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:90000
 - loss:MultipleNegativesRankingLoss
-base_model: Alibaba-NLP/gte-modernbert-base
+base_model: thenlper/gte-small
 widget:
 - source_sentence: who is the publisher of the norton anthology american literature
   sentences:
@@ -154,7 +154,7 @@ 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
@@ -164,49 +164,49 @@ model-index:
       type: NanoMSMARCO
     metrics:
     - type: cosine_accuracy@1
-      value: 0.26
+      value: 0.12
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.48
+      value: 0.32
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.6
+      value: 0.48
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.68
+      value: 0.6
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.26
+      value: 0.12
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.15999999999999998
+      value: 0.10666666666666666
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.12000000000000002
+      value: 0.09600000000000002
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.068
+      value: 0.06
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.26
+      value: 0.12
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.48
+      value: 0.32
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.6
+      value: 0.48
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.68
+      value: 0.6
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.45896424557362947
+      value: 0.3451699142127375
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.38885714285714285
+      value: 0.2649920634920635
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.39926372736834176
+      value: 0.2748673342528789
       name: Cosine Map@100
   - task:
       type: information-retrieval
@@ -216,49 +216,49 @@ model-index:
       type: NanoNQ
     metrics:
     - type: cosine_accuracy@1
-      value: 0.36
+      value: 0.22
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.58
+      value: 0.44
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.64
+      value: 0.5
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.8
+      value: 0.56
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.36
+      value: 0.22
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.19333333333333333
+      value: 0.14666666666666664
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.12800000000000003
+      value: 0.1
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.08
+      value: 0.05800000000000001
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.36
+      value: 0.22
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.57
+      value: 0.43
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.62
+      value: 0.49
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.75
+      value: 0.54
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.5491170117720099
+      value: 0.3853992171360362
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.49174603174603176
+      value: 0.3358888888888889
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.4918572150858902
+      value: 0.3486523060866078
       name: Cosine Map@100
   - task:
       type: nano-beir
@@ -268,63 +268,63 @@ model-index:
       type: NanoBEIR_mean
     metrics:
     - type: cosine_accuracy@1
-      value: 0.31
+      value: 0.16999999999999998
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.53
+      value: 0.38
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.62
+      value: 0.49
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.74
+      value: 0.5800000000000001
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.31
+      value: 0.16999999999999998
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.17666666666666664
+      value: 0.12666666666666665
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.12400000000000003
+      value: 0.098
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.07400000000000001
+      value: 0.059000000000000004
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.31
+      value: 0.16999999999999998
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.5249999999999999
+      value: 0.375
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.61
+      value: 0.485
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.7150000000000001
+      value: 0.5700000000000001
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.5040406286728196
+      value: 0.36528456567438683
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.4403015873015873
+      value: 0.3004404761904762
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.44556047122711595
+      value: 0.31175982016974335
       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 -->
@@ -340,8 +340,9 @@ 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()
 )
 ```
 
@@ -369,14 +370,14 @@ 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,  0.9877, -0.0962],
-#         [ 0.9877,  1.0000, -0.0887],
-#         [-0.0962, -0.0887,  1.0000]])
+# tensor([[ 1.0000,  0.9977, -0.0309],
+#         [ 0.9977,  1.0000, -0.0296],
+#         [-0.0309, -0.0296,  1.0000]])
 ```
 
 <!--
@@ -414,21 +415,21 @@ You can finetune this model on your own dataset.
