add config.json, readme.md
Browse files- config.json +6 -0
- readme.md +11 -0
config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_type": "dia",
|
| 3 |
+
"quantized": true,
|
| 4 |
+
"quantization_type": "dynamic",
|
| 5 |
+
"original_model": "nari-labs/Dia-1.6B"
|
| 6 |
+
}
|
readme.md
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Quantized Dia 1.6B (INT8)
|
| 2 |
+
|
| 3 |
+
This is a dynamic int8 quantized version of [nari-labs/Dia-1.6B](https://huggingface.co/nari-labs/Dia-1.6B).
|
| 4 |
+
It uses dynamic quantization for lighter deployment and faster inference.
|
| 5 |
+
|
| 6 |
+
Original model: **float16**, ~6.4GB
|
| 7 |
+
Quantized model: **int8 dynamic**, ~6.4GB
|
| 8 |
+
|
| 9 |
+
Uploaded by [RobertAgee](https://github.com/RobertAgee) and [RobAgrees](https://huggingface.co/RobAgrees.
|
| 10 |
+
|
| 11 |
+
> Quantized automatically with PyTorch dynamic quantization in Google Colab.
|