337848
-
harness_hellaswag_10
- 分割:2023_10_01T14_35_45.846264
- 路径:
**/details_harness|hellaswag|10_2023-10-01T14-35-45.846264.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|hellaswag|10_2023-10-01T14-35-45.846264.parquet
- 路径:
- 分割:2023_10_01T14_35_45.846264
-
harness_hendrycksTest_5
- 分割:2023_10_01T14_35_45.846264
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-formal_logic|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-global_facts|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_biology|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_chemistry|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_computer_science|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_european_history|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_geography|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_mathematics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_physics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_psychology|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_statistics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_us_history|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-high_school_world_history|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-human_aging|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-human_sexuality|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-international_law|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-jurisprudence|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-logical_fallacies|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-machine_learning|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-management|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-marketing|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-medical_genetics|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-miscellaneous|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-moral_disputes|5_2023-10-01T14-35-45.846264.parquet - 路径:
**/details_harness|hendrycksTest-moral_scenarios|5_2023-10-01T14-35-45.846264.parquet - 路径:`**/details_harness|hendrycksTest-nutrition|5_2023-10-01T14-35-45.846264
- 路径:
- 分割:2023_10_01T14_35_45.846264



