Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: mlx-community/DeepSeek-R1-Distill-Qwen-1.5B-4bit
|
| 3 |
+
library_name: mlx
|
| 4 |
+
tags:
|
| 5 |
+
- lora
|
| 6 |
+
- mlx
|
| 7 |
+
- finance
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# bha6kar/finance-lora-deepseek-r1-1.5b
|
| 11 |
+
|
| 12 |
+
LoRA adapter fine-tuning [`mlx-community/DeepSeek-R1-Distill-Qwen-1.5B-4bit`](https://huggingface.co/mlx-community/DeepSeek-R1-Distill-Qwen-1.5B-4bit) on the [`gbharti/finance-alpaca`](https://huggingface.co/datasets/gbharti/finance-alpaca) financial instruction dataset with [MLX](https://github.com/ml-explore/mlx).
|
| 13 |
+
|
| 14 |
+
- Test perplexity: 10.846 (base) to 9.632 (tuned), across 3 seed(s).
|
| 15 |
+
|
| 16 |
+
## Use
|
| 17 |
+
|
| 18 |
+
```python
|
| 19 |
+
from mlx_lm import load, generate
|
| 20 |
+
model, tok = load("mlx-community/DeepSeek-R1-Distill-Qwen-1.5B-4bit", adapter_path="<downloaded-adapter-dir>")
|
| 21 |
+
```
|
| 22 |
+
|
| 23 |
+
Produced by [slm-training](https://github.com/bha6kar/slm-training).
|