Update README.md
Browse files
README.md
CHANGED
|
@@ -6,6 +6,11 @@ tags:
|
|
| 6 |
- sentence-similarity
|
| 7 |
language:
|
| 8 |
- en
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
# DunnBC22/sentence-t5-base-FT-Quora_Sentence_Similarity-LG
|
|
@@ -33,12 +38,32 @@ embeddings = model.encode(sentences)
|
|
| 33 |
print(embeddings)
|
| 34 |
```
|
| 35 |
|
| 36 |
-
|
| 37 |
-
|
| 38 |
## Evaluation Results
|
| 39 |
|
| 40 |
For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name=DunnBC22/sentence-t5-base-FT-Quora_Sentence_Similarity-LG)
|
| 41 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
|
| 43 |
## Training
|
| 44 |
The model was trained with the parameters:
|
|
@@ -79,7 +104,7 @@ One way to improve the results of this model is to use a larger checkpoint of T5
|
|
| 79 |
The larger checkpoints are:
|
| 80 |
|
| 81 |
| Checkpoint | # of Train Params |
|
| 82 |
-
|
| 83 |
| T5-Base | 220 Million* |
|
| 84 |
| T5-Large | 770 Million |
|
| 85 |
| T5-3B | 3 Billion |
|
|
@@ -98,4 +123,4 @@ SentenceTransformer(
|
|
| 98 |
|
| 99 |
## Citing & Authors
|
| 100 |
|
| 101 |
-
|
|
|
|
| 6 |
- sentence-similarity
|
| 7 |
language:
|
| 8 |
- en
|
| 9 |
+
metrics:
|
| 10 |
+
- accuracy
|
| 11 |
+
- f1
|
| 12 |
+
- recall
|
| 13 |
+
- precision
|
| 14 |
---
|
| 15 |
|
| 16 |
# DunnBC22/sentence-t5-base-FT-Quora_Sentence_Similarity-LG
|
|
|
|
| 38 |
print(embeddings)
|
| 39 |
```
|
| 40 |
|
|
|
|
|
|
|
| 41 |
## Evaluation Results
|
| 42 |
|
| 43 |
For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name=DunnBC22/sentence-t5-base-FT-Quora_Sentence_Similarity-LG)
|
| 44 |
|
| 45 |
+
| Metric | Measure | Value | Notes |
|
| 46 |
+
| :--------: | :--------: | :--------: | :--------: |
|
| 47 |
+
| Accuracy | **Cosine-Similarity** | 85.93 | Threshold: 0.8320 |
|
| 48 |
+
| F1 | Cosine-Similarity | 82.89 | Threshold: 0.8178 |
|
| 49 |
+
| Precision | Cosine-Similarity | 77.43 | - |
|
| 50 |
+
| Recall | Cosine-Similarity | 89.18 | - |
|
| 51 |
+
| Average Precision | Cosine-Similarity | 87.13 | - |
|
| 52 |
+
| Accuracy | **Manhattan-Distance** | 85.95 | Threshold: 12.7721 |
|
| 53 |
+
| F1 | Manhattan-Distance | 82.89 | Threshold: 13.5008 |
|
| 54 |
+
| Precision | Manhattan-Distance | 76.91 | - |
|
| 55 |
+
| Recall | Manhattan-Distance | 89.89 | - |
|
| 56 |
+
| Average Precision | Manhattan-Distance | 87.13 | - |
|
| 57 |
+
| Accuracy | **Euclidean-Distance** | 85.93 | Threshold: 0.5797 |
|
| 58 |
+
| F1 | Euclidean-Distance | 82.89 | Threshold: 0.6037 |
|
| 59 |
+
| Precision | Euclidean-Distance | 77.43 | - |
|
| 60 |
+
| Recall | Euclidean-Distance | 89.18 | - |
|
| 61 |
+
| Average Precision | Euclidean-Distance | 87.13 | - |
|
| 62 |
+
| Accuracy | **Dot-Product** | 85.93 | Threshold: 0.8320 |
|
| 63 |
+
| F1 | Dot-Product | 82.89 | Threshold: 0.8178 |
|
| 64 |
+
| Precision | Dot-Product | 77.43 | - |
|
| 65 |
+
| Recall | Dot-Product | 89.18 | - |
|
| 66 |
+
| Average Precision | Dot-Product | 87.14 | - |
|
| 67 |
|
| 68 |
## Training
|
| 69 |
The model was trained with the parameters:
|
|
|
|
| 104 |
The larger checkpoints are:
|
| 105 |
|
| 106 |
| Checkpoint | # of Train Params |
|
| 107 |
+
| :--------: | :---------------: |
|
| 108 |
| T5-Base | 220 Million* |
|
| 109 |
| T5-Large | 770 Million |
|
| 110 |
| T5-3B | 3 Billion |
|
|
|
|
| 123 |
|
| 124 |
## Citing & Authors
|
| 125 |
|
| 126 |
+
Dataset Source: https://www.kaggle.com/datasets/quora/question-pairs-dataset
|