open-llm-leaderboard-old/details_PulsarAI__2x-LoRA-Assemble-Platypus2-13B
收藏数据集概述
数据集简介
该数据集是在评估模型 PulsarAI/2x-LoRA-Assemble-Platypus2-13B 在 Open LLM Leaderboard 上的自动创建的。
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。每个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_PulsarAI__2x-LoRA-Assemble-Platypus2-13B", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-26T04:49:09.510505 运行的最新结果: python { "all": { "em": 0.14020553691275167, "em_stderr": 0.003555654511760366, "f1": 0.25958473154362444, "f1_stderr": 0.003697673494004961, "acc": 0.3790556094431875, "acc_stderr": 0.007400551365645916 }, "harness|drop|3": { "em": 0.14020553691275167, "em_stderr": 0.003555654511760366, "f1": 0.25958473154362444, "f1_stderr": 0.003697673494004961 }, "harness|gsm8k|5": { "acc": 0.009097801364670205, "acc_stderr": 0.002615326510775672 }, "harness|winogrande|5": { "acc": 0.7490134175217048, "acc_stderr": 0.012185776220516161 } }
配置详情
以下是数据集的配置详情:
配置列表
harness_arc_challenge_25harness_drop_3harness_gsm8k_5harness_hellaswag_10harness_hendrycksTest_5harness_hendrycksTest_abstract_algebra_5harness_hendrycksTest_anatomy_5harness_hendrycksTest_astronomy_5harness_hendrycksTest_business_ethics_5harness_hendrycksTest_clinical_knowledge_5harness_hendrycksTest_college_biology_5harness_hendrycksTest_college_chemistry_5harness_hendrycksTest_college_computer_science_5harness_hendrycksTest_college_mathematics_5harness_hendrycksTest_college_medicine_5harness_hendrycksTest_college_physics_5harness_hendrycksTest_computer_security_5harness_hendrycksTest_conceptual_physics_5harness_hendrycksTest_econometrics_5harness_hendrycksTest_electrical_engineering_5harness_hendrycksTest_elementary_mathematics_5
数据文件路径
每个配置包含多个分片,每个分片对应一个特定的时间戳,路径如下:
harness_arc_challenge_252023_10_08T14_58_33.553023:**/details_harness|arc:challenge|25_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|arc:challenge|25_2023-10-08T14-58-33.553023.parquet
harness_drop_32023_10_26T04_49_09.510505:**/details_harness|drop|3_2023-10-26T04-49-09.510505.parquetlatest:**/details_harness|drop|3_2023-10-26T04-49-09.510505.parquet
harness_gsm8k_52023_10_26T04_49_09.510505:**/details_harness|gsm8k|5_2023-10-26T04-49-09.510505.parquetlatest:**/details_harness|gsm8k|5_2023-10-26T04-49-09.510505.parquet
harness_hellaswag_102023_10_08T14_58_33.553023:**/details_harness|hellaswag|10_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hellaswag|10_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-08T14-58-33.553023.parquet等latest:**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-08T14-58-33.553023.parquet等
harness_hendrycksTest_abstract_algebra_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_anatomy_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-anatomy|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-anatomy|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_astronomy_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-astronomy|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-astronomy|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_business_ethics_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-business_ethics|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-business_ethics|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_clinical_knowledge_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_college_biology_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-college_biology|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-college_biology|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_college_chemistry_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-college_chemistry|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-college_chemistry|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_college_computer_science_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-college_computer_science|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-college_computer_science|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_college_mathematics_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-college_mathematics|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-college_mathematics|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_college_medicine_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-college_medicine|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-college_medicine|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_college_physics_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-college_physics|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-college_physics|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_computer_security_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-computer_security|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-computer_security|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_conceptual_physics_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-08T14-58-33.553023.parquet
harness_hendrycksTest_econometrics_52023_10_08T14_58_33.553023:**/details_harness|hendrycksTest-econometrics|5_2023-10-08T14-58-33.553023.parquetlatest:**/details_harness|hendrycksTest-econometrics|5_2023-10-08T14-58-33.553023.parquet



