Add source-level Stage-2 inventory
Browse files
README.md
CHANGED
|
@@ -14,14 +14,42 @@ Instruction-tuning data accompanying **TimeBraid: Unifying Time Series and Langu
|
|
| 14 |
|
| 15 |
The full Stage-2 training inventory contained **4,881,583 examples** (about **4.88M**), distributed as follows:
|
| 16 |
|
| 17 |
-
|
|
| 18 |
-
|---|---:|
|
| 19 |
-
| Forecasting | 2,640,566 |
|
| 20 |
-
|
|
| 21 |
-
|
|
| 22 |
-
|
|
| 23 |
-
|
|
| 24 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
These are training-example counts, not counts of unique time series. The alignment-retention examples are included in the training total; they will not be packaged again in this dataset alongside the companion [TimeBraid-Alignment](https://huggingface.co/datasets/XinyueWangg/TimeBraid-Alignment) collection.
|
| 27 |
|
|
|
|
| 14 |
|
| 15 |
The full Stage-2 training inventory contained **4,881,583 examples** (about **4.88M**), distributed as follows:
|
| 16 |
|
| 17 |
+
| Category | Source or subcategory | Examples |
|
| 18 |
+
|---|---|---:|
|
| 19 |
+
| **Forecasting** | | **2,640,566** |
|
| 20 |
+
| | CGTSF | 53,926 |
|
| 21 |
+
| | FinMultiTime S&P 500 | 144,387 |
|
| 22 |
+
| | MoTime (News/Wiki) | 32,589 |
|
| 23 |
+
| | CAF-7M | 2,314,216 |
|
| 24 |
+
| | Time-MMD | 20,768 |
|
| 25 |
+
| | Time-MQA Forecasting | 74,680 |
|
| 26 |
+
| **Time-series QA and reasoning** | | **1,065,021** |
|
| 27 |
+
| | CaTS-Bench | 15,995 |
|
| 28 |
+
| | ChatTS | 50,377 |
|
| 29 |
+
| | HiTSR | 157,147 |
|
| 30 |
+
| | OpenSQA | 138,754 |
|
| 31 |
+
| | OpenTSLM ECG | 159,313 |
|
| 32 |
+
| | OpenTSLM HAR | 68,542 |
|
| 33 |
+
| | OpenTSLM Sleep | 7,434 |
|
| 34 |
+
| | OpenTSLM TSQA | 38,400 |
|
| 35 |
+
| | Time-MQA Reasoning | 110,212 |
|
| 36 |
+
| | TSAQA Reasoning | 125,957 |
|
| 37 |
+
| | EngineMT-QA | 100,296 |
|
| 38 |
+
| | RATs40K | 32,994 |
|
| 39 |
+
| | Time-MQA Imputation | 38,607 |
|
| 40 |
+
| | TSAQA Anomaly Detection | 20,993 |
|
| 41 |
+
| **Captioning and description** | | **161,883** |
|
| 42 |
+
| | OpenTSLM M4 Captioning | 80,000 |
|
| 43 |
+
| | SensorCaps | 35,960 |
|
| 44 |
+
| | Time-MMD Open-Ended | 1,925 |
|
| 45 |
+
| | TRACE Captioning | 43,998 |
|
| 46 |
+
| **Unimodal corpora** | | **400,000** |
|
| 47 |
+
| | GIFT-Eval Pretrain | 200,000 |
|
| 48 |
+
| | Dolci-Instruct (No-Tools) | 200,000 |
|
| 49 |
+
| **Alignment retention** | | **614,113** |
|
| 50 |
+
| | Understanding retention | 116,294 |
|
| 51 |
+
| | Forecasting retention | 497,819 |
|
| 52 |
+
| **Total** | | **4,881,583** |
|
| 53 |
|
| 54 |
These are training-example counts, not counts of unique time series. The alignment-retention examples are included in the training total; they will not be packaged again in this dataset alongside the companion [TimeBraid-Alignment](https://huggingface.co/datasets/XinyueWangg/TimeBraid-Alignment) collection.
|
| 55 |
|