Add new SentenceTransformer model

1_Pooling/config.json

@@ -1,7 +1,7 @@
 {
-    "word_embedding_dimension": 768,
-    "pooling_mode_cls_token": true,
-    "pooling_mode_mean_tokens": false,
+    "word_embedding_dimension": 384,
+    "pooling_mode_cls_token": false,
+    "pooling_mode_mean_tokens": true,
     "pooling_mode_max_tokens": false,
     "pooling_mode_mean_sqrt_len_tokens": false,
     "pooling_mode_weightedmean_tokens": false,

README.md

@@ -7,7 +7,7 @@ tags:
 - generated_from_trainer
 - dataset_size:111468
 - loss:MultipleNegativesRankingLoss
-base_model: Alibaba-NLP/gte-modernbert-base
+base_model: thenlper/gte-small
 widget:
 - source_sentence: What is something you do (or don’t do), even though you feel conflicted
     about it?
@@ -60,7 +60,7 @@ metrics:
 - cosine_mrr@10
 - cosine_map@100
 model-index:
-- name: SentenceTransformer based on Alibaba-NLP/gte-modernbert-base
+- name: SentenceTransformer based on thenlper/gte-small
   results:
   - task:
       type: information-retrieval
@@ -70,49 +70,49 @@ model-index:
       type: NanoMSMARCO
     metrics:
     - type: cosine_accuracy@1
-      value: 0.16
+      value: 0.3
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.3
+      value: 0.58
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.32
+      value: 0.6
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.4
+      value: 0.68
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.16
+      value: 0.3
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.09999999999999998
+      value: 0.19333333333333333
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.064
+      value: 0.12000000000000002
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.04000000000000001
+      value: 0.068
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.16
+      value: 0.3
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.3
+      value: 0.58
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.32
+      value: 0.6
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.4
+      value: 0.68
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.2797381138841459
+      value: 0.4950369328373354
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.24177777777777776
+      value: 0.43527777777777776
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.25452652172848145
+      value: 0.4475531768839056
       name: Cosine Map@100
   - task:
       type: information-retrieval
@@ -122,49 +122,49 @@ model-index:
       type: NanoNQ
     metrics:
     - type: cosine_accuracy@1
-      value: 0.1
+      value: 0.26
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.12
+      value: 0.48
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.16
+      value: 0.52
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.26
+      value: 0.64
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.1
+      value: 0.26
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.04
+      value: 0.16666666666666663
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.032
+      value: 0.10800000000000001
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.026000000000000002
+      value: 0.066
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.08
+      value: 0.24
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.09
+      value: 0.45
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.13
+      value: 0.49
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.23
+      value: 0.6
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.13826388457522362
+      value: 0.4279054208986469
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.12794444444444442
+      value: 0.3892142857142856
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.10725400911912068
+      value: 0.3750113241088494
       name: Cosine Map@100
   - task:
       type: nano-beir
@@ -174,63 +174,63 @@ model-index:
       type: NanoBEIR_mean
     metrics:
     - type: cosine_accuracy@1
-      value: 0.13
+      value: 0.28
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.21
+      value: 0.53
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.24
+      value: 0.56
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.33
+      value: 0.66
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.13
+      value: 0.28
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.06999999999999999
+      value: 0.18
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.048
+      value: 0.11400000000000002
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.033
+      value: 0.067
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.12
+      value: 0.27
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.195
+      value: 0.515
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.225
+      value: 0.5449999999999999
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.315
+      value: 0.64
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.20900099922968474
+      value: 0.46147117686799116
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.1848611111111111
+      value: 0.4122460317460317
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.18089026542380107
+      value: 0.4112822504963775
       name: Cosine Map@100
 ---
 
-# SentenceTransformer based on Alibaba-NLP/gte-modernbert-base
+# SentenceTransformer based on thenlper/gte-small
 
-This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [Alibaba-NLP/gte-modernbert-base](https://huggingface.co/Alibaba-NLP/gte-modernbert-base). It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
+This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [thenlper/gte-small](https://huggingface.co/thenlper/gte-small). It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
 
