add model card
Browse files
README.md
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- vi
|
| 4 |
+
- ede
|
| 5 |
+
tags:
|
| 6 |
+
- cross-lingual-retrieval
|
| 7 |
+
- morpheme-tokenizer
|
| 8 |
+
- colbert
|
| 9 |
+
- EViRAL
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# ColBERT Transformer + Morpheme Tokenizer v4 — EViRAL
|
| 13 |
+
|
| 14 |
+
Task: Ede query → Vietnamese passage retrieval
|
| 15 |
+
|
| 16 |
+
## Eval Results
|
| 17 |
+
|
| 18 |
+
| Metric | Validation | Test |
|
| 19 |
+
|---------|-----------|--------|
|
| 20 |
+
| nDCG@1 | 0.0004 | 0.0005 |
|
| 21 |
+
| nDCG@5 | 0.0011 | 0.0013 |
|
| 22 |
+
| nDCG@10 | 0.0017 | 0.0022 |
|
| 23 |
+
| MRR@10 | 0.0020 | 0.0023 |
|
| 24 |
+
| R@50 | 0.0202 | 0.0210 |
|
| 25 |
+
| R@100 | 0.0345 | 0.0374 |
|
| 26 |
+
|
| 27 |
+
## Checkpoints
|
| 28 |
+
| file | description |
|
| 29 |
+
|---|---|
|
| 30 |
+
| mlm.pt | MLM ColBERT encoder weights |
|
| 31 |
+
| align.pt | cross-lingual aligned encoder |
|
| 32 |
+
| finetune.pt | contrastive fine-tuned encoder (best val) |
|