Update README.md
Browse files
README.md
CHANGED
|
@@ -49,6 +49,8 @@ library_name: sentence-transformers
|
|
| 49 |
2. **Adjustable test-time compute scaling.** A single KaLM-Reranker-V1-R2 checkpoint can trade compute for quality by changing the compression ratio. It can also use a coarse-to-fine cascade: cheaply screen all candidates with heavily compressed document representations, then spend more compute only on the most promising candidates.
|
| 50 |
3. **Better multi-domain and multilingual ranking.** At the default 4x compression ratio, every R2 model improves on BEIR, while MIRACL average nDCG@10 increases by **+9.03**, **+7.17**, and **+4.85** points for Nano, Small, and Large, respectively.
|
| 51 |
|
|
|
|
|
|
|
| 52 |
| Capability | KaLM-Reranker-V1 | KaLM-Reranker-V1-R2 |
|
| 53 |
|:--|:--|:--|
|
| 54 |
| Validated MEP range | 1x-32x | **1x-128x** |
|
|
@@ -135,6 +137,7 @@ The reported sizes are **activated parameters**. Nano, Small, and Large are init
|
|
| 135 |
|
| 136 |
```
|
| 137 |
|
|
|
|
| 138 |
|
| 139 |
# Acknowledgements
|
| 140 |
We sincerely thank `jina-reranker-v3` and `Qwen3-Reranker` for their valuable inspiration and contributions to the reranking community, from which we have learned a lot.
|
|
|
|
| 49 |
2. **Adjustable test-time compute scaling.** A single KaLM-Reranker-V1-R2 checkpoint can trade compute for quality by changing the compression ratio. It can also use a coarse-to-fine cascade: cheaply screen all candidates with heavily compressed document representations, then spend more compute only on the most promising candidates.
|
| 50 |
3. **Better multi-domain and multilingual ranking.** At the default 4x compression ratio, every R2 model improves on BEIR, while MIRACL average nDCG@10 increases by **+9.03**, **+7.17**, and **+4.85** points for Nano, Small, and Large, respectively.
|
| 51 |
|
| 52 |
+
|
| 53 |
+
|
| 54 |
| Capability | KaLM-Reranker-V1 | KaLM-Reranker-V1-R2 |
|
| 55 |
|:--|:--|:--|
|
| 56 |
| Validated MEP range | 1x-32x | **1x-128x** |
|
|
|
|
| 137 |
|
| 138 |
```
|
| 139 |
|
| 140 |
+
|
| 141 |
|
| 142 |
# Acknowledgements
|
| 143 |
We sincerely thank `jina-reranker-v3` and `Qwen3-Reranker` for their valuable inspiration and contributions to the reranking community, from which we have learned a lot.
|