open-llm-leaderboard-old/details_BFauber__lora_llama2-13b_10e5_r128_a256
收藏数据集概述
数据集简介
该数据集是在对模型 BFauber/lora_llama2-13b_10e5_r128_a256 进行评估运行期间自动创建的,用于 Open LLM Leaderboard。
数据集结构
- 配置数量:63个配置,每个配置对应一个评估任务。
- 数据来源:数据集来自1次运行,每个运行结果作为一个特定的分割存储在每个配置中,分割名称使用运行的时间戳。
- 最新结果:"train" 分割始终指向最新的结果。
- 汇总结果:一个额外的配置 "results" 存储所有运行的汇总结果,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_BFauber__lora_llama2-13b_10e5_r128_a256", "harness_winogrande_5", split="train")
最新结果
以下是 2024-02-10T01:41:13.878952 运行的最新结果:
python { "all": { "acc": 0.5475513509209767, "acc_stderr": 0.03364466522313527, "acc_norm": 0.5535754457106707, "acc_norm_stderr": 0.03437240325899414, "mc1": 0.2631578947368421, "mc1_stderr": 0.015415241740237017, "mc2": 0.37818229047033813, "mc2_stderr": 0.01371187114283475 }, "harness|arc:challenge|25": { "acc": 0.5674061433447098, "acc_stderr": 0.014478005694182528, "acc_norm": 0.5972696245733788, "acc_norm_stderr": 0.014332236306790149 }, "harness|hellaswag|10": { "acc": 0.6163114917347142, "acc_stderr": 0.004852896681736758, "acc_norm": 0.8207528380800637, "acc_norm_stderr": 0.0038277525727700226 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.29, "acc_stderr": 0.045604802157206845, "acc_norm": 0.29, "acc_norm_stderr": 0.045604802157206845 }, # 其他任务的结果... }
配置详情
-
配置名称:harness_arc_challenge_25
- 数据文件:
- 分割:2024_02_10T01_41_13.878952
- 路径:
**/details_harness|arc:challenge|25_2024-02-10T01-41-13.878952.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2024-02-10T01-41-13.878952.parquet
- 路径:
- 分割:2024_02_10T01_41_13.878952
- 数据文件:
-
配置名称:harness_gsm8k_5
- 数据文件:
- 分割:2024_02_10T01_41_13.878952
- 路径:
**/details_harness|gsm8k|5_2024-02-10T01-41-13.878952.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|gsm8k|5_2024-02-10T01-41-13.878952.parquet
- 路径:
- 分割:2024_02_10T01_41_13.878952
- 数据文件:
-
配置名称:harness_hellaswag_10
- 数据文件:
- 分割:2024_02_10T01_41_13.878952
- 路径:
**/details_harness|hellaswag|10_2024-02-10T01-41-13.878952.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|hellaswag|10_2024-02-10T01-41-13.878952.parquet
- 路径:
- 分割:2024_02_10T01_41_13.878952
- 数据文件:
-
配置名称:harness_hendrycksTest_5
- 数据文件:
- 分割:2024_02_10T01_41_13.878952
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-anatomy|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-astronomy|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-business_ethics|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-college_biology|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-college_chemistry|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-college_computer_science|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-college_mathematics|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-college_medicine|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-college_physics|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-computer_security|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-econometrics|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-electrical_engineering|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-elementary_mathematics|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-formal_logic|5_2024-02-10T01-41-13.878952.parquet**/details_harness|hendrycksTest-global_facts|5_2024-02-10T01-41-13.878952.parquet- `**/details_harness|hendrycksTest-high_school_biology|
- 路径:
- 分割:2024_02_10T01_41_13.878952
- 数据文件:



