open-llm-leaderboard/details_EleutherAI__pythia-160m-deduped
收藏数据集概述
数据集名称
Evaluation run of EleutherAI/pythia-160m-deduped
数据集摘要
该数据集是在评估模型 EleutherAI/pythia-160m-deduped 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" 分片总是指向最新的结果。
- 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_EleutherAI__pythia-160m-deduped", "harness_winogrande_5", split="train")
最新结果
以下是 2023-10-18T14:10:15.721061 运行的最新结果: python { "all": { "em": 0.003145973154362416, "em_stderr": 0.0005734993648436387, "f1": 0.033831795302013495, "f1_stderr": 0.0011064778180343976, "acc": 0.2580433025186501, "acc_stderr": 0.007679640365653923 }, "harness|drop|3": { "em": 0.003145973154362416, "em_stderr": 0.0005734993648436387, "f1": 0.033831795302013495, "f1_stderr": 0.0011064778180343976 }, "harness|gsm8k|5": { "acc": 0.002274450341167551, "acc_stderr": 0.0013121578148674233 }, "harness|winogrande|5": { "acc": 0.5138121546961326, "acc_stderr": 0.014047122916440422 } }
配置详情
-
harness_arc_challenge_25
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|arc:challenge|25_2023-07-19T14:01:37.454131.parquet
-
harness_drop_3
- 分片:2023_10_18T14_10_15.721061, latest
- 路径:
**/details_harness|drop|3_2023-10-18T14-10-15.721061.parquet
-
harness_gsm8k_5
- 分片:2023_10_18T14_10_15.721061, latest
- 路径:
**/details_harness|gsm8k|5_2023-10-18T14-10-15.721061.parquet
-
harness_hellaswag_10
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hellaswag|10_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:多个路径,例如
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_abstract_algebra_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_anatomy_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_astronomy_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_business_ethics_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_college_biology_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_college_medicine_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_college_physics_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_computer_security_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_econometrics_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-electrical_engineering|5_2023-07-19T14:01:37.454131.parquet
-
harness_hendrycksTest_elementary_mathematics_5
- 分片:2023_07_19T14_01_37.454131, latest
- 路径:
**/details_harness|hendrycksTest-elementary_mathematics|5_2023-07-19T14:01:37.454131.parquet
以上是数据集的详细概述,包括数据集名称、摘要、组成、加载示例、最新结果以及各个配置的详细信息。