 ## Model Details
 
 ### Model Description
 - **Model Type:** Sentence Transformer
-- **Base model:** [Alibaba-NLP/gte-modernbert-base](https://huggingface.co/Alibaba-NLP/gte-modernbert-base) <!-- at revision e7f32e3c00f91d699e8c43b53106206bcc72bb22 -->
+- **Base model:** [thenlper/gte-small](https://huggingface.co/thenlper/gte-small) <!-- at revision 17e1f347d17fe144873b1201da91788898c639cd -->
 - **Maximum Sequence Length:** 128 tokens
-- **Output Dimensionality:** 768 dimensions
+- **Output Dimensionality:** 384 dimensions
 - **Similarity Function:** Cosine Similarity
 <!-- - **Training Dataset:** Unknown -->
 <!-- - **Language:** Unknown -->
@@ -246,8 +246,9 @@ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [A
 
 ```
 SentenceTransformer(
-  (0): Transformer({'max_seq_length': 128, 'do_lower_case': False, 'architecture': 'ModernBertModel'})
-  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
+  (0): Transformer({'max_seq_length': 128, 'do_lower_case': False, 'architecture': 'BertModel'})
+  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
+  (2): Normalize()
 )
 ```
 
@@ -275,14 +276,14 @@ sentences = [
 ]
 embeddings = model.encode(sentences)
 print(embeddings.shape)
-# [3, 768]
+# [3, 384]
 
 # Get the similarity scores for the embeddings
 similarities = model.similarity(embeddings, embeddings)
 print(similarities)
-# tensor([[1.0000, 1.0000, 0.1015],
-#         [1.0000, 1.0000, 0.1015],
-#         [0.1015, 0.1015, 1.0000]])
+# tensor([[1.0000, 1.0000, 0.3917],
+#         [1.0000, 1.0000, 0.3917],
+#         [0.3917, 0.3917, 1.0000]])
 ```
 
 <!--
@@ -320,21 +321,21 @@ You can finetune this model on your own dataset.
 
 | Metric              | NanoMSMARCO | NanoNQ     |
 |:--------------------|:------------|:-----------|
-| cosine_accuracy@1   | 0.16        | 0.1        |
-| cosine_accuracy@3   | 0.3         | 0.12       |
-| cosine_accuracy@5   | 0.32        | 0.16       |
-| cosine_accuracy@10  | 0.4         | 0.26       |
-| cosine_precision@1  | 0.16        | 0.1        |
-| cosine_precision@3  | 0.1         | 0.04       |
-| cosine_precision@5  | 0.064       | 0.032      |
-| cosine_precision@10 | 0.04        | 0.026      |
-| cosine_recall@1     | 0.16        | 0.08       |
-| cosine_recall@3     | 0.3         | 0.09       |
-| cosine_recall@5     | 0.32        | 0.13       |
-| cosine_recall@10    | 0.4         | 0.23       |
-| **cosine_ndcg@10**  | **0.2797**  | **0.1383** |
-| cosine_mrr@10       | 0.2418      | 0.1279     |
-| cosine_map@100      | 0.2545      | 0.1073     |
+| cosine_accuracy@1   | 0.3         | 0.26       |
+| cosine_accuracy@3   | 0.58        | 0.48       |
+| cosine_accuracy@5   | 0.6         | 0.52       |
+| cosine_accuracy@10  | 0.68        | 0.64       |
+| cosine_precision@1  | 0.3         | 0.26       |
+| cosine_precision@3  | 0.1933      | 0.1667     |
+| cosine_precision@5  | 0.12        | 0.108      |
+| cosine_precision@10 | 0.068       | 0.066      |
+| cosine_recall@1     | 0.3         | 0.24       |
+| cosine_recall@3     | 0.58        | 0.45       |
+| cosine_recall@5     | 0.6         | 0.49       |
+| cosine_recall@10    | 0.68        | 0.6        |
+| **cosine_ndcg@10**  | **0.495**   | **0.4279** |
+| cosine_mrr@10       | 0.4353      | 0.3892     |
+| cosine_map@100      | 0.4476      | 0.375      |
 
 #### Nano BEIR
 
@@ -350,23 +351,23 @@ You can finetune this model on your own dataset.
   }
   ```
 
