Add new SentenceTransformer model

1_Pooling/config.json

@@ -1,7 +1,7 @@
 {
-    "word_embedding_dimension": 384,
-    "pooling_mode_cls_token": false,
-    "pooling_mode_mean_tokens": true,
+    "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,

README.md

@@ -7,7 +7,7 @@ tags:
 - generated_from_trainer
 - dataset_size:90000
 - loss:MultipleNegativesRankingLoss
-base_model: sentence-transformers/all-MiniLM-L12-v2
+base_model: Alibaba-NLP/gte-modernbert-base
 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 sentence-transformers/all-MiniLM-L12-v2
+- name: SentenceTransformer based on Alibaba-NLP/gte-modernbert-base
   results:
   - task:
       type: information-retrieval
@@ -164,49 +164,49 @@ model-index:
       type: NanoMSMARCO
     metrics:
     - type: cosine_accuracy@1
-      value: 0.34
+      value: 0.38
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.54
+      value: 0.68
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.64
+      value: 0.8
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.78
+      value: 0.86
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.34
+      value: 0.38
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.18
+      value: 0.22666666666666668
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.128
+      value: 0.16
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.07800000000000001
+      value: 0.08599999999999998
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.34
+      value: 0.38
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.54
+      value: 0.68
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.64
+      value: 0.8
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.78
+      value: 0.86
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.5447080049645561
+      value: 0.6232981077766904
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.47073809523809523
+      value: 0.5465555555555556
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.4806962957327628
+      value: 0.5540526315789474
       name: Cosine Map@100
   - task:
       type: information-retrieval
@@ -216,49 +216,49 @@ model-index:
       type: NanoNQ
     metrics:
     - type: cosine_accuracy@1
-      value: 0.44
+      value: 0.64
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.62
+      value: 0.7
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.7
+      value: 0.78
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.78
+      value: 0.82
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.44
+      value: 0.64
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.21333333333333332
+      value: 0.24
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.14800000000000002
+      value: 0.16
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.08199999999999999
+      value: 0.08800000000000001
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.43
+      value: 0.61
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.61
+      value: 0.66
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.67
+      value: 0.73
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.74
+      value: 0.78
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.5924173512360595
+      value: 0.6987067579229547
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.5506349206349206
+      value: 0.69
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.5491036387356644
+      value: 0.6733088641959746
       name: Cosine Map@100
   - task:
       type: nano-beir
@@ -268,63 +268,63 @@ model-index:
       type: NanoBEIR_mean
     metrics:
     - type: cosine_accuracy@1
-      value: 0.39
+      value: 0.51
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.5800000000000001
+      value: 0.69
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.6699999999999999
+      value: 0.79
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.78
+      value: 0.84
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.39
+      value: 0.51
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.19666666666666666
+      value: 0.23333333333333334
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.138
+      value: 0.16
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.08
+      value: 0.087
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.385
+      value: 0.495
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.575
+      value: 0.67
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.655
+      value: 0.765
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.76
+      value: 0.8200000000000001
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.5685626781003078
+      value: 0.6610024328498225
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.5106865079365079
+      value: 0.6182777777777777
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.5148999672342136
+      value: 0.613680747887461
       name: Cosine Map@100
 ---
 
-# SentenceTransformer based on sentence-transformers/all-MiniLM-L12-v2
+# SentenceTransformer based on Alibaba-NLP/gte-modernbert-base
 
-This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [sentence-transformers/all-MiniLM-L12-v2](https://huggingface.co/sentence-transformers/all-MiniLM-L12-v2). 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.
+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.
 
 ## Model Details
 
 ### Model Description
 - **Model Type:** Sentence Transformer
-- **Base model:** [sentence-transformers/all-MiniLM-L12-v2](https://huggingface.co/sentence-transformers/all-MiniLM-L12-v2) <!-- at revision 936af83a2ecce5fe87a09109ff5cbcefe073173a -->
+- **Base model:** [Alibaba-NLP/gte-modernbert-base](https://huggingface.co/Alibaba-NLP/gte-modernbert-base) <!-- at revision e7f32e3c00f91d699e8c43b53106206bcc72bb22 -->
 - **Maximum Sequence Length:** 128 tokens
-- **Output Dimensionality:** 384 dimensions
+- **Output Dimensionality:** 768 dimensions
 - **Similarity Function:** Cosine Similarity
 <!-- - **Training Dataset:** Unknown -->
 <!-- - **Language:** Unknown -->
@@ -340,9 +340,8 @@ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [s
 
 ```
 SentenceTransformer(
-  (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()
+  (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})
 )
 ```
 
