Update README.md
Browse files
README.md
CHANGED
|
@@ -56,7 +56,7 @@ They're simple to use, fast and very accurate.
|
|
| 56 |
|
| 57 |
Like many people, I was excited about the architecture and training uplift from the ModernBERT architecture (`answerdotai/ModernBERT-base`).
|
| 58 |
So I've applied it to the stsb cross encoder, which is a very handy model. Additionally, I've added
|
| 59 |
-
pretraining from
|
| 60 |
The inference performance efficiency, expanded context and simplicity make this a really nice platform as an evaluator model.
|
| 61 |
|
| 62 |
---
|
|
|
|
| 56 |
|
| 57 |
Like many people, I was excited about the architecture and training uplift from the ModernBERT architecture (`answerdotai/ModernBERT-base`).
|
| 58 |
So I've applied it to the stsb cross encoder, which is a very handy model. Additionally, I've added
|
| 59 |
+
pretraining from a much larger semi-synthetic dataset `dleemiller/wiki-sim` that targets this kind of objective.
|
| 60 |
The inference performance efficiency, expanded context and simplicity make this a really nice platform as an evaluator model.
|
| 61 |
|
| 62 |
---
|