Add new SentenceTransformer model.

README.md

@@ -46,7 +46,7 @@ tags:
 - feature-extraction
 - generated_from_trainer
 - dataset_size:560
-- loss:MultipleNegativesRankingLoss
+- loss:CoSENTLoss
 widget:
 - source_sentence: Let's search inside
   sentences:
@@ -84,109 +84,109 @@ model-index:
       type: custom-arc-semantics-data
     metrics:
     - type: cosine_accuracy
-      value: 0.85
+      value: 0.9285714285714286
       name: Cosine Accuracy
     - type: cosine_accuracy_threshold
-      value: 0.49632835388183594
+      value: 0.42927420139312744
       name: Cosine Accuracy Threshold
     - type: cosine_f1
-      value: 0.8727272727272727
+      value: 0.9425287356321839
       name: Cosine F1
     - type: cosine_f1_threshold
-      value: 0.48691314458847046
+      value: 0.2269928753376007
       name: Cosine F1 Threshold
     - type: cosine_precision
-      value: 0.8888888888888888
+      value: 0.9111111111111111
       name: Cosine Precision
     - type: cosine_recall
-      value: 0.8571428571428571
+      value: 0.9761904761904762
       name: Cosine Recall
     - type: cosine_ap
-      value: 0.927175101411552
+      value: 0.9720863676601571
       name: Cosine Ap
     - type: dot_accuracy
-      value: 0.85
+      value: 0.9285714285714286
       name: Dot Accuracy
     - type: dot_accuracy_threshold
-      value: 0.4963283836841583
+      value: 0.42927438020706177
       name: Dot Accuracy Threshold
     - type: dot_f1
-      value: 0.8727272727272727
+      value: 0.9425287356321839
       name: Dot F1
     - type: dot_f1_threshold
-      value: 0.48691320419311523
+      value: 0.22699296474456787
       name: Dot F1 Threshold
     - type: dot_precision
-      value: 0.8888888888888888
+      value: 0.9111111111111111
       name: Dot Precision
     - type: dot_recall
-      value: 0.8571428571428571
+      value: 0.9761904761904762
       name: Dot Recall
     - type: dot_ap
-      value: 0.927175101411552
+      value: 0.9720863676601571
       name: Dot Ap
     - type: manhattan_accuracy
-      value: 0.8428571428571429
+      value: 0.9285714285714286
       name: Manhattan Accuracy
     - type: manhattan_accuracy_threshold
-      value: 15.624195098876953
+      value: 16.630834579467773
       name: Manhattan Accuracy Threshold
     - type: manhattan_f1
-      value: 0.8681318681318683
+      value: 0.9431818181818182
       name: Manhattan F1
     - type: manhattan_f1_threshold
-      value: 18.23479461669922
+      value: 19.740108489990234
       name: Manhattan F1 Threshold
     - type: manhattan_precision
-      value: 0.8061224489795918
+      value: 0.9021739130434783
       name: Manhattan Precision
     - type: manhattan_recall
-      value: 0.9404761904761905
+      value: 0.9880952380952381
       name: Manhattan Recall
     - type: manhattan_ap
-      value: 0.9264219833665228
+      value: 0.9728353486982702
       name: Manhattan Ap
     - type: euclidean_accuracy
-      value: 0.85
+      value: 0.9285714285714286
       name: Euclidean Accuracy
     - type: euclidean_accuracy_threshold
-      value: 1.00364351272583
+      value: 1.068155288696289
       name: Euclidean Accuracy Threshold
     - type: euclidean_f1
-      value: 0.8727272727272727
+      value: 0.9425287356321839
       name: Euclidean F1
     - type: euclidean_f1_threshold
-      value: 1.0129987001419067
+      value: 1.2433418035507202
       name: Euclidean F1 Threshold
     - type: euclidean_precision
-      value: 0.8888888888888888
+      value: 0.9111111111111111
       name: Euclidean Precision
     - type: euclidean_recall
-      value: 0.8571428571428571
+      value: 0.9761904761904762
       name: Euclidean Recall
     - type: euclidean_ap
-      value: 0.927175101411552
+      value: 0.9720863676601571
       name: Euclidean Ap
     - type: max_accuracy
-      value: 0.85
+      value: 0.9285714285714286
       name: Max Accuracy
     - type: max_accuracy_threshold
-      value: 15.624195098876953
+      value: 16.630834579467773
       name: Max Accuracy Threshold
     - type: max_f1
-      value: 0.8727272727272727
+      value: 0.9431818181818182
       name: Max F1
     - type: max_f1_threshold
-      value: 18.23479461669922
+      value: 19.740108489990234
       name: Max F1 Threshold
     - type: max_precision
-      value: 0.8888888888888888
+      value: 0.9111111111111111
       name: Max Precision
     - type: max_recall
-      value: 0.9404761904761905
+      value: 0.9880952380952381
       name: Max Recall
     - type: max_ap
-      value: 0.927175101411552
+      value: 0.9728353486982702
       name: Max Ap
 ---
 
