Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
inference: false
|
| 3 |
+
datasets:
|
| 4 |
+
- answerdotai/MMARCO-japanese-32-scored-triplets
|
| 5 |
+
- miracl/miracl
|
| 6 |
+
- hotchpotch/JQaRA
|
| 7 |
+
- matsuxr/JaGovFaqs-22k
|
| 8 |
+
- unicamp-dl/mmarco
|
| 9 |
+
language:
|
| 10 |
+
- ja
|
| 11 |
+
pipeline_tag: sentence-similarity
|
| 12 |
+
tags:
|
| 13 |
+
- ColBERT
|
| 14 |
+
base_model:
|
| 15 |
+
- cl-tohoku/bert-base-japanese-v3
|
| 16 |
+
- bclavie/JaColBERT
|
| 17 |
+
license: mit
|
| 18 |
+
library_name: RAGatouille
|
| 19 |
+
---
|
| 20 |
+
|
| 21 |
+
Model weights for the final JaColBERTv2.5 checkpoint, using an entirely overhauled training recipe and trained on just 40% of the data of JaColBERTv2.
|
| 22 |
+
|
| 23 |
+
This model largely outperforms all previous approaches, including JaColBERTV2 multilingual models such as BGE-M3, on all datasets.
|
| 24 |
+
|
| 25 |
+
This page will be updated with the full details and the model report in the next few days.
|