Spaces:
Running
Running
Add github link
Browse files- src/about.py +3 -1
src/about.py
CHANGED
|
@@ -111,7 +111,9 @@ TITLE = """<h1 align="center" id="space-title"> It's TIME</h1>"""
|
|
| 111 |
|
| 112 |
# What does your leaderboard evaluate?
|
| 113 |
INTRODUCTION_TEXT = """
|
| 114 |
-
TIME introduces a unified benchmark for time series probabilistic forecasting that supports evaluation at **multiple granularities**, ranging from overall performance across datasets to dataset-level, variate-level, and even individual test windows (with visualization). Beyond conventional analysis, the benchmark enables **pattern-driven, cross-dataset benchmarking** by grouping variates with similar temporal features, where patterns are defined based on groups of tsfeatures that capture properties such as trend, seasonality, and stationarity, offering a more systematic understanding of model behavior.
|
|
|
|
|
|
|
| 115 |
"""
|
| 116 |
# An integrated archive further enriches the platform by providing structural tsfeatures and statistical descriptors of all variates,
|
| 117 |
# ensuring both comprehensive evaluation and transparent interpretability across diverse forecasting scenarios
|
|
|
|
| 111 |
|
| 112 |
# What does your leaderboard evaluate?
|
| 113 |
INTRODUCTION_TEXT = """
|
| 114 |
+
TIME introduces a unified benchmark for time series probabilistic forecasting that supports evaluation at **multiple granularities**, ranging from overall performance across datasets to dataset-level, variate-level, and even individual test windows (with visualization). Beyond conventional analysis, the benchmark enables **pattern-driven, cross-dataset benchmarking** by grouping variates with similar temporal features, where patterns are defined based on groups of tsfeatures that capture properties such as trend, seasonality, and stationarity, offering a more systematic understanding of model behavior.
|
| 115 |
+
For data and results, please refer to 🤗 [dataset](https://huggingface.co/datasets/TIME-benchmark/TIME-1.0/tree/main).
|
| 116 |
+
For result submissions, please refer to our [Github repo](https://github.com/zqiao11/TIME/tree/main).
|
| 117 |
"""
|
| 118 |
# An integrated archive further enriches the platform by providing structural tsfeatures and statistical descriptors of all variates,
|
| 119 |
# ensuring both comprehensive evaluation and transparent interpretability across diverse forecasting scenarios
|