helenqu/astro-classification-noaugment
收藏Hugging Face2024-02-27 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/helenqu/astro-classification-noaugment
下载链接
链接失效反馈官方服务:
资源简介:
---
license: mit
---
# Data Format
The dataset has been split into `train`, `val`, and `test` jsonl files.
Each line of each jsonl file has the following fields:
* `object_id`: unique identifier for the object
* `times_wv`: an array of shape _(T, 2)_, where the i-th element contains (t_i, w_i), the time and wavelength of the i-th measured flux value. If t_i = 0, then this is a padded value and the associated flux f_i will also be 0.
* `lightcurve`: an array of shape _(T, 2)_, where the i-th element contains (f_i, f_err_i), the i-th flux and uncertainty on that flux measurement (flux error).
* `label`: an integer label corresponding to an astronomical object type. This is the value to be predicted.
# Usage
The directory includes a custom dataset loading script (`raw_train_with_labels.py`), which will be used when calling
```
load_dataset("/path/to/current/dir")
```
提供机构:
helenqu
原始信息汇总
数据集格式
数据集被分为 train、val 和 test 三个 jsonl 文件。每个 jsonl 文件的每一行包含以下字段:
object_id:对象的唯一标识符。times_wv:形状为 (T, 2) 的数组,其中第 i 个元素包含 (t_i, w_i),表示第 i 个测量通量值的时间和波长。如果 t_i = 0,则这是一个填充值,关联的通量 f_i 也将为 0。lightcurve:形状为 (T, 2) 的数组,其中第 i 个元素包含 (f_i, f_err_i),表示第 i 个通量及其测量不确定度(通量误差)。label:对应于天文对象类型的整数标签。这是需要预测的值。
使用方法
目录中包含一个自定义的数据集加载脚本(raw_train_with_labels.py),在调用以下命令时将使用该脚本:
load_dataset("/path/to/current/dir")



