Add collection overview and source distribution
Browse files
README.md
CHANGED
|
@@ -10,10 +10,24 @@ tags:
|
|
| 10 |
|
| 11 |
# TimeBraid-Alignment
|
| 12 |
|
| 13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
|
| 15 |
## Coming soon
|
| 16 |
|
| 17 |
-
Dataset files are not available here while we review the licenses and redistribution permissions of the underlying sources. We will update this page when we know what can be shared.
|
| 18 |
|
| 19 |
Stay tuned.
|
|
|
|
| 10 |
|
| 11 |
# TimeBraid-Alignment
|
| 12 |
|
| 13 |
+
Time-series and language alignment data accompanying **TimeBraid: Unifying Time Series and Language for Understanding and Forecasting**. The collection brings together synthetic and real time-series descriptions, multivariate relationships, contextual understanding, and forecasting examples.
|
| 14 |
+
|
| 15 |
+
The previously prepared collection contained **2,216,597 records**, distributed as follows:
|
| 16 |
+
|
| 17 |
+
| Component | Records |
|
| 18 |
+
|---|---:|
|
| 19 |
+
| ChatTS univariate | 260,000 |
|
| 20 |
+
| ChatTS multivariate | 108,850 |
|
| 21 |
+
| Synthetic temporal causal models | 100,000 |
|
| 22 |
+
| TimeSeriesExam-derived descriptions | 100,000 |
|
| 23 |
+
| GIFT-derived morphology and forecasting | 1,610,650 |
|
| 24 |
+
| Context-rich descriptions from nine sources | 37,097 |
|
| 25 |
+
| **Total** | **2,216,597** |
|
| 26 |
+
|
| 27 |
+
These are record counts, not counts of unique time series. The companion [TimeBraid-SFT](https://huggingface.co/datasets/XinyueWangg/TimeBraid-SFT) collection adds instruction-tuning examples without repeating this alignment collection.
|
| 28 |
|
| 29 |
## Coming soon
|
| 30 |
|
| 31 |
+
Dataset files are not available here while we review the licenses and redistribution permissions of the underlying sources. The distribution above describes the prepared collection; the final shared subset may differ. We will update this page when we know what can be shared.
|
| 32 |
|
| 33 |
Stay tuned.
|