Update README.md
Browse files
README.md
CHANGED
|
@@ -30,7 +30,13 @@ samples_per_second: 2984.379
|
|
| 30 |
|
| 31 |
steps_per_second: 5.83
|
| 32 |
|
| 33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
|
| 35 |
## Usage
|
| 36 |
|
|
|
|
| 30 |
|
| 31 |
steps_per_second: 5.83
|
| 32 |
|
| 33 |
+
## Info
|
| 34 |
+
|
| 35 |
+
This bert-tiny finetuned on 1.5B tokens for embedding/feature extraction, for English and Brazillian Portuguese languages.
|
| 36 |
+
|
| 37 |
+
The output vector size is 128.
|
| 38 |
+
|
| 39 |
+
This model only has 4.4M params but the quality of the embeddings punch way above its size after tuning.
|
| 40 |
|
| 41 |
## Usage
|
| 42 |
|