open-llm-leaderboard-old/details_LordNoah__latent_gpt2_medium_alpaca_e2
收藏数据集概述
数据集来源
该数据集是在评估模型 LordNoah/latent_gpt2_medium_alpaca_e2 在 Open LLM Leaderboard 上的运行过程中自动创建的。
数据集组成
数据集由63个配置组成,每个配置对应一个评估任务。数据集从1次运行中创建,每个运行可以在每个配置中找到特定的分割,分割名称使用运行的时间戳。"train" 分割始终指向最新的结果。
额外配置
一个额外的配置 "results" 存储了所有运行的聚合结果,用于计算和显示在 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_LordNoah__latent_gpt2_medium_alpaca_e2", "harness_winogrande_5", split="train")
最新结果
以下是 最新结果 的摘要:
python { "all": { "acc": 0.2696207378617201, "acc_stderr": 0.031238054238067026, "acc_norm": 0.27144095071550417, "acc_norm_stderr": 0.03204425561234846, "mc1": 0.2460220318237454, "mc1_stderr": 0.01507721920066259, "mc2": 0.41013083928221167, "mc2_stderr": 0.014408997014495538 }, "harness|arc:challenge|25": { "acc": 0.23293515358361774, "acc_stderr": 0.012352507042617391, "acc_norm": 0.2696245733788396, "acc_norm_stderr": 0.012968040686869157 }, "harness|hellaswag|10": { "acc": 0.3300139414459271, "acc_stderr": 0.004692567655961772, "acc_norm": 0.3972316271659032, "acc_norm_stderr": 0.004883246579496671 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.25, "acc_stderr": 0.04351941398892446, "acc_norm": 0.25, "acc_norm_stderr": 0.04351941398892446 }, # 其他任务的结果... }
配置详情
-
config_name: harness_arc_challenge_25
- 数据文件:
- 分割: 2024_02_18T11_27_26.489335
- 路径:
**/details_harness|arc:challenge|25_2024-02-18T11-27-26.489335.parquet
- 路径:
- 分割: latest
- 路径:
**/details_harness|arc:challenge|25_2024-02-18T11-27-26.489335.parquet
- 路径:
- 分割: 2024_02_18T11_27_26.489335
- 数据文件:
-
config_name: harness_gsm8k_5
- 数据文件:
- 分割: 2024_02_18T11_27_26.489335
- 路径:
**/details_harness|gsm8k|5_2024-02-18T11-27-26.489335.parquet
- 路径:
- 分割: latest
- 路径:
**/details_harness|gsm8k|5_2024-02-18T11-27-26.489335.parquet
- 路径:
- 分割: 2024_02_18T11_27_26.489335
- 数据文件:
-
config_name: harness_hellaswag_10
- 数据文件:
- 分割: 2024_02_18T11_27_26.489335
- 路径:
**/details_harness|hellaswag|10_2024-02-18T11-27-26.489335.parquet
- 路径:
- 分割: latest
- 路径:
**/details_harness|hellaswag|10_2024-02-18T11-27-26.489335.parquet
- 路径:
- 分割: 2024_02_18T11_27_26.489335
- 数据文件:
-
config_name: harness_hendrycksTest_5
- 数据文件:
- 分割: 2024_02_18T11_27_26.489335
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-anatomy|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-astronomy|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-business_ethics|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-college_biology|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-college_chemistry|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-college_computer_science|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-college_mathematics|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-college_medicine|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-college_physics|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-computer_security|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-econometrics|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-electrical_engineering|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-elementary_mathematics|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-formal_logic|5_2024-02-18T11-27-26.489335.parquet**/details_harness|hendrycksTest-global_facts|5_2024-02-18T11-27-26.489335.parquet
其他路径...
- 路径:
- 分割: 2024_02_18T11_27_26.489335
- 数据文件:
}



