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:111470
 - loss:MultipleNegativesRankingLoss
-base_model: Alibaba-NLP/gte-modernbert-base
+base_model: thenlper/gte-small
 widget:
 - source_sentence: why are some rocks radioactive
   sentences:
@@ -106,7 +106,7 @@ 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
@@ -116,49 +116,49 @@ model-index:
       type: NanoMSMARCO
     metrics:
     - type: cosine_accuracy@1
-      value: 0.42
+      value: 0.4
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.64
+      value: 0.5
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.78
+      value: 0.58
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.84
+      value: 0.68
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.42
+      value: 0.4
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.21333333333333332
+      value: 0.16666666666666669
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.156
+      value: 0.11600000000000002
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.08399999999999999
+      value: 0.068
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.42
+      value: 0.4
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.64
+      value: 0.5
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.78
+      value: 0.58
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.84
+      value: 0.68
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.6273713143801162
+      value: 0.523462400095123
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.5593571428571429
+      value: 0.4752380952380952
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.567451526639622
+      value: 0.48914996910623887
       name: Cosine Map@100
   - task:
       type: information-retrieval
@@ -168,49 +168,49 @@ model-index:
       type: NanoNQ
     metrics:
     - type: cosine_accuracy@1
-      value: 0.44
+      value: 0.48
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
       value: 0.56
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.62
+      value: 0.64
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
       value: 0.74
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.44
+      value: 0.48
       name: Cosine Precision@1
     - type: cosine_precision@3
       value: 0.18666666666666665
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.128
+      value: 0.132
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.08
+      value: 0.078
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.4
+      value: 0.45
       name: Cosine Recall@1
     - type: cosine_recall@3
       value: 0.52
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.59
+      value: 0.61
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.71
+      value: 0.7
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.5468372621429358
+      value: 0.5662755634296394
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.5185555555555555
+      value: 0.5436031746031746
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.49953000242452567
+      value: 0.5289614062254305
       name: Cosine Map@100
   - task:
       type: nano-beir
@@ -220,63 +220,63 @@ model-index:
       type: NanoBEIR_mean
     metrics:
     - type: cosine_accuracy@1
-      value: 0.43
+      value: 0.44
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.6000000000000001
+      value: 0.53
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.7
+      value: 0.61
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.79
+      value: 0.71
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.43
+      value: 0.44
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.19999999999999998
+      value: 0.17666666666666667
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.14200000000000002
+      value: 0.12400000000000001
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.08199999999999999
+      value: 0.07300000000000001
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.41000000000000003
+      value: 0.42500000000000004
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.5800000000000001
+      value: 0.51
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.685
+      value: 0.595
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.7749999999999999
+      value: 0.69
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.587104288261526
+      value: 0.5448689817623812
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.5389563492063492
+      value: 0.5094206349206349
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.5334907645320738
+      value: 0.5090556876658348
       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 -->
@@ -292,8 +292,9 @@ 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()
 )
 ```
 
@@ -321,14 +322,14 @@ 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, 0.3177],
-#         [1.0000, 1.0000, 0.3177],
-#         [0.3177, 0.3177, 1.0000]])
+# tensor([[1.0001, 1.0001, 0.9810],
+#         [1.0001, 1.0001, 0.9810],
+#         [0.9810, 0.9810, 1.0000]])
 ```
 
 <!--
@@ -366,21 +367,21 @@ You can finetune this model on your own dataset.
 
