open-llm-leaderboard-old/details_OpenAssistant__galactica-6.7b-finetuned
收藏数据集概述
数据集名称
Evaluation run of OpenAssistant/galactica-6.7b-finetuned
数据集描述
该数据集是在模型 OpenAssistant/galactica-6.7b-finetuned 在 Open LLM Leaderboard 上的评估运行期间自动创建的。
数据集结构
- 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_OpenAssistant__galactica-6.7b-finetuned", "harness_winogrande_5", split="train")
最新结果
这些是最新的结果,来自 2023-10-22T02:17:57.155970 的运行(注意可能还有其他任务的结果,可以在结果和每个评估的 "latest" 分割中找到): python { "all": { "em": 0.0037751677852348995, "em_stderr": 0.0006280387809484433, "f1": 0.07303901006711401, "f1_stderr": 0.001555851204252822, "acc": 0.3040187939848238, "acc_stderr": 0.009332676038724909 }, "harness|drop|3": { "em": 0.0037751677852348995, "em_stderr": 0.0006280387809484433, "f1": 0.07303901006711401, "f1_stderr": 0.001555851204252822 }, "harness|gsm8k|5": { "acc": 0.0310841546626232, "acc_stderr": 0.004780296718393351 }, "harness|winogrande|5": { "acc": 0.5769534333070244, "acc_stderr": 0.013885055359056465 } }
配置详情
-
harness_arc_challenge_25
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|arc:challenge|25_2023-08-25T20:22:41.470589.parquet
-
harness_drop_3
- 分割: 2023_10_22T02_17_57.155970, latest
- 路径: **/details_harness|drop|3_2023-10-22T02-17-57.155970.parquet
-
harness_gsm8k_5
- 分割: 2023_10_22T02_17_57.155970, latest
- 路径: **/details_harness|gsm8k|5_2023-10-22T02-17-57.155970.parquet
-
harness_hellaswag_10
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hellaswag|10_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: 多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_anatomy_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-anatomy|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_astronomy_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-astronomy|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-business_ethics|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-clinical_knowledge|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_college_biology_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-college_biology|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-college_chemistry|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-college_computer_science|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-college_mathematics|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-college_medicine|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_college_physics_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-college_physics|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_computer_security_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-computer_security|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-conceptual_physics|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_econometrics_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-econometrics|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-electrical_engineering|5_2023-08-25T20:22:41.470589.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分割: 2023_08_25T20_22_41.470589, latest
- 路径: **/details_harness|hendrycksTest-elementary_mathematics|5_2023-08-25T20:22:41.470589.parquet
以上是数据集的详细概述,包括数据集名称、描述、结构、加载示例、最新结果以及各个配置的详细信息。