-| Metric              | Value     |
-|:--------------------|:----------|
-| cosine_accuracy@1   | 0.13      |
-| cosine_accuracy@3   | 0.21      |
-| cosine_accuracy@5   | 0.24      |
-| cosine_accuracy@10  | 0.33      |
-| cosine_precision@1  | 0.13      |
-| cosine_precision@3  | 0.07      |
-| cosine_precision@5  | 0.048     |
-| cosine_precision@10 | 0.033     |
-| cosine_recall@1     | 0.12      |
-| cosine_recall@3     | 0.195     |
-| cosine_recall@5     | 0.225     |
-| cosine_recall@10    | 0.315     |
-| **cosine_ndcg@10**  | **0.209** |
-| cosine_mrr@10       | 0.1849    |
-| cosine_map@100      | 0.1809    |
+| Metric              | Value      |
+|:--------------------|:-----------|
+| cosine_accuracy@1   | 0.28       |
+| cosine_accuracy@3   | 0.53       |
+| cosine_accuracy@5   | 0.56       |
+| cosine_accuracy@10  | 0.66       |
+| cosine_precision@1  | 0.28       |
+| cosine_precision@3  | 0.18       |
+| cosine_precision@5  | 0.114      |
+| cosine_precision@10 | 0.067      |
+| cosine_recall@1     | 0.27       |
+| cosine_recall@3     | 0.515      |
+| cosine_recall@5     | 0.545      |
+| cosine_recall@10    | 0.64       |
+| **cosine_ndcg@10**  | **0.4615** |
+| cosine_mrr@10       | 0.4122     |
+| cosine_map@100      | 0.4113     |
 