@@ -370,14 +369,14 @@ sentences = [
 ]
 embeddings = model.encode(sentences)
 print(embeddings.shape)
-# [3, 384]
+# [3, 768]
 
 # Get the similarity scores for the embeddings
 similarities = model.similarity(embeddings, embeddings)
 print(similarities)
-# tensor([[ 1.0000,  0.7187, -0.0053],
-#         [ 0.7187,  1.0000,  0.0412],
-#         [-0.0053,  0.0412,  1.0000]])
+# tensor([[ 1.0001,  0.6919, -0.0133],
+#         [ 0.6919,  1.0000, -0.0985],
+#         [-0.0133, -0.0985,  1.0000]])
 ```
 
 <!--
@@ -415,21 +414,21 @@ You can finetune this model on your own dataset.
 
 | Metric              | NanoMSMARCO | NanoNQ     |
 |:--------------------|:------------|:-----------|
-| cosine_accuracy@1   | 0.34        | 0.44       |
-| cosine_accuracy@3   | 0.54        | 0.62       |
-| cosine_accuracy@5   | 0.64        | 0.7        |
-| cosine_accuracy@10  | 0.78        | 0.78       |
-| cosine_precision@1  | 0.34        | 0.44       |
-| cosine_precision@3  | 0.18        | 0.2133     |
-| cosine_precision@5  | 0.128       | 0.148      |
-| cosine_precision@10 | 0.078       | 0.082      |
-| cosine_recall@1     | 0.34        | 0.43       |
-| cosine_recall@3     | 0.54        | 0.61       |
-| cosine_recall@5     | 0.64        | 0.67       |
-| cosine_recall@10    | 0.78        | 0.74       |
-| **cosine_ndcg@10**  | **0.5447**  | **0.5924** |
-| cosine_mrr@10       | 0.4707      | 0.5506     |
-| cosine_map@100      | 0.4807      | 0.5491     |
+| cosine_accuracy@1   | 0.38        | 0.64       |
+| cosine_accuracy@3   | 0.68        | 0.7        |
+| cosine_accuracy@5   | 0.8         | 0.78       |
+| cosine_accuracy@10  | 0.86        | 0.82       |
+| cosine_precision@1  | 0.38        | 0.64       |
+| cosine_precision@3  | 0.2267      | 0.24       |
+| cosine_precision@5  | 0.16        | 0.16       |
+| cosine_precision@10 | 0.086       | 0.088      |
+| cosine_recall@1     | 0.38        | 0.61       |
+| cosine_recall@3     | 0.68        | 0.66       |
+| cosine_recall@5     | 0.8         | 0.73       |
+| cosine_recall@10    | 0.86        | 0.78       |
+| **cosine_ndcg@10**  | **0.6233**  | **0.6987** |
+| cosine_mrr@10       | 0.5466      | 0.69       |
+| cosine_map@100      | 0.5541      | 0.6733     |
 
 #### Nano BEIR
 
@@ -445,23 +444,23 @@ You can finetune this model on your own dataset.
   }
   ```
 
-| Metric              | Value      |
-|:--------------------|:-----------|
-| cosine_accuracy@1   | 0.39       |
-| cosine_accuracy@3   | 0.58       |
-| cosine_accuracy@5   | 0.67       |
-| cosine_accuracy@10  | 0.78       |
-| cosine_precision@1  | 0.39       |
-| cosine_precision@3  | 0.1967     |
-| cosine_precision@5  | 0.138      |
-| cosine_precision@10 | 0.08       |
-| cosine_recall@1     | 0.385      |
-| cosine_recall@3     | 0.575      |
-| cosine_recall@5     | 0.655      |
-| cosine_recall@10    | 0.76       |
-| **cosine_ndcg@10**  | **0.5686** |
-| cosine_mrr@10       | 0.5107     |
-| cosine_map@100      | 0.5149     |
+| Metric              | Value     |
+|:--------------------|:----------|
+| cosine_accuracy@1   | 0.51      |
+| cosine_accuracy@3   | 0.69      |
+| cosine_accuracy@5   | 0.79      |
+| cosine_accuracy@10  | 0.84      |
+| cosine_precision@1  | 0.51      |
+| cosine_precision@3  | 0.2333    |
+| cosine_precision@5  | 0.16      |
+| cosine_precision@10 | 0.087     |
+| cosine_recall@1     | 0.495     |
+| cosine_recall@3     | 0.67      |
+| cosine_recall@5     | 0.765     |
+| cosine_recall@10    | 0.82      |
+| **cosine_ndcg@10**  | **0.661** |
+| cosine_mrr@10       | 0.6183    |
+| cosine_map@100      | 0.6137    |
 