 | Metric              | NanoMSMARCO | NanoNQ     |
 |:--------------------|:------------|:-----------|
-| cosine_accuracy@1   | 0.42        | 0.44       |
-| cosine_accuracy@3   | 0.64        | 0.56       |
-| cosine_accuracy@5   | 0.78        | 0.62       |
-| cosine_accuracy@10  | 0.84        | 0.74       |
-| cosine_precision@1  | 0.42        | 0.44       |
-| cosine_precision@3  | 0.2133      | 0.1867     |
-| cosine_precision@5  | 0.156       | 0.128      |
-| cosine_precision@10 | 0.084       | 0.08       |
-| cosine_recall@1     | 0.42        | 0.4        |
-| cosine_recall@3     | 0.64        | 0.52       |
-| cosine_recall@5     | 0.78        | 0.59       |
-| cosine_recall@10    | 0.84        | 0.71       |
-| **cosine_ndcg@10**  | **0.6274**  | **0.5468** |
-| cosine_mrr@10       | 0.5594      | 0.5186     |
-| cosine_map@100      | 0.5675      | 0.4995     |
+| cosine_accuracy@1   | 0.4         | 0.48       |
+| cosine_accuracy@3   | 0.5         | 0.56       |
+| cosine_accuracy@5   | 0.58        | 0.64       |
+| cosine_accuracy@10  | 0.68        | 0.74       |
+| cosine_precision@1  | 0.4         | 0.48       |
+| cosine_precision@3  | 0.1667      | 0.1867     |
+| cosine_precision@5  | 0.116       | 0.132      |
+| cosine_precision@10 | 0.068       | 0.078      |
+| cosine_recall@1     | 0.4         | 0.45       |
+| cosine_recall@3     | 0.5         | 0.52       |
+| cosine_recall@5     | 0.58        | 0.61       |
+| cosine_recall@10    | 0.68        | 0.7        |
+| **cosine_ndcg@10**  | **0.5235**  | **0.5663** |
+| cosine_mrr@10       | 0.4752      | 0.5436     |
+| cosine_map@100      | 0.4891      | 0.529      |
 
 #### Nano BEIR
 
@@ -398,21 +399,21 @@ You can finetune this model on your own dataset.
 
 | Metric              | Value      |
 |:--------------------|:-----------|
-| cosine_accuracy@1   | 0.43       |
-| cosine_accuracy@3   | 0.6        |
-| cosine_accuracy@5   | 0.7        |
-| cosine_accuracy@10  | 0.79       |
-| cosine_precision@1  | 0.43       |
-| cosine_precision@3  | 0.2        |
-| cosine_precision@5  | 0.142      |
-| cosine_precision@10 | 0.082      |
-| cosine_recall@1     | 0.41       |
-| cosine_recall@3     | 0.58       |
-| cosine_recall@5     | 0.685      |
-| cosine_recall@10    | 0.775      |
-| **cosine_ndcg@10**  | **0.5871** |
-| cosine_mrr@10       | 0.539      |
-| cosine_map@100      | 0.5335     |
+| cosine_accuracy@1   | 0.44       |
+| cosine_accuracy@3   | 0.53       |
+| cosine_accuracy@5   | 0.61       |
+| cosine_accuracy@10  | 0.71       |
+| cosine_precision@1  | 0.44       |
+| cosine_precision@3  | 0.1767     |
+| cosine_precision@5  | 0.124      |
+| cosine_precision@10 | 0.073      |
+| cosine_recall@1     | 0.425      |
+| cosine_recall@3     | 0.51       |
+| cosine_recall@5     | 0.595      |
+| cosine_recall@10    | 0.69       |
+| **cosine_ndcg@10**  | **0.5449** |
+| cosine_mrr@10       | 0.5094     |
+| cosine_map@100      | 0.5091     |
 
