open-llm-leaderboard-old/details_adamo1139__Yi-34B-200K-AEZAKMI-RAW-2301-LoRA
收藏数据集概述
数据集简介
该数据集是在评估模型 adamo1139/Yi-34B-200K-AEZAKMI-RAW-2301-LoRA 在 Open LLM Leaderboard 上的自动创建的。
数据集组成
- 该数据集包含 63 个配置,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_adamo1139__Yi-34B-200K-AEZAKMI-RAW-2301-LoRA", "harness_winogrande_5", split="train")
最新结果
这些是最新结果的示例: python { "all": { "acc": 0.7413605787370283, "acc_stderr": 0.02895069135836259, "acc_norm": 0.7476488274301629, "acc_norm_stderr": 0.029481162291123596, "mc1": 0.40758873929008566, "mc1_stderr": 0.017201949234553107, "mc2": 0.5708422092704679, "mc2_stderr": 0.015184723749426742 }, "harness|arc:challenge|25": { "acc": 0.6262798634812287, "acc_stderr": 0.014137708601759091, "acc_norm": 0.659556313993174, "acc_norm_stderr": 0.01384746051889298 }, "harness|hellaswag|10": { "acc": 0.6420035849432384, "acc_stderr": 0.0047843129724954, "acc_norm": 0.8388767177853017, "acc_norm_stderr": 0.003668932629672556 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.4, "acc_stderr": 0.049236596391733084, "acc_norm": 0.4, "acc_norm_stderr": 0.049236596391733084 }, # 其他任务的结果... }
配置详情
-
config_name: harness_arc_challenge_25
- 数据文件:
- 分割: 2024_01_27T16_09_42.767487
- 路径: **/details_harness|arc:challenge|25_2024-01-27T16-09-42.767487.parquet
- 分割: latest
- 路径: **/details_harness|arc:challenge|25_2024-01-27T16-09-42.767487.parquet
- 分割: 2024_01_27T16_09_42.767487
- 数据文件:
-
config_name: harness_gsm8k_5
- 数据文件:
- 分割: 2024_01_27T16_09_42.767487
- 路径: **/details_harness|gsm8k|5_2024-01-27T16-09-42.767487.parquet
- 分割: latest
- 路径: **/details_harness|gsm8k|5_2024-01-27T16-09-42.767487.parquet
- 分割: 2024_01_27T16_09_42.767487
- 数据文件:
-
config_name: harness_hellaswag_10
- 数据文件:
- 分割: 2024_01_27T16_09_42.767487
- 路径: **/details_harness|hellaswag|10_2024-01-27T16-09-42.767487.parquet
- 分割: latest
- 路径: **/details_harness|hellaswag|10_2024-01-27T16-09-42.767487.parquet
- 分割: 2024_01_27T16_09_42.767487
- 数据文件:
-
config_name: harness_hendrycksTest_5
- 数据文件:
- 分割: 2024_01_27T16_09_42.767487
- 路径:
- **/details_harness|hendrycksTest-abstract_algebra|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-anatomy|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-astronomy|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-business_ethics|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-clinical_knowledge|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-college_biology|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-college_chemistry|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-college_computer_science|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-college_mathematics|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-college_medicine|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-college_physics|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-computer_security|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-conceptual_physics|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-econometrics|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-electrical_engineering|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-elementary_mathematics|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-formal_logic|5_2024-01-27T16-09-42.767487.parquet
- **/details_harness|hendrycksTest-global_facts|5_2024-01-27T16-09-42.767487.parquet
其他任务的路径...
- 路径:
- 分割: 2024_01_27T16_09_42.767487
- 数据文件:



