open-llm-leaderboard/details_YeungNLP__firefly-llama2-7b-pretrain
收藏数据集概述
数据集来源
该数据集是在评估模型 YeungNLP/firefly-llama2-7b-pretrain 在 Open LLM Leaderboard 上的运行过程中自动创建的。
数据集结构
- 数据集由64个配置组成,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_YeungNLP__firefly-llama2-7b-pretrain", "harness_winogrande_5", split="train")
最新结果
以下是 最新结果来自2023-10-24T22:23:31.822890 的摘要: python { "all": { "em": 0.0007340604026845638, "em_stderr": 0.0002773614457335575, "f1": 0.0473752097315439, "f1_stderr": 0.0011829405023092946, "acc": 0.36752358971812016, "acc_stderr": 0.008870377138277116 }, "harness|drop|3": { "em": 0.0007340604026845638, "em_stderr": 0.0002773614457335575, "f1": 0.0473752097315439, "f1_stderr": 0.0011829405023092946 }, "harness|gsm8k|5": { "acc": 0.032600454890068235, "acc_stderr": 0.004891669021939581 }, "harness|winogrande|5": { "acc": 0.7024467245461721, "acc_stderr": 0.012849085254614652 } }
配置详情
以下是部分配置的详细信息:
-
harness_arc_challenge_25
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|arc:challenge|25_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|arc:challenge|25_2023-09-11T15-29-37.507273.parquet
-
harness_drop_3
- 分割:2023_10_24T22_23_31.822890
- 路径:
**/details_harness|drop|3_2023-10-24T22-23-31.822890.parquet - 分割:latest
- 路径:
**/details_harness|drop|3_2023-10-24T22-23-31.822890.parquet
-
harness_gsm8k_5
- 分割:2023_10_24T22_23_31.822890
- 路径:
**/details_harness|gsm8k|5_2023-10-24T22-23-31.822890.parquet - 分割:latest
- 路径:
**/details_harness|gsm8k|5_2023-10-24T22-23-31.822890.parquet
-
harness_hellaswag_10
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hellaswag|10_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hellaswag|10_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-11T15-29-37.507273.parquet等 - 分割:latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-11T15-29-37.507273.parquet等
-
harness_hendrycksTest_abstract_algebra_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_anatomy_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-anatomy|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_astronomy_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-astronomy|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_business_ethics_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-business_ethics|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_clinical_knowledge_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-clinical_knowledge|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_college_biology_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_biology|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_college_chemistry_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_chemistry|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_college_computer_science_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_computer_science|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_college_mathematics_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_mathematics|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_college_medicine_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_medicine|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_college_physics_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-college_physics|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_computer_security_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-computer_security|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_conceptual_physics_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-conceptual_physics|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_econometrics_5
- 分割:2023_09_11T15_29_37.507273
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-09-11T15-29-37.507273.parquet - 分割:latest
- 路径:
**/details_harness|hendrycksTest-econometrics|5_2023-09-11T15-29-37.507273.parquet
-
harness_hendrycksTest_electrical_engineering_5
- 分割:2023_09_11T15_29_37.507273
- 路径:`**/details_harness|hendrycksTest-electrical_engineering|5_2023-09-11T15-29-37.507273.par



