open-llm-leaderboard/details_dhmeltzer__llama-7b-SFT_ds_eli5_1024_r_64_alpha_16_merged

该数据集是在Open LLM Leaderboard上对特定模型进行评估时自动创建的。数据集由64个配置组成，每个配置对应一个评估任务。数据集是从3次运行中创建的，每次运行在每个配置中表示为特定的分割。train分割始终指向最新的结果。一个名为results的额外配置存储了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。README还提供了如何使用Python代码加载运行细节的示例，并包含了特定运行的最新结果。

This dataset was automatically created during the evaluation of specific models on the Open LLM Leaderboard. The dataset consists of 64 configurations, each corresponding to one evaluation task. It is constructed from 3 runs, where each run is represented as a specific split under each configuration. The train split always points to the most up-to-date results. An additional configuration named "results" stores the aggregated results across all runs, which are used to calculate and display the aggregate metrics on the Open LLM Leaderboard. The README also provides examples of how to load run details using Python code, and includes the latest results from specific runs.

原始信息汇总

数据集概述

数据集信息

• 名称: Evaluation run of dhmeltzer/llama-7b-SFT_ds_eli5_1024_r_64_alpha_16_merged
• 来源: 自动创建于模型 dhmeltzer/llama-7b-SFT_ds_eli5_1024_r_64_alpha_16_merged 在 Open LLM Leaderboard 的评估运行中。
• 组成: 包含64个配置，每个配置对应一个评估任务。
• 创建: 从3次运行中创建，每次运行对应一个特定的时间戳分割。"train" 分割始终指向最新的结果。
• 额外配置: "results" 配置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_dhmeltzer__llama-7b-SFT_ds_eli5_1024_r_64_alpha_16_merged", "harness_winogrande_5", split="train")

最新结果

• 来源: 最新结果来自 run 2023-10-23T02:38:18.626473
• 详细结果: python { "all": { "em": 0.014681208053691275, "em_stderr": 0.0012317113143108561, "f1": 0.07373846476510039, "f1_stderr": 0.0018229608118759215, "acc": 0.3983262056052844, "acc_stderr": 0.009142329658293176 }, "harness|drop|3": { "em": 0.014681208053691275, "em_stderr": 0.0012317113143108561, "f1": 0.07373846476510039, "f1_stderr": 0.0018229608118759215 }, "harness|gsm8k|5": { "acc": 0.05079605761940864, "acc_stderr": 0.006048352096878091 }, "harness|winogrande|5": { "acc": 0.7458563535911602, "acc_stderr": 0.012236307219708262 } }

配置详情

• harness_arc_challenge_25:

  • 分割: 2023_08_31T20_30_17.516134
    • 路径: **/details_harness|arc:challenge|25_2023-08-31T20:30:17.516134.parquet
  • 分割: 2023_08_31T20_49_05.320050
    • 路径: **/details_harness|arc:challenge|25_2023-08-31T20:49:05.320050.parquet
  • 分割: latest
    • 路径: **/details_harness|arc:challenge|25_2023-08-31T20:49:05.320050.parquet

• harness_drop_3:

  • 分割: 2023_10_23T02_38_18.626473
    • 路径: **/details_harness|drop|3_2023-10-23T02-38-18.626473.parquet
  • 分割: latest
    • 路径: **/details_harness|drop|3_2023-10-23T02-38-18.626473.parquet

• harness_gsm8k_5:

  • 分割: 2023_10_23T02_38_18.626473
    • 路径: **/details_harness|gsm8k|5_2023-10-23T02-38-18.626473.parquet
  • 分割: latest
    • 路径: **/details_harness|gsm8k|5_2023-10-23T02-38-18.626473.parquet

• harness_hellaswag_10:

  • 分割: 2023_08_31T20_30_17.516134
    • 路径: **/details_harness|hellaswag|10_2023-08-31T20:30:17.516134.parquet
  • 分割: 2023_08_31T20_49_05.320050
    • 路径: **/details_harness|hellaswag|10_2023-08-31T20:49:05.320050.parquet
  • 分割: latest
    • 路径: **/details_harness|hellaswag|10_2023-08-31T20:49:05.320050.parquet

• harness_hendrycksTest_5:

  • 分割: 2023_08_31T20_30_17.516134
    • 路径: 多个文件路径，例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-31T20:30:17.516134.parquet
  • 分割: 2023_08_31T20_49_05.320050
    • 路径: 多个文件路径，例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-31T20:49:05.320050.parquet
  • 分割: latest
    • 路径: 多个文件路径，例如 **/details_harness|hendrycksTest-abstract_algebra|5_2023-08-31T20:49:05.320050.parquet