open-llm-leaderboard-old/details_anhnv125__llama-op-v4
收藏数据集概述
该数据集是在评估模型 anhnv125/llama-op-v4 在 Open LLM Leaderboard 上的自动创建的。
数据集组成
- 数据集包含 64 个配置,每个配置对应一个评估任务。
- 数据集从 2 次运行中创建,每次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。
- 一个额外的配置 "results" 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_anhnv125__llama-op-v4", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-24T15:15:17.062289 运行 的最新结果: python { "all": { "em": 0.006187080536912751, "em_stderr": 0.0008030357040874047, "f1": 0.07929949664429545, "f1_stderr": 0.0016902591729910574, "acc": 0.4277794698693672, "acc_stderr": 0.010070411670220034 }, "harness|drop|3": { "em": 0.006187080536912751, "em_stderr": 0.0008030357040874047, "f1": 0.07929949664429545, "f1_stderr": 0.0016902591729910574 }, "harness|gsm8k|5": { "acc": 0.09628506444275967, "acc_stderr": 0.008125264128215882 }, "harness|winogrande|5": { "acc": 0.7592738752959748, "acc_stderr": 0.012015559212224185 } }
配置详情
以下是部分配置及其数据文件路径:
-
harness_arc_challenge_25
- 分割:2023_10_01T16_49_40.614622
- 路径:
**/details_harness|arc:challenge|25_2023-10-01T16-49-40.614622.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2023-10-01T16-49-40.614622.parquet
- 路径:
- 分割:2023_10_01T16_49_40.614622
-
harness_drop_3
- 分割:2023_10_24T15_15_17.062289
- 路径:
**/details_harness|drop|3_2023-10-24T15-15-17.062289.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|drop|3_2023-10-24T15-15-17.062289.parquet
- 路径:
- 分割:2023_10_24T15_15_17.062289
-
harness_gsm8k_5
- 分割:2023_10_24T15_15_17.062289
- 路径:
**/details_harness|gsm8k|5_2023-10-24T15-15-17.062289.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|gsm8k|5_2023-10-24T15-15-17.062289.parquet
- 路径:
- 分割:2023_10_24T15_15_17.062289
-
harness_hellaswag_10
- 分割:2023_10_01T16_49_40.614622
- 路径:
**/details_harness|hellaswag|10_2023-10-01T16-49-40.614622.parquet
- 路径:
- 分割:latest
- 路径:
**/details_harness|hellaswag|10_2023-10-01T16-49-40.614622.parquet
- 路径:
- 分割:2023_10_01T16_49_40.614622
-
harness_hendrycksTest_5
- 分割:2023_10_01T16_49_40.614622
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-formal_logic|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-global_facts|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_biology|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_chemistry|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_computer_science|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_european_history|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_geography|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_government_and_politics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_macroeconomics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_mathematics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_microeconomics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_physics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_psychology|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_statistics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_us_history|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-high_school_world_history|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-human_aging|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-human_sexuality|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-international_law|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-jurisprudence|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-logical_fallacies|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-machine_learning|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-management|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-marketing|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-medical_genetics|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-miscellaneous|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-moral_disputes|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-moral_scenarios|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-nutrition|5_2023-10-01T16-49-40.614622.parquet - 路径:
**/details_harness|hendrycksTest-philosophy|5_2023-10-01T16-49-40.614622.parquet - 路径:`**/details_harness|hendrycksTest-prehistory|5_2023-
- 路径:
- 分割:2023_10_01T16_49_40.614622