@@ -288,41 +288,41 @@ You can finetune this model on your own dataset.
 
 | Metric                       | Value      |
 |:-----------------------------|:-----------|
-| cosine_accuracy              | 0.85       |
-| cosine_accuracy_threshold    | 0.4963     |
-| cosine_f1                    | 0.8727     |
-| cosine_f1_threshold          | 0.4869     |
-| cosine_precision             | 0.8889     |
-| cosine_recall                | 0.8571     |
-| cosine_ap                    | 0.9272     |
-| dot_accuracy                 | 0.85       |
-| dot_accuracy_threshold       | 0.4963     |
-| dot_f1                       | 0.8727     |
-| dot_f1_threshold             | 0.4869     |
-| dot_precision                | 0.8889     |
-| dot_recall                   | 0.8571     |
-| dot_ap                       | 0.9272     |
-| manhattan_accuracy           | 0.8429     |
-| manhattan_accuracy_threshold | 15.6242    |
-| manhattan_f1                 | 0.8681     |
-| manhattan_f1_threshold       | 18.2348    |
-| manhattan_precision          | 0.8061     |
-| manhattan_recall             | 0.9405     |
-| manhattan_ap                 | 0.9264     |
-| euclidean_accuracy           | 0.85       |
-| euclidean_accuracy_threshold | 1.0036     |
-| euclidean_f1                 | 0.8727     |
-| euclidean_f1_threshold       | 1.013      |
-| euclidean_precision          | 0.8889     |
-| euclidean_recall             | 0.8571     |
-| euclidean_ap                 | 0.9272     |
-| max_accuracy                 | 0.85       |
-| max_accuracy_threshold       | 15.6242    |
-| max_f1                       | 0.8727     |
-| max_f1_threshold             | 18.2348    |
-| max_precision                | 0.8889     |
-| max_recall                   | 0.9405     |
-| **max_ap**                   | **0.9272** |
+| cosine_accuracy              | 0.9286     |
+| cosine_accuracy_threshold    | 0.4293     |
+| cosine_f1                    | 0.9425     |
+| cosine_f1_threshold          | 0.227      |
+| cosine_precision             | 0.9111     |
+| cosine_recall                | 0.9762     |
+| cosine_ap                    | 0.9721     |
+| dot_accuracy                 | 0.9286     |
+| dot_accuracy_threshold       | 0.4293     |
+| dot_f1                       | 0.9425     |
+| dot_f1_threshold             | 0.227      |
+| dot_precision                | 0.9111     |
+| dot_recall                   | 0.9762     |
+| dot_ap                       | 0.9721     |
+| manhattan_accuracy           | 0.9286     |
+| manhattan_accuracy_threshold | 16.6308    |
+| manhattan_f1                 | 0.9432     |
+| manhattan_f1_threshold       | 19.7401    |
+| manhattan_precision          | 0.9022     |
+| manhattan_recall             | 0.9881     |
+| manhattan_ap                 | 0.9728     |
+| euclidean_accuracy           | 0.9286     |
+| euclidean_accuracy_threshold | 1.0682     |
+| euclidean_f1                 | 0.9425     |
+| euclidean_f1_threshold       | 1.2433     |
+| euclidean_precision          | 0.9111     |
+| euclidean_recall             | 0.9762     |
+| euclidean_ap                 | 0.9721     |
+| max_accuracy                 | 0.9286     |
+| max_accuracy_threshold       | 16.6308    |
+| max_f1                       | 0.9432     |
+| max_f1_threshold             | 19.7401    |
+| max_precision                | 0.9111     |
+| max_recall                   | 0.9881     |
+| **max_ap**                   | **0.9728** |
 
 <!--
 ## Bias, Risks and Limitations
@@ -356,11 +356,11 @@ You can finetune this model on your own dataset.
   | <code>When it was dinner</code>                      | <code>Dinner time</code>                                                  | <code>1</code> |
   | <code>Did you cook chicken noodle last night?</code> | <code>Did you make chicken noodle for dinner?</code>                      | <code>1</code> |
   | <code>Someone who can change item</code>             | <code>Someone who uses magic that turns something into something. </code> | <code>1</code> |
-* Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
+* Loss: [<code>CoSENTLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cosentloss) with these parameters:
   ```json
   {
       "scale": 20.0,
-      "similarity_fct": "cos_sim"
+      "similarity_fct": "pairwise_cos_sim"
   }
   ```
 
