Hugging Face's logo

AITeamVN
/
Vietnamese_Embedding

Sentence Similarity
sentence-transformers
ONNX
Safetensors
Vietnamese
xlm-roberta
Embedding
text-embeddings-inference
Model card Files
xet
Community
3
AITeamVN commited on
Commit
17a88ad
·
verified ·
1 Parent(s): 49e3b82

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -8
README.md CHANGED
@@ -56,14 +56,14 @@ array([[0.66212064, 0.33066642],
56
  ### Evaluation:
57
 
58
  - Dataset: Entire training dataset of Legal Zalo 2021. Our model was not trained on this dataset.
59
- 79.443 93.242 95.369 97.403 86.717
60
- | Model | Accuracy@1 | Accuracy@3 | Accuracy@5 | Accuracy@10 | Accuracy@100 | MRR@10 |
61
- |----------------------|------------|------------|------------|-------------|-------------|--------------|
62
- | Vietnamese Reranker (Phase 2) | 0.7944 | 0.9324 | 0.9537 | 0.9740 | NA | 0.8672 |
63
- | Vietnamese_Embedding (Phase 2) | 0.7262 | 0.8927 | 0.9268 | 0.9578 | 0.9925 | 0.8149 |
64
- | Vietnamese_Embedding (public) | 0.7274 | 0.8992 | 0.9305 | 0.9568 | 0.9922 | 0.8181 |
65
- | Vietnamese-bi-encoder (BKAI) | 0.7109 | 0.8680 | 0.9014 | 0.9299 | 0.9772 | 0.7951 |
66
- | BGE-M3 | 0.5682 | 0.7728 | 0.8382 | 0.8921 | 0.9772 | 0.6822 |
67
 
68
  Vietnamese Reranker (Phase 2) and Vietnamese Reranker (Phased) was trained on 1100000 triplets. Although the score on the legal domain drops a bit on Vietnamese_Embedding (Phase 2), since this phase data is much larger, it is very good for other domains.
69
 
 
56
  ### Evaluation:
57
 
58
  - Dataset: Entire training dataset of Legal Zalo 2021. Our model was not trained on this dataset.
59
+
60
+ | Model | Accuracy@1 | Accuracy@3 | Accuracy@5 | Accuracy@10 | MRR@10 |
61
+ |----------------------|------------|------------|------------|-------------|--------------|
62
+ | Vietnamese Reranker (Phase 2) | 0.7944 | 0.9324 | 0.9537 | 0.9740 | 0.8672 |
63
+ | Vietnamese_Embedding (Phase 2) | 0.7262 | 0.8927 | 0.9268 | 0.9578 | 0.8149 |
64
+ | Vietnamese_Embedding (public) | 0.7274 | 0.8992 | 0.9305 | 0.9568 | 0.8181 |
65
+ | Vietnamese-bi-encoder (BKAI) | 0.7109 | 0.8680 | 0.9014 | 0.9299 | 0.7951 |
66
+ | BGE-M3 | 0.5682 | 0.7728 | 0.8382 | 0.8921 | 0.6822 |
67
 
68
  Vietnamese Reranker (Phase 2) and Vietnamese Reranker (Phased) was trained on 1100000 triplets. Although the score on the legal domain drops a bit on Vietnamese_Embedding (Phase 2), since this phase data is much larger, it is very good for other domains.
69