open-llm-leaderboard/details_speechlessai__speechless-codellama-dolphin-orca-platypus-13b
收藏数据集概述
数据集创建
- 数据集是在模型
speechlessai/speechless-codellama-dolphin-orca-platypus-13b的评估运行期间自动创建的。 - 该数据集用于 Open LLM Leaderboard 上的评估。
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,每次运行在每个配置中作为一个特定的分片存在,分片名称使用运行的时间戳。
- "train" 分片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_speechlessai__speechless-codellama-dolphin-orca-platypus-13b", "harness_winogrande_5", split="train")
最新结果
- 以下是来自 2023-10-24T03:55:53.446205 运行的最新结果: python { "all": { "em": 0.28366191275167785, "em_stderr": 0.004616354866148243, "f1": 0.3476940016778539, "f1_stderr": 0.004573065876077725, "acc": 0.36921252317982634, "acc_stderr": 0.010525993118068228 }, "harness|drop|3": { "em": 0.28366191275167785, "em_stderr": 0.004616354866148243, "f1": 0.3476940016778539, "f1_stderr": 0.004573065876077725 }, "harness|gsm8k|5": { "acc": 0.08491281273692192, "acc_stderr": 0.0076782128244508 }, "harness|winogrande|5": { "acc": 0.6535122336227308, "acc_stderr": 0.013373773411685655 } }
配置详情
-
harness_arc_challenge_25
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|arc:challenge|25_2023-09-13T12-05-20.709991.parquet
-
harness_drop_3
- 分片:2023_10_24T03_55_53.446205, latest
- 路径:
**/details_harness|drop|3_2023-10-24T03-55-53.446205.parquet
-
harness_gsm8k_5
- 分片:2023_10_24T03_55_53.446205, latest
- 路径:
**/details_harness|gsm8k|5_2023-10-24T03-55-53.446205.parquet
-
harness_hellaswag_10
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hellaswag|10_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_anatomy_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_astronomy_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_business_ethics_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_college_biology_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_college_medicine_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_college_physics_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_computer_security_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_econometrics_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-09-13T12-05-20.709991.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分片:2023_09_13T12_05_20.709991, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-09-13T12-05-20.709991.parquet
以上是数据集的详细概述,包括数据集的创建背景、结构、加载示例、最新结果以及各个配置的详细信息。



