Add new SentenceTransformer model

README.md

@@ -7,7 +7,7 @@ tags:
 - generated_from_trainer
 - dataset_size:111470
 - loss:MultipleNegativesRankingLoss
-base_model: thenlper/gte-small
+base_model: sentence-transformers/all-MiniLM-L6-v2
 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 thenlper/gte-small
+- name: SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2
   results:
   - task:
       type: information-retrieval
@@ -116,49 +116,49 @@ model-index:
       type: NanoMSMARCO
     metrics:
     - type: cosine_accuracy@1
-      value: 0.38
+      value: 0.3
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.6
+      value: 0.5
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.66
+      value: 0.58
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
       value: 0.68
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.38
+      value: 0.3
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.2
+      value: 0.16666666666666669
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.132
+      value: 0.11599999999999999
       name: Cosine Precision@5
     - type: cosine_precision@10
       value: 0.068
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.38
+      value: 0.3
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.6
+      value: 0.5
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.66
+      value: 0.58
       name: Cosine Recall@5
     - type: cosine_recall@10
       value: 0.68
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.5369233576215849
+      value: 0.48741389266955737
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.48966666666666664
+      value: 0.4262222222222222
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.5049307222721455
+      value: 0.44072094685707097
       name: Cosine Map@100
   - task:
       type: information-retrieval
@@ -168,49 +168,49 @@ model-index:
       type: NanoNQ
     metrics:
     - type: cosine_accuracy@1
-      value: 0.42
+      value: 0.26
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.56
+      value: 0.4
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.62
+      value: 0.48
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.74
+      value: 0.54
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.42
+      value: 0.26
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.19333333333333333
+      value: 0.14
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.128
+      value: 0.1
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.07600000000000001
+      value: 0.05600000000000001
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.39
+      value: 0.23
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.53
+      value: 0.37
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.59
+      value: 0.45
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.69
+      value: 0.51
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.5406765177061442
+      value: 0.3745207998751907
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.5126031746031746
+      value: 0.35074603174603175
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.49920094784101693
+      value: 0.3364191132763434
       name: Cosine Map@100
   - task:
       type: nano-beir
@@ -220,61 +220,61 @@ model-index:
       type: NanoBEIR_mean
     metrics:
     - type: cosine_accuracy@1
-      value: 0.4
+      value: 0.28
       name: Cosine Accuracy@1
     - type: cosine_accuracy@3
-      value: 0.5800000000000001
+      value: 0.45
       name: Cosine Accuracy@3
     - type: cosine_accuracy@5
-      value: 0.64
+      value: 0.53
       name: Cosine Accuracy@5
     - type: cosine_accuracy@10
-      value: 0.71
+      value: 0.6100000000000001
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.4
+      value: 0.28
       name: Cosine Precision@1
     - type: cosine_precision@3
-      value: 0.19666666666666666
+      value: 0.15333333333333335
       name: Cosine Precision@3
     - type: cosine_precision@5
-      value: 0.13
+      value: 0.108
       name: Cosine Precision@5
     - type: cosine_precision@10
-      value: 0.07200000000000001
+      value: 0.062000000000000006
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.385
+      value: 0.265
       name: Cosine Recall@1
     - type: cosine_recall@3
-      value: 0.565
+      value: 0.435
       name: Cosine Recall@3
     - type: cosine_recall@5
-      value: 0.625
+      value: 0.515
       name: Cosine Recall@5
     - type: cosine_recall@10
-      value: 0.685
+      value: 0.595
       name: Cosine Recall@10
     - type: cosine_ndcg@10
-      value: 0.5387999376638646
+      value: 0.43096734627237404
       name: Cosine Ndcg@10
     - type: cosine_mrr@10
-      value: 0.5011349206349206
+      value: 0.388484126984127
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.5020658350565812
+      value: 0.3885700300667072
       name: Cosine Map@100
 ---
 
-# SentenceTransformer based on thenlper/gte-small
+# SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2
 
-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.
+This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [sentence-transformers/all-MiniLM-L6-v2](https://huggingface.co/sentence-transformers/all-MiniLM-L6-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.
 