@@ -382,11 +382,11 @@ You can finetune this model on your own dataset.
   | <code>Let's check inside</code>          | <code>Let's search inside</code>         | <code>1</code> |
   | <code>Sohpie, are you okay?</code>       | <code>Sophie Are you pressured?</code>   | <code>0</code> |
   | <code>This wine glass is related.</code> | <code>This sword looks important.</code> | <code>0</code> |
-* Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
+* Loss: [<code>CoSENTLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cosentloss) with these parameters:
   ```json
   {
       "scale": 20.0,
-      "similarity_fct": "cos_sim"
+      "similarity_fct": "pairwise_cos_sim"
   }
   ```
 
@@ -521,19 +521,19 @@ You can finetune this model on your own dataset.
 | Epoch | Step | Training Loss | loss   | custom-arc-semantics-data_max_ap |
 |:-----:|:----:|:-------------:|:------:|:--------------------------------:|
 | None  | 0    | -             | -      | 0.9254                           |
-| 1.0   | 70   | 1.1722        | 1.2175 | 0.9237                           |
-| 2.0   | 140  | 0.7774        | 1.0454 | 0.9291                           |
-| 3.0   | 210  | 0.4122        | 1.0024 | 0.9316                           |
-| 4.0   | 280  | 0.229         | 0.9819 | 0.9285                           |
-| 5.0   | 350  | 0.1509        | 0.9215 | 0.9321                           |
-| 6.0   | 420  | 0.0988        | 0.9119 | 0.9312                           |
-| 7.0   | 490  | 0.0772        | 0.8962 | 0.9303                           |
-| 8.0   | 560  | 0.0564        | 0.8905 | 0.9272                           |
-| 9.0   | 630  | 0.0449        | 0.8878 | 0.9266                           |
-| 10.0  | 700  | 0.037         | 0.8841 | 0.9273                           |
-| 11.0  | 770  | 0.0387        | 0.8881 | 0.9265                           |
-| 12.0  | 840  | 0.0332        | 0.8884 | 0.9274                           |
-| 13.0  | 910  | 0.032         | 0.8890 | 0.9272                           |
+| 1.0   | 70   | 2.9684        | 1.4087 | 0.9425                           |
+| 2.0   | 140  | 1.4461        | 1.0942 | 0.9629                           |
+| 3.0   | 210  | 0.6005        | 0.8398 | 0.9680                           |
+| 4.0   | 280  | 0.3021        | 0.7577 | 0.9703                           |
+| 5.0   | 350  | 0.2412        | 0.7216 | 0.9715                           |
+| 6.0   | 420  | 0.1816        | 0.7538 | 0.9722                           |
+| 7.0   | 490  | 0.1512        | 0.8049 | 0.9726                           |
+| 8.0   | 560  | 0.1208        | 0.7602 | 0.9726                           |
+| 9.0   | 630  | 0.0915        | 0.7286 | 0.9729                           |
+| 10.0  | 700  | 0.0553        | 0.7072 | 0.9729                           |
+| 11.0  | 770  | 0.0716        | 0.6984 | 0.9730                           |
+| 12.0  | 840  | 0.0297        | 0.7063 | 0.9725                           |
+| 13.0  | 910  | 0.0462        | 0.6997 | 0.9728                           |
 
 
 ### Framework Versions
@@ -562,15 +562,14 @@ You can finetune this model on your own dataset.
 }
 ```
 
-#### MultipleNegativesRankingLoss
+#### CoSENTLoss
 ```bibtex
-@misc{henderson2017efficient,
-    title={Efficient Natural Language Response Suggestion for Smart Reply}, 
-    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
-    year={2017},
-    eprint={1705.00652},
-    archivePrefix={arXiv},
-    primaryClass={cs.CL}
+@online{kexuefm-8847,
+    title={CoSENT: A more efficient sentence vector scheme than Sentence-BERT},
+    author={Su Jianlin},
+    year={2022},
+    month={Jan},
+    url={https://kexue.fm/archives/8847},
 }
 ```
 

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f724ea45fc6e76f2fe28ae0d75a450d3e7365c6fb93d8edd41724b13cde80da5
+oid sha256:2d9ab6b7472780e4b9271e02f535d125c33cef1b145ab2f8d3135ed97c72aea5
 size 90864192