Update README.md
Browse files
README.md
CHANGED
|
@@ -89,4 +89,11 @@ Then just call `push_to_hub` method or `save_pretrained` method if you want to s
|
|
| 89 |
model.push_to_hub("{your_username}/bloom-1b7-8bit")
|
| 90 |
```
|
| 91 |
|
| 92 |
-
That's it!
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 89 |
model.push_to_hub("{your_username}/bloom-1b7-8bit")
|
| 90 |
```
|
| 91 |
|
| 92 |
+
That's it!
|
| 93 |
+
|
| 94 |
+
## What is inside the model's `state_dict`?
|
| 95 |
+
|
| 96 |
+
Inside the state dict of the model (`pytorch_model.bin` file) you have
|
| 97 |
+
|
| 98 |
+
- the quantized `int8` weights
|
| 99 |
+
- the quantization statistics in `float16`
|