Instructions to use talant28/Kronos-TW-Predictor with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- KRONOS
How to use talant28/Kronos-TW-Predictor with KRONOS:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Upload folder using huggingface_hub
Browse files- README.md +45 -0
- config.json +13 -0
- model.safetensors +3 -0
README.md
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: zh
|
| 3 |
+
license: mit
|
| 4 |
+
tags:
|
| 5 |
+
- finance
|
| 6 |
+
- time-series
|
| 7 |
+
- taiwan
|
| 8 |
+
- stock
|
| 9 |
+
- kronos
|
| 10 |
+
base_model: NeoQuasar/Kronos-base
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# Kronos-TW-Predictor
|
| 14 |
+
|
| 15 |
+
Kronos Autoregressive Predictor fine-tuned on Taiwan Stock Exchange (TWSE) daily data.
|
| 16 |
+
|
| 17 |
+
## Base Model
|
| 18 |
+
[NeoQuasar/Kronos-base](https://huggingface.co/NeoQuasar/Kronos-base)
|
| 19 |
+
|
| 20 |
+
## Training Data
|
| 21 |
+
- Source: TWSE listed stocks (613 stocks after quality filtering)
|
| 22 |
+
- Period: 2010-01-01 ~ 2024-12-31
|
| 23 |
+
- Features: open, high, low, close, volume, amount
|
| 24 |
+
- Predict window: 10 trading days
|
| 25 |
+
|
| 26 |
+
## Training Config
|
| 27 |
+
- Epochs: 3
|
| 28 |
+
- Batch size: 8 (effective 256 with gradient accumulation)
|
| 29 |
+
- Learning rate: 4e-5
|
| 30 |
+
- Hardware: RTX 4080 Laptop 12GB
|
| 31 |
+
|
| 32 |
+
## Usage
|
| 33 |
+
Use with [Kronos-TW-Tokenizer](https://huggingface.co/talant28/Kronos-TW-Tokenizer)
|
| 34 |
+
|
| 35 |
+
## Citation
|
| 36 |
+
This model is fine-tuned from Kronos. If you use this work, please cite the original paper:
|
| 37 |
+
|
| 38 |
+
```bibtex
|
| 39 |
+
@article{shi2025kronos,
|
| 40 |
+
title={Kronos: A Foundation Model for Time Series},
|
| 41 |
+
author={Shi, Yu and others},
|
| 42 |
+
journal={arXiv preprint arXiv:2504.03249},
|
| 43 |
+
year={2025}
|
| 44 |
+
}
|
| 45 |
+
```
|
config.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"attn_dropout_p": 0.0,
|
| 3 |
+
"d_model": 832,
|
| 4 |
+
"ff_dim": 2048,
|
| 5 |
+
"ffn_dropout_p": 0.2,
|
| 6 |
+
"learn_te": true,
|
| 7 |
+
"n_heads": 16,
|
| 8 |
+
"n_layers": 12,
|
| 9 |
+
"resid_dropout_p": 0.2,
|
| 10 |
+
"s1_bits": 10,
|
| 11 |
+
"s2_bits": 10,
|
| 12 |
+
"token_dropout_p": 0.0
|
| 13 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6ddef65a1e2f7c328700f43420c7957f4fd8f793a7b441b269df4de5989ebf38
|
| 3 |
+
size 409264008
|