open-llm-leaderboard/details_vonjack__phi-3-mini-4k-instruct-llamafied
收藏Hugging Face2024-04-26 更新2024-06-12 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_vonjack__phi-3-mini-4k-instruct-llamafied
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是在模型vonjack/phi-3-mini-4k-instruct-llamafied在Open LLM Leaderboard上的评估运行期间自动创建的。数据集由63个配置组成,每个配置对应一个评估任务。数据集是从1次运行中创建的,每次运行可以在每个配置中找到,分割名称使用运行的时间戳。train分割始终指向最新结果。此外,results配置存储了所有运行的聚合结果,并用于计算和显示Open LLM Leaderboard上的聚合指标。
该数据集是在模型vonjack/phi-3-mini-4k-instruct-llamafied在Open LLM Leaderboard上的评估运行期间自动创建的。数据集由63个配置组成,每个配置对应一个评估任务。数据集是从1次运行中创建的,每次运行可以在每个配置中找到,分割名称使用运行的时间戳。train分割始终指向最新结果。此外,results配置存储了所有运行的聚合结果,并用于计算和显示Open LLM Leaderboard上的聚合指标。
提供机构:
open-llm-leaderboard
原始信息汇总
数据集概述
数据集名称: Evaluation run of vonjack/phi-3-mini-4k-instruct-llamafied
创建目的: 该数据集是在评估模型vonjack/phi-3-mini-4k-instruct-llamafied的过程中自动创建的,用于Open LLM Leaderboard的评估。
数据集组成:
- 配置数量: 63个
- 数据来源: 1次运行
- 数据结构: 每个配置对应一个评估任务,每个运行以特定的时间戳命名,并作为数据集的一个分割。
- 特殊配置: 包含一个名为"results"的配置,用于存储所有运行的聚合结果,用于计算和显示聚合指标。
数据集加载示例
python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_vonjack__phi-3-mini-4k-instruct-llamafied", "harness_winogrande_5", split="train")
最新结果
最新结果来自2024-04-24T21:46:20.884882的运行,具体结果包括多个任务的准确率(acc)和标准误差(acc_stderr)等指标。例如:
- harness|arc:challenge|25:
- acc: 0.598976
- acc_stderr: 0.014322
- harness|hellaswag|10:
- acc: 0.620394
- acc_stderr: 0.004843
数据集配置详情
-
配置名称: harness_arc_challenge_25
- 数据文件:
- split: 2024_04_24T21_46_20.884882
- path: /details_harness|arc:challenge|25_2024-04-24T21-46-20.884882.parquet
- split: latest
- path: /details_harness|arc:challenge|25_2024-04-24T21-46-20.884882.parquet
- split: 2024_04_24T21_46_20.884882
- 数据文件:
-
配置名称: harness_gsm8k_5
- 数据文件:
- split: 2024_04_24T21_46_20.884882
- path: /details_harness|gsm8k|5_2024-04-24T21-46-20.884882.parquet
- split: latest
- path: /details_harness|gsm8k|5_2024-04-24T21-46-20.884882.parquet
- split: 2024_04_24T21_46_20.884882
- 数据文件:
-
配置名称: harness_hellaswag_10
- 数据文件:
- split: 2024_04_24T21_46_20.884882
- path: /details_harness|hellaswag|10_2024-04-24T21-46-20.884882.parquet
- split: latest
- path: /details_harness|hellaswag|10_2024-04-24T21-46-20.884882.parquet
- split: 2024_04_24T21_46_20.884882
- 数据文件:
-
配置名称: harness_hendrycksTest_5
- 数据文件:
- split: 2024_04_24T21_46_20.884882
- path:
- /details_harness|hendrycksTest-abstract_algebra|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-anatomy|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-astronomy|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-business_ethics|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-clinical_knowledge|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-college_biology|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-college_chemistry|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-college_computer_science|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-college_mathematics|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-college_medicine|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-college_physics|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-computer_security|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-conceptual_physics|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-econometrics|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-electrical_engineering|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-elementary_mathematics|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-formal_logic|5_2024-04-24T21-46-20.884882.parquet
- /details_harness|hendrycksTest-global_facts|5_2024-04-24T21-46-20.884882.parquet
- path:
- split: 2024_04_24T21_46_20.884882
- 数据文件:



