update readme.md
Browse files
README.md
CHANGED
|
@@ -45,9 +45,16 @@ print(embeddings)
|
|
| 45 |
```
|
| 46 |
|
| 47 |
### Important usage notes
|
| 48 |
-
- "ošišana
|
| 49 |
- The usage of uppercase letters for named entities can significantly improve search quality
|
| 50 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
|
| 52 |
## Evaluation
|
| 53 |
|
|
@@ -86,7 +93,6 @@ Evaluation datasets will be published as Part of [MTEB benchmark](https://huggin
|
|
| 86 |
|
| 87 |
If you have any question or sugestion related to this project, you can open an issue or pull request. You can also email me at novakzivanic@gmail.com
|
| 88 |
|
| 89 |
-
|
| 90 |
## Full Model Architecture
|
| 91 |
```
|
| 92 |
SentenceTransformer(
|
|
|
|
| 45 |
```
|
| 46 |
|
| 47 |
### Important usage notes
|
| 48 |
+
- "ošišana latinica" (usage of c instead of ć, etc...) significantly deacreases search quality
|
| 49 |
- The usage of uppercase letters for named entities can significantly improve search quality
|
| 50 |
|
| 51 |
+
## Training
|
| 52 |
+
|
| 53 |
+
- Embedić models are fine-tuned from multilingual-e5 models and they come in 3 sizes (small, base, large).
|
| 54 |
+
|
| 55 |
+
- Training is done on a single 4070ti super GPU
|
| 56 |
+
|
| 57 |
+
- 3-step training: distillation, training on (query, text) pairs and finally fine-tuning with triplets.
|
| 58 |
|
| 59 |
## Evaluation
|
| 60 |
|
|
|
|
| 93 |
|
| 94 |
If you have any question or sugestion related to this project, you can open an issue or pull request. You can also email me at novakzivanic@gmail.com
|
| 95 |
|
|
|
|
| 96 |
## Full Model Architecture
|
| 97 |
```
|
| 98 |
SentenceTransformer(
|