open-llm-leaderboard/details_Aspik101__30B-Lazarus-instruct-PL-lora_unload

该数据集是在模型Aspik101/30B-Lazarus-instruct-PL-lora_unload在Open LLM Leaderboard上的评估运行过程中自动创建的。数据集由3个配置组成，每个配置对应一个评估任务。数据集包含1次运行的详细信息，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，并用于在Open LLM Leaderboard上计算和显示聚合指标。

This dataset was automatically created during the evaluation run of the model Aspik101/30B-Lazarus-instruct-PL-lora_unload on the Open LLM Leaderboard. The dataset consists of 3 configurations, each corresponding to one evaluation task. It contains detailed information for a single run, which is available under each configuration, with split names using the run's timestamp. The "train" split always points to the most recent results. In addition, the "results" configuration stores the aggregated results of all runs and is used to calculate and display the aggregate metrics on the Open LLM Leaderboard.

原始信息汇总

数据集概述

数据集描述

• 数据集摘要： 该数据集是在模型Aspik101/30B-Lazarus-instruct-PL-lora_unload的评估运行期间自动创建的，用于Open LLM Leaderboard。

  数据集由3个配置组成，每个配置对应一个评估任务。数据集从1次运行中创建，每次运行在每个配置中作为一个特定的分割存在，分割名称使用运行的时间戳。"train"分割始终指向最新的结果。

  额外的配置"results"存储了运行的所有聚合结果，并用于计算和显示Open LLM Leaderboard上的聚合指标。

• 最新结果： 以下是来自2023-10-17T08:13:24.195120运行的最新结果： python { "all": { "em": 0.01164010067114094, "em_stderr": 0.0010984380734032925, "f1": 0.07800545302013438, "f1_stderr": 0.0017935902090569574, "acc": 0.4522835158298991, "acc_stderr": 0.010087630088457804 }, "harness|drop|3": { "em": 0.01164010067114094, "em_stderr": 0.0010984380734032925, "f1": 0.07800545302013438, "f1_stderr": 0.0017935902090569574 }, "harness|gsm8k|5": { "acc": 0.11372251705837756, "acc_stderr": 0.008744810131034036 }, "harness|winogrande|5": { "acc": 0.7908445146014207, "acc_stderr": 0.01143045004588157 } }

数据集结构

配置

• harness_drop_3：

  • 分割：2023_10_17T08_13_24.195120
  • 路径：**/details_harness|drop|3_2023-10-17T08-13-24.195120.parquet
  • 分割：latest
  • 路径：**/details_harness|drop|3_2023-10-17T08-13-24.195120.parquet

• harness_gsm8k_5：

  • 分割：2023_10_17T08_13_24.195120
  • 路径：**/details_harness|gsm8k|5_2023-10-17T08-13-24.195120.parquet
  • 分割：latest
  • 路径：**/details_harness|gsm8k|5_2023-10-17T08-13-24.195120.parquet

• harness_winogrande_5：

  • 分割：2023_10_17T08_13_24.195120
  • 路径：**/details_harness|winogrande|5_2023-10-17T08-13-24.195120.parquet
  • 分割：latest
  • 路径：**/details_harness|winogrande|5_2023-10-17T08-13-24.195120.parquet

• results：

  • 分割：2023_10_17T08_13_24.195120
  • 路径：results_2023-10-17T08-13-24.195120.parquet
  • 分割：latest
  • 路径：results_2023-10-17T08-13-24.195120.parquet