open-llm-leaderboard/details_winglian__llama-2-4b
收藏数据集概述
该数据集是在对模型 winglian/llama-2-4b 进行评估运行期间自动创建的,用于 Open LLM Leaderboard。
数据集组成
- 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。
- 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_winglian__llama-2-4b", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-26T09:12:20.462280 运行 的最新结果:
python { "all": { "em": 0.0016778523489932886, "em_stderr": 0.0004191330178826824, "f1": 0.039830117449664484, "f1_stderr": 0.0011176272810803495, "acc": 0.2895672680207256, "acc_stderr": 0.007874406355324997 }, "harness|drop|3": { "em": 0.0016778523489932886, "em_stderr": 0.0004191330178826824, "f1": 0.039830117449664484, "f1_stderr": 0.0011176272810803495 }, "harness|gsm8k|5": { "acc": 0.004548900682335102, "acc_stderr": 0.0018535550440036202 }, "harness|winogrande|5": { "acc": 0.574585635359116, "acc_stderr": 0.013895257666646373 } }
配置详情
-
harness_arc_challenge_25
- 分割: 2023_10_03T14_22_33.156570
- 路径: **/details_harness|arc:challenge|25_2023-10-03T14-22-33.156570.parquet
- 分割: latest
- 路径: **/details_harness|arc:challenge|25_2023-10-03T14-22-33.156570.parquet
- 分割: 2023_10_03T14_22_33.156570
-
harness_drop_3
- 分割: 2023_10_26T09_12_20.462280
- 路径: **/details_harness|drop|3_2023-10-26T09-12-20.462280.parquet
- 分割: latest
- 路径: **/details_harness|drop|3_2023-10-26T09-12-20.462280.parquet
- 分割: 2023_10_26T09_12_20.462280
-
harness_gsm8k_5
- 分割: 2023_10_26T09_12_20.462280
- 路径: **/details_harness|gsm8k|5_2023-10-26T09-12-20.462280.parquet
- 分割: latest
- 路径: **/details_harness|gsm8k|5_2023-10-26T09-12-20.462280.parquet
- 分割: 2023_10_26T09_12_20.462280
-
harness_hellaswag_10
- 分割: 2023_10_03T14_22_33.156570
- 路径: **/details_harness|hellaswag|10_2023-10-03T14-22-33.156570.parquet
- 分割: latest
- 路径: **/details_harness|hellaswag|10_2023-10-03T14-22-33.156570.parquet
- 分割: 2023_10_03T14_22_33.156570
-
harness_hendrycksTest_5
- 分割: 2023_10_03T14_22_33.156570
- 路径:
- **/details_harness|hendrycksTest-abstract_algebra|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-anatomy|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-astronomy|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-business_ethics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-college_biology|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-college_chemistry|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-college_computer_science|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-college_mathematics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-college_medicine|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-college_physics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-computer_security|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-conceptual_physics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-econometrics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-electrical_engineering|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-elementary_mathematics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-formal_logic|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-global_facts|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_biology|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_chemistry|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_computer_science|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_european_history|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_geography|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_mathematics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_microeconomics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_physics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_psychology|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_statistics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_us_history|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-high_school_world_history|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-human_aging|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-human_sexuality|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-international_law|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-jurisprudence|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-logical_fallacies|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-machine_learning|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-management|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-marketing|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-medical_genetics|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-miscellaneous|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-moral_disputes|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-moral_scenarios|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-nutrition|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-philosophy|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-prehistory|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-professional_accounting|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest-professional_law|5_2023-10-03T14-22-33.156570.parquet
- **/details_harness|hendrycksTest
- 路径:
- 分割: 2023_10_03T14_22_33.156570