 
 | Metric              | NanoMSMARCO | NanoNQ     |
 |:--------------------|:------------|:-----------|
-| cosine_accuracy@1   | 0.26        | 0.36       |
-| cosine_accuracy@3   | 0.48        | 0.58       |
-| cosine_accuracy@5   | 0.6         | 0.64       |
-| cosine_accuracy@10  | 0.68        | 0.8        |
-| cosine_precision@1  | 0.26        | 0.36       |
-| cosine_precision@3  | 0.16        | 0.1933     |
-| cosine_precision@5  | 0.12        | 0.128      |
-| cosine_precision@10 | 0.068       | 0.08       |
-| cosine_recall@1     | 0.26        | 0.36       |
-| cosine_recall@3     | 0.48        | 0.57       |
-| cosine_recall@5     | 0.6         | 0.62       |
-| cosine_recall@10    | 0.68        | 0.75       |
-| **cosine_ndcg@10**  | **0.459**   | **0.5491** |
-| cosine_mrr@10       | 0.3889      | 0.4917     |
-| cosine_map@100      | 0.3993      | 0.4919     |
+| cosine_accuracy@1   | 0.12        | 0.22       |
+| cosine_accuracy@3   | 0.32        | 0.44       |
+| cosine_accuracy@5   | 0.48        | 0.5        |
+| cosine_accuracy@10  | 0.6         | 0.56       |
+| cosine_precision@1  | 0.12        | 0.22       |
+| cosine_precision@3  | 0.1067      | 0.1467     |
+| cosine_precision@5  | 0.096       | 0.1        |
+| cosine_precision@10 | 0.06        | 0.058      |
+| cosine_recall@1     | 0.12        | 0.22       |
+| cosine_recall@3     | 0.32        | 0.43       |
+| cosine_recall@5     | 0.48        | 0.49       |
+| cosine_recall@10    | 0.6         | 0.54       |
+| **cosine_ndcg@10**  | **0.3452**  | **0.3854** |
+| cosine_mrr@10       | 0.265       | 0.3359     |
+| cosine_map@100      | 0.2749      | 0.3487     |
 
 #### Nano BEIR
 
@@ -444,23 +445,23 @@ You can finetune this model on your own dataset.
   }
   ```
 
-| Metric              | Value     |
-|:--------------------|:----------|
-| cosine_accuracy@1   | 0.31      |
-| cosine_accuracy@3   | 0.53      |
-| cosine_accuracy@5   | 0.62      |
-| cosine_accuracy@10  | 0.74      |
-| cosine_precision@1  | 0.31      |
-| cosine_precision@3  | 0.1767    |
-| cosine_precision@5  | 0.124     |
-| cosine_precision@10 | 0.074     |
-| cosine_recall@1     | 0.31      |
-| cosine_recall@3     | 0.525     |
-| cosine_recall@5     | 0.61      |
-| cosine_recall@10    | 0.715     |
-| **cosine_ndcg@10**  | **0.504** |
-| cosine_mrr@10       | 0.4403    |
-| cosine_map@100      | 0.4456    |
+| Metric              | Value      |
+|:--------------------|:-----------|
+| cosine_accuracy@1   | 0.17       |
+| cosine_accuracy@3   | 0.38       |
+| cosine_accuracy@5   | 0.49       |
+| cosine_accuracy@10  | 0.58       |
+| cosine_precision@1  | 0.17       |
+| cosine_precision@3  | 0.1267     |
+| cosine_precision@5  | 0.098      |
+| cosine_precision@10 | 0.059      |
+| cosine_recall@1     | 0.17       |
+| cosine_recall@3     | 0.375      |
+| cosine_recall@5     | 0.485      |
+| cosine_recall@10    | 0.57       |
+| **cosine_ndcg@10**  | **0.3653** |
+| cosine_mrr@10       | 0.3004     |
+| cosine_map@100      | 0.3118     |
 
 <!--
 ## Bias, Risks and Limitations
@@ -483,10 +484,10 @@ You can finetune this model on your own dataset.