 <!--
 ## Bias, Risks and Limitations
@@ -438,7 +439,7 @@ You can finetune this model on your own dataset.
   |         | anchor                                                                            | positive                                                                           | negative                                                                           |
   |:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
   | type    | string                                                                            | string                                                                             | string                                                                             |
-  | details | <ul><li>min: 4 tokens</li><li>mean: 11.17 tokens</li><li>max: 59 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 68.53 tokens</li><li>max: 128 tokens</li></ul> | <ul><li>min: 7 tokens</li><li>mean: 67.56 tokens</li><li>max: 128 tokens</li></ul> |
+  | details | <ul><li>min: 4 tokens</li><li>mean: 10.95 tokens</li><li>max: 60 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 67.57 tokens</li><li>max: 128 tokens</li></ul> | <ul><li>min: 7 tokens</li><li>mean: 66.64 tokens</li><li>max: 128 tokens</li></ul> |
 * Samples:
   | anchor                                                                | positive                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               | negative                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                     |
   |:----------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
@@ -464,7 +465,7 @@ You can finetune this model on your own dataset.
   |         | anchor                                                                            | positive                                                                           | negative                                                                           |
   |:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
   | type    | string                                                                            | string                                                                             | string                                                                             |
-  | details | <ul><li>min: 4 tokens</li><li>mean: 11.35 tokens</li><li>max: 64 tokens</li></ul> | <ul><li>min: 7 tokens</li><li>mean: 68.67 tokens</li><li>max: 128 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 67.03 tokens</li><li>max: 128 tokens</li></ul> |
+  | details | <ul><li>min: 4 tokens</li><li>mean: 11.11 tokens</li><li>max: 66 tokens</li></ul> | <ul><li>min: 7 tokens</li><li>mean: 67.99 tokens</li><li>max: 128 tokens</li></ul> | <ul><li>min: 7 tokens</li><li>mean: 66.08 tokens</li><li>max: 128 tokens</li></ul> |
 * Samples:
   | anchor                                                                                                    | positive                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                | negative                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      |
   |:----------------------------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
@@ -486,9 +487,9 @@ You can finetune this model on your own dataset.
 - `eval_strategy`: steps
 - `per_device_train_batch_size`: 128
 - `per_device_eval_batch_size`: 128
-- `learning_rate`: 4e-05
-- `weight_decay`: 0.01
-- `max_steps`: 703
+- `learning_rate`: 8e-05
+- `weight_decay`: 0.005
+- `max_steps`: 3375
 - `warmup_ratio`: 0.1
 - `fp16`: True
 - `dataloader_drop_last`: True
@@ -515,14 +516,14 @@ You can finetune this model on your own dataset.
 - `gradient_accumulation_steps`: 1
 - `eval_accumulation_steps`: None
 - `torch_empty_cache_steps`: None
-- `learning_rate`: 4e-05
-- `weight_decay`: 0.01
+- `learning_rate`: 8e-05
+- `weight_decay`: 0.005
 - `adam_beta1`: 0.9
 - `adam_beta2`: 0.999
 - `adam_epsilon`: 1e-08
 - `max_grad_norm`: 1.0
 - `num_train_epochs`: 3.0
-- `max_steps`: 703
+- `max_steps`: 3375
 - `lr_scheduler_type`: linear
 - `lr_scheduler_kwargs`: {}
 - `warmup_ratio`: 0.1
@@ -629,9 +630,20 @@ 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    | -             | 2.5772          | 0.6530                     | 0.6552                | 0.6541                       |
-| 0.2874 | 250  | 0.9649        | 0.7574          | 0.6170                     | 0.5720                | 0.5945                       |
-| 0.5747 | 500  | 0.7456        | 0.7372          | 0.6274                     | 0.5468                | 0.5871                       |
+| 0      | 0    | -             | 4.0678          | 0.6259                     | 0.6583                | 0.6421                       |
+| 0.2874 | 250  | 1.9612        | 0.8621          | 0.5278                     | 0.5499                | 0.5389                       |
+| 0.5747 | 500  | 0.9657        | 0.8137          | 0.5173                     | 0.5494                | 0.5333                       |
+| 0.8621 | 750  | 0.9272        | 0.7913          | 0.5136                     | 0.5526                | 0.5331                       |
+| 1.1494 | 1000 | 0.8618        | 0.7881          | 0.5281                     | 0.5308                | 0.5295                       |
+| 1.4368 | 1250 | 0.8225        | 0.7816          | 0.5286                     | 0.5265                | 0.5276                       |
+| 1.7241 | 1500 | 0.8128        | 0.7757          | 0.5442                     | 0.5449                | 0.5445                       |
+| 2.0115 | 1750 | 0.8077        | 0.7718          | 0.5665                     | 0.5175                | 0.5420                       |
+| 2.2989 | 2000 | 0.734         | 0.7798          | 0.5271                     | 0.5350                | 0.5311                       |
+| 2.5862 | 2250 | 0.7325        | 0.7758          | 0.5371                     | 0.5666                | 0.5518                       |
+| 2.8736 | 2500 | 0.7245        | 0.7776          | 0.5173                     | 0.5766                | 0.5470                       |
+| 3.1609 | 2750 | 0.7067        | 0.7787          | 0.5209                     | 0.5548                | 0.5379                       |
+| 3.4483 | 3000 | 0.6855        | 0.7807          | 0.5180                     | 0.5611                | 0.5395                       |
+| 3.7356 | 3250 | 0.6831        | 0.7813          | 0.5235                     | 0.5663                | 0.5449                       |
 
 
 ### 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"
   }
 ]