Update README.md
Browse files
README.md
CHANGED
|
@@ -1,19 +1,34 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
|
|
|
| 2 |
base_model:
|
| 3 |
- Qwen/Qwen3-4B
|
| 4 |
-
pipeline_tag:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
---
|
|
|
|
| 6 |
|
| 7 |
-
|
| 8 |
|
| 9 |
-
This is a smaller version of ZeroEntropy Reranker v0.3.
|
| 10 |
|
| 11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
```python
|
| 14 |
from sentence_transformers import CrossEncoder
|
| 15 |
|
| 16 |
-
model = CrossEncoder("zeroentropy/
|
| 17 |
|
| 18 |
query_documents = [
|
| 19 |
("What is 2+2?", "4"),
|
|
@@ -22,4 +37,34 @@ query_documents = [
|
|
| 22 |
|
| 23 |
scores = model.predict(query_documents)
|
| 24 |
|
| 25 |
-
print(scores)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license: cc-by-nc-4.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
base_model:
|
| 6 |
- Qwen/Qwen3-4B
|
| 7 |
+
pipeline_tag: text-ranking
|
| 8 |
+
tags:
|
| 9 |
+
- finance
|
| 10 |
+
- legal
|
| 11 |
+
- code
|
| 12 |
+
- stem
|
| 13 |
+
- medical
|
| 14 |
---
|
| 15 |
+
# zerank-1: ZeroEntropy Inc.'s SoTA reranker
|
| 16 |
|
| 17 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 18 |
|
|
|
|
| 19 |
|
| 20 |
+
This model is the smaller version of [zeroentropy/zerank-1](https://huggingface.co/zeroentropy/zerank-1). This model is over 2x smaller, but maintains nearly the same standard of performance, continuing to outperform other popular rerankers.
|
| 21 |
+
|
| 22 |
+
It is an open-weights reranker model meant to be integrated into RAG applications to rerank results from preliminary search methods such as embeddings, BM25, and hybrid search.
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
## How to Use
|
| 27 |
|
| 28 |
```python
|
| 29 |
from sentence_transformers import CrossEncoder
|
| 30 |
|
| 31 |
+
model = CrossEncoder("zeroentropy/zerank-1-small", trust_remote_code=True)
|
| 32 |
|
| 33 |
query_documents = [
|
| 34 |
("What is 2+2?", "4"),
|
|
|
|
| 37 |
|
| 38 |
scores = model.predict(query_documents)
|
| 39 |
|
| 40 |
+
print(scores)
|
| 41 |
+
```
|
| 42 |
+
|
| 43 |
+
## Evaluations
|
| 44 |
+
|
| 45 |
+
Comparing NDCG@10 starting from top 100 documents by embedding (using text-3-embedding-small):
|
| 46 |
+
|
| 47 |
+
| Task | Embedding | cohere-rerank-v3.5 | Salesforce/Llama-rank-v1 | **zerank-1-small** | zerank-1 |
|
| 48 |
+
|----------------|-----------|--------------------|--------------------------|----------------|----------|
|
| 49 |
+
| Code | 0.678 | 0.724 | 0.694 | **0.730** | 0.754 |
|
| 50 |
+
| Conversational | 0.250 | 0.571 | 0.484 | **0.556** | 0.596 |
|
| 51 |
+
| Finance | 0.839 | 0.824 | 0.828 | **0.861** | 0.894 |
|
| 52 |
+
| Legal | 0.703 | 0.804 | 0.767 | **0.817** | 0.821 |
|
| 53 |
+
| Medical | 0.619 | 0.750 | 0.719 | **0.773** | 0.796 |
|
| 54 |
+
| STEM | 0.401 | 0.510 | 0.595 | **0.680** | 0.694 |
|
| 55 |
+
|
| 56 |
+
Comparing BM25 and Hybrid Search without and with zerank-1:
|
| 57 |
+
|
| 58 |
+
<img src="https://cdn-uploads.huggingface.co/production/uploads/67776f9dcd9c9435499eafc8/2GPVHFrI39FspnSNklhsM.png" alt="Description" width="400"/> <img src="https://cdn-uploads.huggingface.co/production/uploads/67776f9dcd9c9435499eafc8/dwYo2D7hoL8QiE8u3yqr9.png" alt="Description" width="400"/>
|
| 59 |
+
|
| 60 |
+
|
| 61 |
+
## Citation
|
| 62 |
+
|
| 63 |
+
**BibTeX:**
|
| 64 |
+
|
| 65 |
+
Coming soon!
|
| 66 |
+
|
| 67 |
+
**APA:**
|
| 68 |
+
|
| 69 |
+
Coming soon!
|
| 70 |
+
|