 * Size: 90,000 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: 10 tokens</li><li>mean: 12.57 tokens</li><li>max: 28 tokens</li></ul> | <ul><li>min: 19 tokens</li><li>mean: 107.04 tokens</li><li>max: 128 tokens</li></ul> | <ul><li>min: 15 tokens</li><li>mean: 105.42 tokens</li><li>max: 128 tokens</li></ul> |
+  |         | anchor                                                                            | positive                                                                            | negative                                                                             |
+  |:--------|:----------------------------------------------------------------------------------|:------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|
+  | type    | string                                                                            | string                                                                              | string                                                                               |
+  | details | <ul><li>min: 9 tokens</li><li>mean: 11.82 tokens</li><li>max: 27 tokens</li></ul> | <ul><li>min: 19 tokens</li><li>mean: 106.2 tokens</li><li>max: 128 tokens</li></ul> | <ul><li>min: 16 tokens</li><li>mean: 104.63 tokens</li><li>max: 128 tokens</li></ul> |
 * Samples:
   | anchor                                                          | positive                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      | negative                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
   |:----------------------------------------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
@@ -512,7 +513,7 @@ You can finetune this model on your own dataset.
   |         | anchor                                                                            | positive                                                                             | negative                                                                             |
   |:--------|:----------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|
   | type    | string                                                                            | string                                                                               | string                                                                               |
-  | details | <ul><li>min: 9 tokens</li><li>mean: 12.46 tokens</li><li>max: 25 tokens</li></ul> | <ul><li>min: 16 tokens</li><li>mean: 106.89 tokens</li><li>max: 128 tokens</li></ul> | <ul><li>min: 11 tokens</li><li>mean: 106.57 tokens</li><li>max: 128 tokens</li></ul> |
+  | details | <ul><li>min: 9 tokens</li><li>mean: 11.76 tokens</li><li>max: 24 tokens</li></ul> | <ul><li>min: 15 tokens</li><li>mean: 105.95 tokens</li><li>max: 128 tokens</li></ul> | <ul><li>min: 11 tokens</li><li>mean: 105.69 tokens</li><li>max: 128 tokens</li></ul> |
 * Samples:
   | anchor                                                                                           | positive                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 | negative                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    |
   |:-------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
@@ -534,9 +535,9 @@ 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
@@ -563,14 +564,14 @@ 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
@@ -677,9 +678,20 @@ 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    | -             | 4.4513          | 0.6530                     | 0.6552                | 0.6541                       |
-| 0.3556 | 250  | 3.1939        | 2.9908          | 0.4651                     | 0.5551                | 0.5101                       |
-| 0.7112 | 500  | 2.9769        | 2.9599          | 0.4590                     | 0.5491                | 0.5040                       |
+| 0      | 0    | -             | 5.0014          | 0.6259                     | 0.6583                | 0.6421                       |
+| 0.3556 | 250  | 3.7345        | 3.0513          | 0.4721                     | 0.4567                | 0.4644                       |
+| 0.7112 | 500  | 3.1165        | 2.9938          | 0.4464                     | 0.4306                | 0.4385                       |
+| 1.0669 | 750  | 3.055         | 2.9656          | 0.4028                     | 0.4675                | 0.4351                       |
+| 1.4225 | 1000 | 3.0018        | 2.9558          | 0.3668                     | 0.4309                | 0.3989                       |
+| 1.7781 | 1250 | 2.988         | 2.9463          | 0.4017                     | 0.4426                | 0.4221                       |
+| 2.1337 | 1500 | 2.9625        | 2.9372          | 0.3571                     | 0.4003                | 0.3787                       |
+| 2.4893 | 1750 | 2.9363        | 2.9311          | 0.3729                     | 0.4068                | 0.3898                       |
+| 2.8450 | 2000 | 2.9287        | 2.9274          | 0.3728                     | 0.3778                | 0.3753                       |
+| 3.2006 | 2250 | 2.907         | 2.9254          | 0.3770                     | 0.3713                | 0.3742                       |
+| 3.5562 | 2500 | 2.8979        | 2.9242          | 0.3606                     | 0.3884                | 0.3745                       |
+| 3.9118 | 2750 | 2.8931        | 2.9215          | 0.3446                     | 0.3955                | 0.3700                       |
+| 4.2674 | 3000 | 2.883         | 2.9207          | 0.3511                     | 0.3777                | 0.3644                       |
+| 4.6230 | 3250 | 2.8762        | 2.9201          | 0.3452                     | 0.3854                | 0.3653                       |
 
 
 ### 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"
   }
 ]