 <!--
 ## Bias, Risks and Limitations
@@ -484,10 +483,10 @@ 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: 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> |
+  |         | 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> |
 * Samples:
   | anchor                                                          | positive                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      | negative                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
   |:----------------------------------------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
@@ -513,7 +512,7 @@ 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: 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> |
+  | 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> |
 * Samples:
   | anchor                                                                                           | positive                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 | negative                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    |
   |:-------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
@@ -535,9 +534,9 @@ 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`: 8e-05
-- `weight_decay`: 0.005
-- `max_steps`: 1125
+- `learning_rate`: 4e-05
+- `weight_decay`: 0.01
+- `max_steps`: 500
 - `warmup_ratio`: 0.1
 - `fp16`: True
 - `dataloader_drop_last`: True
@@ -564,14 +563,14 @@ You can finetune this model on your own dataset.
 - `gradient_accumulation_steps`: 1
 - `eval_accumulation_steps`: None
 - `torch_empty_cache_steps`: None
-- `learning_rate`: 8e-05
-- `weight_decay`: 0.005
+- `learning_rate`: 4e-05
+- `weight_decay`: 0.01
 - `adam_beta1`: 0.9
 - `adam_beta2`: 0.999
 - `adam_epsilon`: 1e-08
 - `max_grad_norm`: 1.0
 - `num_train_epochs`: 3.0
-- `max_steps`: 1125
+- `max_steps`: 500
 - `lr_scheduler_type`: linear
 - `lr_scheduler_kwargs`: {}
 - `warmup_ratio`: 0.1
@@ -676,14 +675,13 @@ You can finetune this model on your own dataset.
 </details>
 
 ### Training Logs
-| Epoch  | Step | Training Loss | Validation Loss | NanoMSMARCO_cosine_ndcg@10 | NanoNQ_cosine_ndcg@10 | NanoBEIR_mean_cosine_ndcg@10 |
-|:------:|:----:|:-------------:|:---------------:|:--------------------------:|:---------------------:|:----------------------------:|
-| 0      | 0    | -             | 0.0731          | 0.5887                     | 0.5786                | 0.5836                       |
-| 0.3556 | 250  | 0.0821        | 0.0701          | 0.5325                     | 0.5977                | 0.5651                       |
-| 0.7112 | 500  | 0.0805        | 0.0640          | 0.5523                     | 0.5631                | 0.5577                       |
-| 1.0669 | 750  | 0.0712        | 0.0572          | 0.5369                     | 0.5819                | 0.5594                       |
-| 1.4225 | 1000 | 0.0371        | 0.0551          | 0.5447                     | 0.5924                | 0.5686                       |
+| Epoch      | Step    | Training Loss | Validation Loss | NanoMSMARCO_cosine_ndcg@10 | NanoNQ_cosine_ndcg@10 | NanoBEIR_mean_cosine_ndcg@10 |
+|:----------:|:-------:|:-------------:|:---------------:|:--------------------------:|:---------------------:|:----------------------------:|
+| 0          | 0       | -             | 0.4265          | 0.6530                     | 0.6552                | 0.6541                       |
+| 0.3556     | 250     | 0.0816        | 0.0565          | 0.6334                     | 0.6822                | 0.6578                       |
+| **0.7112** | **500** | **0.0517**    | **0.052**       | **0.6233**                 | **0.6987**            | **0.661**                    |
 
+* The bold row denotes the saved checkpoint.
 
 ### Framework Versions
 - Python: 3.10.18

config_sentence_transformers.json

@@ -4,11 +4,11 @@
     "transformers": "4.57.3",
     "pytorch": "2.9.1+cu128"
   },
-  "model_type": "SentenceTransformer",
   "prompts": {
     "query": "",
     "document": ""
   },
   "default_prompt_name": null,
-  "similarity_fn_name": "cosine"
+  "similarity_fn_name": "cosine",
+  "model_type": "SentenceTransformer"
 }

modules.json

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