open-llm-leaderboard-old/details_DatPySci__pythia-1b-sft-full
收藏Hugging Face2024-02-13 更新2024-06-22 收录
下载链接:
https://hf-mirror.com/datasets/open-llm-leaderboard-old/details_DatPySci__pythia-1b-sft-full
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DatPySci/pythia-1b-sft-full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,每个配置对应一个被评估的任务。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,分割以运行的时间戳命名。train分割始终指向最新的结果。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DatPySci/pythia-1b-sft-full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,每个配置对应一个被评估的任务。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,分割以运行的时间戳命名。train分割始终指向最新的结果。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-old
原始信息汇总
数据集概述
数据集基本信息
- 数据集名称: Evaluation run of DatPySci/pythia-1b-sft-full
- 数据集摘要: 该数据集是在模型 DatPySci/pythia-1b-sft-full 在 Open LLM Leaderboard 上的评估运行期间自动创建的。
- 数据集组成: 包含63个配置,每个配置对应一个评估任务。
- 数据集来源: 从1次运行中创建,每个运行可以在每个配置中作为一个特定的分片找到,分片名称使用运行的时间戳。"train" 分片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_DatPySci__pythia-1b-sft-full", "harness_winogrande_5", split="train")
最新结果
- 最新结果时间戳: 2024-02-13T16:10:25.536341
- 结果示例:
python
{
"all": {
"acc": 0.2437500378442946,
"acc_stderr": 0.030213863245287735,
"acc_norm": 0.24468974101675026,
"acc_norm_stderr": 0.03094305925119546,
"mc1": 0.2252141982864137,
"mc1_stderr": 0.014623240768023496,
"mc2": 0.37081334738032573,
"mc2_stderr": 0.014356461899633393
},
"harness|arc:challenge|25": {
"acc": 0.27303754266211605,
"acc_stderr": 0.013019332762635753,
"acc_norm": 0.295221843003413,
"acc_norm_stderr": 0.013329750293382316
},
"harness|hellaswag|10": {
"acc": 0.38697470623381797,
"acc_stderr": 0.004860623733461137,
"acc_norm": 0.48914558852818163,
"acc_norm_stderr": 0.004988605498273906
},
"harness|hendrycksTest-abstract_algebra|5": {
"acc": 0.25,
"acc_stderr": 0.04351941398892446,
"acc_norm": 0.25,
"acc_norm_stderr": 0.04351941398892446
},
更多任务结果...
}
配置信息
-
配置名称: harness_arc_challenge_25
- 数据文件:
- 分片: 2024_02_13T16_10_25.536341
- 路径: **/details_harness|arc:challenge|25_2024-02-13T16-10-25.536341.parquet
- 分片: latest
- 路径: **/details_harness|arc:challenge|25_2024-02-13T16-10-25.536341.parquet
- 分片: 2024_02_13T16_10_25.536341
- 数据文件:
-
配置名称: harness_gsm8k_5
- 数据文件:
- 分片: 2024_02_13T16_10_25.536341
- 路径: **/details_harness|gsm8k|5_2024-02-13T16-10-25.536341.parquet
- 分片: latest
- 路径: **/details_harness|gsm8k|5_2024-02-13T16-10-25.536341.parquet
- 分片: 2024_02_13T16_10_25.536341
- 数据文件:
-
配置名称: harness_hellaswag_10
- 数据文件:
- 分片: 2024_02_13T16_10_25.536341
- 路径: **/details_harness|hellaswag|10_2024-02-13T16-10-25.536341.parquet
- 分片: latest
- 路径: **/details_harness|hellaswag|10_2024-02-13T16-10-25.536341.parquet
- 分片: 2024_02_13T16_10_25.536341
- 数据文件:
-
配置名称: harness_hendrycksTest_5
- 数据文件:
- 分片: 2024_02_13T16_10_25.536341
- 路径:
- **/details_harness|hendrycksTest-abstract_algebra|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-anatomy|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-astronomy|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-business_ethics|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-clinical_knowledge|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-college_biology|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-college_chemistry|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-college_computer_science|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-college_mathematics|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-college_medicine|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-college_physics|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-computer_security|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-conceptual_physics|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-econometrics|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-electrical_engineering|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-elementary_mathematics|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-formal_logic|5_2024-02-13T16-10-25.536341.parquet
- **/details_harness|hendrycksTest-global_facts|5_2024-02-13T16-10-25.536341.parquet
更多路径...
- 路径:
- 分片: 2024_02_13T16_10_25.536341
- 数据文件:



