Commit
•
9c4dbc1
1
Parent(s):
d93f1b8
Update README.md
Browse files
README.md
CHANGED
|
@@ -53,6 +53,8 @@ datasets:
|
|
| 53 |
Multilingual OpenFlamingo is a multilingual version of [OpenFlamingo](https://arxiv.org/abs/2308.01390) trained on [mOSCAR](https://arxiv.org/abs/2406.08707) and a translated version of [LAION-400M](https://arxiv.org/abs/2111.02114). The model was trained on 43 languages and is based on `google/gemma-2b`.
|
| 54 |
Multilingual OpenFlamingo models process arbitrarily interleaved sequences of images and text to output text in multiple languages. The model will output the language provided in the prompt, no special token for specifying the language is required.
|
| 55 |
|
|
|
|
|
|
|
| 56 |
## Installation
|
| 57 |
```
|
| 58 |
git clone https://github.com/MatthieuFP/open_flamingo
|
|
|
|
| 53 |
Multilingual OpenFlamingo is a multilingual version of [OpenFlamingo](https://arxiv.org/abs/2308.01390) trained on [mOSCAR](https://arxiv.org/abs/2406.08707) and a translated version of [LAION-400M](https://arxiv.org/abs/2111.02114). The model was trained on 43 languages and is based on `google/gemma-2b`.
|
| 54 |
Multilingual OpenFlamingo models process arbitrarily interleaved sequences of images and text to output text in multiple languages. The model will output the language provided in the prompt, no special token for specifying the language is required.
|
| 55 |
|
| 56 |
+
Multilingual OpenFlamingo is only available for research purpose. We did not conduct any safety alignment training so the model could output harmful content if prompted to.
|
| 57 |
+
|
| 58 |
## Installation
|
| 59 |
```
|
| 60 |
git clone https://github.com/MatthieuFP/open_flamingo
|