Update README.md
Browse files
README.md
CHANGED
|
@@ -77,6 +77,8 @@ score, joint_embedding = model.encode_multimodal(
|
|
| 77 |
|
| 78 |
There are two options to calculate semantic compatibility between an image and a text: cosine similarity and [Matching Score](#matching-score).
|
| 79 |
|
|
|
|
|
|
|
| 80 |
__Pros__:
|
| 81 |
|
| 82 |
- Computationally cheap.
|
|
@@ -88,7 +90,7 @@ __Cons__:
|
|
| 88 |
- Takes into account only coarse-grained features.
|
| 89 |
|
| 90 |
|
| 91 |
-
### Matching Score
|
| 92 |
|
| 93 |
Unlike cosine similarity, unimodal embedding are not enough.
|
| 94 |
Joint embedding will be needed and the resulting `score` will belong to the `[0, 1]` range, `1` meaning the absolute match.
|
|
|
|
| 77 |
|
| 78 |
There are two options to calculate semantic compatibility between an image and a text: cosine similarity and [Matching Score](#matching-score).
|
| 79 |
|
| 80 |
+
### Cosine Similarity
|
| 81 |
+
|
| 82 |
__Pros__:
|
| 83 |
|
| 84 |
- Computationally cheap.
|
|
|
|
| 90 |
- Takes into account only coarse-grained features.
|
| 91 |
|
| 92 |
|
| 93 |
+
### Matching Score
|
| 94 |
|
| 95 |
Unlike cosine similarity, unimodal embedding are not enough.
|
| 96 |
Joint embedding will be needed and the resulting `score` will belong to the `[0, 1]` range, `1` meaning the absolute match.
|