Commit
•
e537c76
1
Parent(s):
94df737
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,52 @@
|
|
1 |
---
|
2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
+
library_name: mlx-llm
|
4 |
+
language:
|
5 |
+
- en
|
6 |
+
tags:
|
7 |
+
- mlx
|
8 |
+
- exbert
|
9 |
+
datasets:
|
10 |
+
- bookcorpus
|
11 |
+
- wikipedia
|
12 |
---
|
13 |
+
|
14 |
+
|
15 |
+
# BERT base model (uncased) - MLX
|
16 |
+
|
17 |
+
Pretrained model on English language using a masked language modeling (MLM) objective. It was introduced in
|
18 |
+
[this paper](https://arxiv.org/abs/1810.04805) and first released in
|
19 |
+
[this repository](https://github.com/google-research/bert). This model is uncased: it does not make a difference
|
20 |
+
between english and English.
|
21 |
+
|
22 |
+
Disclaimer: The team releasing BERT did not write a model card for this model so this model card has been written by
|
23 |
+
the Hugging Face team.
|
24 |
+
|
25 |
+
## Model description
|
26 |
+
|
27 |
+
Please, refer to the [original model card](https://huggingface.co/bert-base-uncased) for more details on bert-base-uncased.
|
28 |
+
|
29 |
+
## Use it with mlx-llm
|
30 |
+
|
31 |
+
Install `mlx-llm` from GitHub.
|
32 |
+
```bash
|
33 |
+
git clone https://github.com/riccardomusmeci/mlx-llm
|
34 |
+
cd mlx-llm
|
35 |
+
pip install .
|
36 |
+
```
|
37 |
+
|
38 |
+
Run
|
39 |
+
```python
|
40 |
+
from mlx_llm.model import create_model
|
41 |
+
from transformers import BertTokenizer
|
42 |
+
import mlx.core as mx
|
43 |
+
|
44 |
+
model = create_model("bert-base-uncased") # it will download weights from this repository
|
45 |
+
tokenizer = BertTokenizer.from_pretrained("bert-large-uncased")
|
46 |
+
|
47 |
+
batch = ["This is an example of BERT working on MLX."]
|
48 |
+
tokens = tokenizer(batch, return_tensors="np", padding=True)
|
49 |
+
tokens = {key: mx.array(v) for key, v in tokens.items()}
|
50 |
+
|
51 |
+
output, pooled = model(**tokens)
|
52 |
+
```
|