open-llm-leaderboard-old/details_facebook__opt-30b
收藏数据集概述
数据集名称
- Evaluation run of facebook/opt-30b
数据集描述
- 该数据集是在对模型 facebook/opt-30b 进行评估运行期间自动创建的。
- 数据集包含 122 个配置,每个配置对应一个评估任务。
- 数据集从 5 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_facebook__opt-30b", "harness_gsm8k_5", split="train")
最新结果
- 这些是最新结果,来自 2023-12-02T22:44:52.889942 的运行。 python { "all": { "acc": 0.021986353297952996, "acc_stderr": 0.004039162758110029 }, "harness|gsm8k|5": { "acc": 0.021986353297952996, "acc_stderr": 0.004039162758110029 } }
配置详情
-
harness_arc_challenge_25
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|arc:challenge|25_2023-07-19T22:24:27.765109.parquet
-
harness_drop_3
- 分割: 2023_09_09T12_38_44.623374, 2023_09_17T20_41_44.085857, latest
- 路径: **/details_harness|drop|3_2023-09-09T12-38-44.623374.parquet, **/details_harness|drop|3_2023-09-17T20-41-44.085857.parquet
-
harness_gsm8k_5
- 分割: 2023_09_09T12_38_44.623374, 2023_09_17T20_41_44.085857, 2023_12_02T22_44_52.889942, latest
- 路径: **/details_harness|gsm8k|5_2023-09-09T12-38-44.623374.parquet, **/details_harness|gsm8k|5_2023-09-17T20-41-44.085857.parquet, **/details_harness|gsm8k|5_2023-12-02T22-44-52.889942.parquet
-
harness_hellaswag_10
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hellaswag|10_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: 多个路径,涵盖多个学科的测试结果。
-
harness_hendrycksTest_abstract_algebra_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_anatomy_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-anatomy|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_astronomy_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-astronomy|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_college_biology_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-college_biology|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_college_physics_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-college_physics|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_computer_security_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-computer_security|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_econometrics_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-econometrics|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T22:24:27.765109.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割: 2023_07_19T22_24_27.765109, latest
- 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T22:24:27.765109.parquet



