Time Series Forecasting
Transformers
Safetensors
Timer-S1
time series
time-series
forecasting
foundation models
pretrained models
time series foundation models
custom_code
Instructions to use bytedance-research/Timer-S1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use bytedance-research/Timer-S1 with Transformers:
# Load model directly from transformers import Timer-S1 model = Timer-S1.from_pretrained("bytedance-research/Timer-S1", trust_remote_code=True, dtype="auto") - Notebooks
- Google Colab
- Kaggle
Update README.md (#1)
Browse files- Update README.md (6a59992785a526ce74cc9875307dea8faee9001a)
Co-authored-by: Yong Liu <Yong99@users.noreply.huggingface.co>
README.md
CHANGED
|
@@ -88,7 +88,9 @@ print(output[0][4])
|
|
| 88 |
```
|
| 89 |
|
| 90 |
|
| 91 |
-
|
|
|
|
|
|
|
| 92 |
> ```python
|
| 93 |
> # Option 1: reduce batch size or context length
|
| 94 |
> batch_size, lookback_length = 1, 2880
|
|
|
|
| 88 |
```
|
| 89 |
|
| 90 |
|
| 91 |
+
This model support inference using either CPU or GPU. To load this model on GPU, we recommend a GPU with **at least 40GB VRAM** (e.g., A100 40GB/80GB, or H100).
|
| 92 |
+
|
| 93 |
+
> **Encounter out-of-memory at runtime?** Try the following options:
|
| 94 |
> ```python
|
| 95 |
> # Option 1: reduce batch size or context length
|
| 96 |
> batch_size, lookback_length = 1, 2880
|