Update README.md
Browse files
README.md
CHANGED
|
@@ -10,6 +10,10 @@ tags:
|
|
| 10 |
---
|
| 11 |
# dark-potion-base-150M Model Card
|
| 12 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
This [Model2Vec](https://github.com/MinishLab/model2vec) model is a distilled version of a **dark secret model** 🤫. This model was created by [Bhavnick Minhas](https://github.com/bhavnicksm) using the [Model2Vec](https://github.com/MinishLab/model2vec) library, as a proof of concept, answering the question: "Can we distill static embeddings from a Sentence Embedding API, without any access to the original model?".
|
| 14 |
|
| 15 |
The original model is a proprietary model, and the only information we have about it is the tokenizer.json file, which contains the vocabulary and special tokens. We use this information to create a new tokenizer, and then use the Model2Vec library to distill a static embedding model from the API. Note that this is perfectly legal as embeddings are owned by the user as long as they are paid for, and the user is free to use them in any way they want.
|
|
|
|
| 10 |
---
|
| 11 |
# dark-potion-base-150M Model Card
|
| 12 |
|
| 13 |
+
<div align="center">
|
| 14 |
+
<img width="35%" alt="Model2Vec logo" src="https://raw.githubusercontent.com/MinishLab/model2vec/main/assets/images/logo_v2.png">
|
| 15 |
+
</div>
|
| 16 |
+
|
| 17 |
This [Model2Vec](https://github.com/MinishLab/model2vec) model is a distilled version of a **dark secret model** 🤫. This model was created by [Bhavnick Minhas](https://github.com/bhavnicksm) using the [Model2Vec](https://github.com/MinishLab/model2vec) library, as a proof of concept, answering the question: "Can we distill static embeddings from a Sentence Embedding API, without any access to the original model?".
|
| 18 |
|
| 19 |
The original model is a proprietary model, and the only information we have about it is the tokenizer.json file, which contains the vocabulary and special tokens. We use this information to create a new tokenizer, and then use the Model2Vec library to distill a static embedding model from the API. Note that this is perfectly legal as embeddings are owned by the user as long as they are paid for, and the user is free to use them in any way they want.
|