Add KernelSynth training data

README.md

@@ -919,7 +919,19 @@ dataset_info:
   dataset_size: 57560596
   license: Apache 2.0
   homepage: https://github.com/mbohlkeschneider/gluon-ts/tree/mv_release/datasets
-- config_name: tsmixup_10m
+- config_name: training_corpus_kernel_synth_1m
+  features:
+  - name: target
+    sequence: float64
+  - name: id
+    dtype: string
+  - name: timestamp
+    sequence: timestamp[ms]
+  splits:
+  - name: train
+    num_examples: 1000000
+  download_size: 8313239368
+- config_name: training_corpus_tsmixup_10m
   features:
   - name: target
     sequence: float64
@@ -1566,10 +1578,14 @@ configs:
   data_files:
   - split: train
     path: taxi_30min/train-*
-- config_name: tsmixup_10m
+- config_name: training_corpus_kernel_synth_1m
+  data_files:
+  - split: train
+    path: training_corpus/kernel_synth_1m/train-*
+- config_name: training_corpus_tsmixup_10m
   data_files:
   - split: train
-    path: tsmixup_10m/train-*
+    path: training_corpus/tsmixup_10m/train-*
 - config_name: uber_tlc_daily
   data_files:
   - split: train
@@ -1731,12 +1747,12 @@ or, for the largest datasets like `weatherbench_hourly_temperature`, reading the
 ds = datasets.load_dataset("autogluon/chronos_datasets", "weatherbench_hourly_temperature", streaming=True, split="train")
 ```
 
-## Chronos training set with TSMixup
-The dataset used for training the Chronos model (including KernelSynth synthetic data + TSMixup augmentations) can be loaded via the config `tsmixup_10m`
+## Chronos training corpus with TSMixup & KernelSynth
+The training corpus used for training the Chronos models can be loaded via the configs `training_corpus_tsmixup_10m` (10M TSMixup augmentations of real-world data) and `training_corpus_kernel_synth_1m` (1M synthetic time series generated with KernelSynth), e.g.,
 ```python
-ds = datasets.load_dataset("autogluon/chronos_datasets", "tsmixup_10m", streaming=True, split="train")
+ds = datasets.load_dataset("autogluon/chronos_datasets", "training_corpus_tsmixup_10m", streaming=True, split="train")
 ```
-Note that since data in the `tsmixup_10m` config was obtained by combining various synthetic & real-world time series, the timestamps contain dummy values that have no connection to the original data.
+Note that since data in the training corpus was obtained by combining various synthetic & real-world time series, the timestamps contain dummy values that have no connection to the original data.
 
 
 ## License