five

open-llm-leaderboard/details_CHIH-HUNG__llama-2-13b-FINETUNE5_4w-r4-q_k_v_o

收藏
Hugging Face2023-10-28 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard/details_CHIH-HUNG__llama-2-13b-FINETUNE5_4w-r4-q_k_v_o
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CHIH-HUNG/llama-2-13b-FINETUNE5_4w-r4-q_k_v_o在Open LLM Leaderboard上的评估运行期间自动创建的。它由64个配置组成,每个配置对应一个被评估的任务。数据集是从2次运行中生成的,每次运行在每个配置中表示为特定的拆分。train拆分始终指向最新的结果。一个额外的配置results存储了所有运行的聚合结果,这些结果用于计算和显示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

数据集来源

该数据集是在评估模型 CHIH-HUNG/llama-2-13b-FINETUNE5_4w-r4-q_k_v_oOpen LLM Leaderboard 上的运行过程中自动创建的。

数据集组成

  • 数据集包含 64 个配置,每个配置对应一个评估任务。
  • 数据集由 2 次运行创建,每次运行的结果作为特定分片存储,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_CHIH-HUNG__llama-2-13b-FINETUNE5_4w-r4-q_k_v_o", "harness_winogrande_5", split="train")

最新结果

以下是 2023-10-28T03:32:51.454817 运行的最新结果: python { "all": { "em": 0.37258808724832215, "em_stderr": 0.004951428522573584, "f1": 0.41863255033557134, "f1_stderr": 0.004838761301543826, "acc": 0.4445987937813739, "acc_stderr": 0.010466651540029098 }, "harness|drop|3": { "em": 0.37258808724832215, "em_stderr": 0.004951428522573584, "f1": 0.41863255033557134, "f1_stderr": 0.004838761301543826 }, "harness|gsm8k|5": { "acc": 0.12282031842304776, "acc_stderr": 0.009041108602874675 }, "harness|winogrande|5": { "acc": 0.7663772691397001, "acc_stderr": 0.011892194477183524 } }

配置详情

  • harness_arc_challenge_25

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|arc:challenge|25_2023-10-10T09-48-52.263585.parquet
  • harness_drop_3

    • 分片:2023_10_28T03_32_51.454817, latest
    • 路径:**/details_harness|drop|3_2023-10-28T03-32-51.454817.parquet
  • harness_gsm8k_5

    • 分片:2023_10_28T03_32_51.454817, latest
    • 路径:**/details_harness|gsm8k|5_2023-10-28T03-32-51.454817.parquet
  • harness_hellaswag_10

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hellaswag|10_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:多个路径,例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_abstract_algebra_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_anatomy_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-anatomy|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_astronomy_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-astronomy|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_business_ethics_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-business_ethics|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_clinical_knowledge_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_college_biology_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-college_biology|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_college_chemistry_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-college_chemistry|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_college_computer_science_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-college_computer_science|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_college_mathematics_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-college_mathematics|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_college_medicine_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-college_medicine|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_college_physics_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-college_physics|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_computer_security_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-computer_security|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_conceptual_physics_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_econometrics_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-econometrics|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_electrical_engineering_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-electrical_engineering|5_2023-10-10T09-48-52.263585.parquet
  • harness_hendrycksTest_elementary_mathematics_5

    • 分片:2023_10_10T09_48_52.263585, latest
    • 路径:**/details_harness|hendrycksTest-elementary_mathematics|5_2023-10-10T09-48-52.263585.parquet

以上是数据集的详细概述,包括数据集的来源、组成、加载示例、最新结果以及各个配置的详细信息。

二维码
社区交流群
二维码
科研交流群
商业服务