| title: tinycast | |
| emoji: 📉 | |
| colorFrom: blue | |
| colorTo: red | |
| sdk: static | |
| pinned: false | |
| license: mit | |
| short_description: Tiny models vs transformers on the ETTh1 benchmark | |
| # tinycast | |
| Do you need a transformer to forecast? In-browser demo of tiny forecasting | |
| models (NLinear, DLinear + baselines) on the ETTh1 benchmark — inference runs | |
| as plain JavaScript matrix products, no backend. The ETTh1 CSV is fetched | |
| directly from [ETDataset](https://github.com/zhouhaoyi/ETDataset) at load time. | |
| Source, training code and full evaluation (TCN, LSTM, published transformer | |
| comparison): https://github.com/UsmarHaider/tinycast | |