 <!--
 ## Bias, Risks and Limitations
@@ -389,10 +390,10 @@ You can finetune this model on your own dataset.
 * Size: 111,468 training samples
 * Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
 * Approximate statistics based on the first 1000 samples:
-  |         | anchor                                                                           | positive                                                                          | negative                                                                          |
-  |:--------|:---------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|
-  | type    | string                                                                           | string                                                                            | string                                                                            |
-  | details | <ul><li>min: 6 tokens</li><li>mean: 16.1 tokens</li><li>max: 68 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 16.16 tokens</li><li>max: 68 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 17.35 tokens</li><li>max: 76 tokens</li></ul> |
+  |         | anchor                                                                            | positive                                                                          | negative                                                                          |
+  |:--------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|
+  | type    | string                                                                            | string                                                                            | string                                                                            |
+  | details | <ul><li>min: 6 tokens</li><li>mean: 16.11 tokens</li><li>max: 71 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 16.16 tokens</li><li>max: 71 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 17.35 tokens</li><li>max: 76 tokens</li></ul> |
 * Samples:
   | anchor                                                                                | positive                                                                              | negative                                                                                 |
   |:--------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------|
@@ -415,10 +416,10 @@ You can finetune this model on your own dataset.
 * Size: 12,386 evaluation samples
 * Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
 * Approximate statistics based on the first 1000 samples:
-  |         | anchor                                                                           | positive                                                                          | negative                                                                          |
-  |:--------|:---------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|
-  | type    | string                                                                           | string                                                                            | string                                                                            |
-  | details | <ul><li>min: 6 tokens</li><li>mean: 16.3 tokens</li><li>max: 62 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 16.36 tokens</li><li>max: 62 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 17.41 tokens</li><li>max: 65 tokens</li></ul> |
+  |         | anchor                                                                            | positive                                                                          | negative                                                                          |
+  |:--------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|
+  | type    | string                                                                            | string                                                                            | string                                                                            |
+  | details | <ul><li>min: 6 tokens</li><li>mean: 16.22 tokens</li><li>max: 62 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 16.28 tokens</li><li>max: 62 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 17.39 tokens</li><li>max: 66 tokens</li></ul> |
 * Samples:
   | anchor                                                                                                                              | positive                                                                                                                            | negative                                                                     |
   |:------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------|
@@ -583,19 +584,19 @@ You can finetune this model on your own dataset.
 ### Training Logs
 | Epoch  | Step | Training Loss | Validation Loss | NanoMSMARCO_cosine_ndcg@10 | NanoNQ_cosine_ndcg@10 | NanoBEIR_mean_cosine_ndcg@10 |
 |:------:|:----:|:-------------:|:---------------:|:--------------------------:|:---------------------:|:----------------------------:|
-| 0      | 0    | -             | 2.1536          | 0.6530                     | 0.6552                | 0.6541                       |
-| 0.2874 | 250  | 0.8149        | 0.3782          | 0.5748                     | 0.6239                | 0.5994                       |
-| 0.5747 | 500  | 0.3649        | 0.3561          | 0.5529                     | 0.5583                | 0.5556                       |
-| 0.8621 | 750  | 0.3496        | 0.3456          | 0.5747                     | 0.4299                | 0.5023                       |
-| 1.1494 | 1000 | 0.3364        | 0.3416          | 0.5479                     | 0.4669                | 0.5074                       |
-| 1.4368 | 1250 | 0.3231        | 0.3385          | 0.4802                     | 0.4214                | 0.4508                       |
-| 1.7241 | 1500 | 0.3222        | 0.3349          | 0.4694                     | 0.3547                | 0.4120                       |
-| 2.0115 | 1750 | 0.3187        | 0.3340          | 0.4398                     | 0.2723                | 0.3560                       |
-| 2.2989 | 2000 | 0.3051        | 0.3347          | 0.3702                     | 0.2144                | 0.2923                       |
-| 2.5862 | 2250 | 0.3039        | 0.3339          | 0.4180                     | 0.2575                | 0.3377                       |
-| 2.8736 | 2500 | 0.2994        | 0.3328          | 0.3687                     | 0.2662                | 0.3174                       |
-| 3.1609 | 2750 | 0.2969        | 0.3339          | 0.2906                     | 0.1283                | 0.2094                       |
-| 3.4483 | 3000 | 0.2933        | 0.3343          | 0.2797                     | 0.1383                | 0.2090                       |
+| 0      | 0    | -             | 3.6560          | 0.6259                     | 0.6583                | 0.6421                       |
+| 0.2874 | 250  | 2.1436        | 0.4823          | 0.5264                     | 0.5634                | 0.5449                       |
+| 0.5747 | 500  | 0.5891        | 0.4299          | 0.5280                     | 0.5051                | 0.5165                       |
+| 0.8621 | 750  | 0.5393        | 0.4123          | 0.5246                     | 0.4755                | 0.5001                       |
+| 1.1494 | 1000 | 0.5173        | 0.4027          | 0.5068                     | 0.4549                | 0.4809                       |
+| 1.4368 | 1250 | 0.5022        | 0.3954          | 0.5055                     | 0.4513                | 0.4784                       |
+| 1.7241 | 1500 | 0.4958        | 0.3909          | 0.5033                     | 0.4466                | 0.4749                       |
+| 2.0115 | 1750 | 0.4908        | 0.3890          | 0.4897                     | 0.4416                | 0.4656                       |
+| 2.2989 | 2000 | 0.4824        | 0.3859          | 0.4912                     | 0.4359                | 0.4636                       |
+| 2.5862 | 2250 | 0.4797        | 0.3847          | 0.4987                     | 0.4387                | 0.4687                       |
+| 2.8736 | 2500 | 0.4728        | 0.3834          | 0.4969                     | 0.4256                | 0.4613                       |
+| 3.1609 | 2750 | 0.4721        | 0.3824          | 0.4863                     | 0.4279                | 0.4571                       |
+| 3.4483 | 3000 | 0.4694        | 0.3822          | 0.4950                     | 0.4279                | 0.4615                       |
 
 
 ### Framework Versions

config_sentence_transformers.json

@@ -1,4 +1,5 @@
 {
+  "model_type": "SentenceTransformer",
   "__version__": {
     "sentence_transformers": "5.2.0",
     "transformers": "4.57.3",
@@ -9,6 +10,5 @@
     "document": ""
   },
   "default_prompt_name": null,
-  "similarity_fn_name": "cosine",
-  "model_type": "SentenceTransformer"
+  "similarity_fn_name": "cosine"
 }

modules.json

@@ -10,5 +10,11 @@
     "name": "1",
     "path": "1_Pooling",
     "type": "sentence_transformers.models.Pooling"
+  },
+  {
+    "idx": 2,
+    "name": "2",
+    "path": "2_Normalize",
+    "type": "sentence_transformers.models.Normalize"
   }
 ]