 ## Model Details
 
 ### Model Description
 - **Model Type:** Sentence Transformer
-- **Base model:** [thenlper/gte-small](https://huggingface.co/thenlper/gte-small) <!-- at revision 17e1f347d17fe144873b1201da91788898c639cd -->
+- **Base model:** [sentence-transformers/all-MiniLM-L6-v2](https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2) <!-- at revision c9745ed1d9f207416be6d2e6f8de32d1f16199bf -->
 - **Maximum Sequence Length:** 128 tokens
 - **Output Dimensionality:** 384 dimensions
 - **Similarity Function:** Cosine Similarity
@@ -327,9 +327,9 @@ print(embeddings.shape)
 # Get the similarity scores for the embeddings
 similarities = model.similarity(embeddings, embeddings)
 print(similarities)
-# tensor([[1.0001, 1.0001, 0.9814],
-#         [1.0001, 1.0001, 0.9814],
-#         [0.9814, 0.9814, 1.0000]])
+# tensor([[1.0000, 1.0000, 0.9587],
+#         [1.0000, 1.0000, 0.9587],
+#         [0.9587, 0.9587, 1.0000]])
 ```
 
 <!--
@@ -367,21 +367,21 @@ You can finetune this model on your own dataset.
 
 | Metric              | NanoMSMARCO | NanoNQ     |
 |:--------------------|:------------|:-----------|
-| cosine_accuracy@1   | 0.38        | 0.42       |
-| cosine_accuracy@3   | 0.6         | 0.56       |
-| cosine_accuracy@5   | 0.66        | 0.62       |
-| cosine_accuracy@10  | 0.68        | 0.74       |
-| cosine_precision@1  | 0.38        | 0.42       |
-| cosine_precision@3  | 0.2         | 0.1933     |
-| cosine_precision@5  | 0.132       | 0.128      |
-| cosine_precision@10 | 0.068       | 0.076      |
-| cosine_recall@1     | 0.38        | 0.39       |
-| cosine_recall@3     | 0.6         | 0.53       |
-| cosine_recall@5     | 0.66        | 0.59       |
-| cosine_recall@10    | 0.68        | 0.69       |
-| **cosine_ndcg@10**  | **0.5369**  | **0.5407** |
-| cosine_mrr@10       | 0.4897      | 0.5126     |
-| cosine_map@100      | 0.5049      | 0.4992     |
+| cosine_accuracy@1   | 0.3         | 0.26       |
+| cosine_accuracy@3   | 0.5         | 0.4        |
+| cosine_accuracy@5   | 0.58        | 0.48       |
+| cosine_accuracy@10  | 0.68        | 0.54       |
+| cosine_precision@1  | 0.3         | 0.26       |
+| cosine_precision@3  | 0.1667      | 0.14       |
+| cosine_precision@5  | 0.116       | 0.1        |
+| cosine_precision@10 | 0.068       | 0.056      |
+| cosine_recall@1     | 0.3         | 0.23       |
+| cosine_recall@3     | 0.5         | 0.37       |
+| cosine_recall@5     | 0.58        | 0.45       |
+| cosine_recall@10    | 0.68        | 0.51       |
+| **cosine_ndcg@10**  | **0.4874**  | **0.3745** |
+| cosine_mrr@10       | 0.4262      | 0.3507     |
+| cosine_map@100      | 0.4407      | 0.3364     |
 
 #### Nano BEIR
 
@@ -397,23 +397,23 @@ You can finetune this model on your own dataset.
   }
   ```
 
-| Metric              | Value      |
-|:--------------------|:-----------|
-| cosine_accuracy@1   | 0.4        |
-| cosine_accuracy@3   | 0.58       |
-| cosine_accuracy@5   | 0.64       |
-| cosine_accuracy@10  | 0.71       |
-| cosine_precision@1  | 0.4        |
-| cosine_precision@3  | 0.1967     |
-| cosine_precision@5  | 0.13       |
-| cosine_precision@10 | 0.072      |
-| cosine_recall@1     | 0.385      |
-| cosine_recall@3     | 0.565      |
-| cosine_recall@5     | 0.625      |
-| cosine_recall@10    | 0.685      |
-| **cosine_ndcg@10**  | **0.5388** |
-| cosine_mrr@10       | 0.5011     |
-| cosine_map@100      | 0.5021     |
+| Metric              | Value     |
+|:--------------------|:----------|
+| cosine_accuracy@1   | 0.28      |
+| cosine_accuracy@3   | 0.45      |
+| cosine_accuracy@5   | 0.53      |
+| cosine_accuracy@10  | 0.61      |
+| cosine_precision@1  | 0.28      |
+| cosine_precision@3  | 0.1533    |
+| cosine_precision@5  | 0.108     |
+| cosine_precision@10 | 0.062     |
+| cosine_recall@1     | 0.265     |
+| cosine_recall@3     | 0.435     |
+| cosine_recall@5     | 0.515     |
+| cosine_recall@10    | 0.595     |
+| **cosine_ndcg@10**  | **0.431** |
+| cosine_mrr@10       | 0.3885    |
+| cosine_map@100      | 0.3886    |
 
 <!--
 ## Bias, Risks and Limitations
@@ -487,9 +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`: 8e-05
-- `weight_decay`: 0.005
-- `max_steps`: 1687
+- `learning_rate`: 0.0001
+- `weight_decay`: 0.001
+- `max_steps`: 5062
 - `warmup_ratio`: 0.1
 - `fp16`: True
 - `dataloader_drop_last`: True
@@ -516,14 +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`: 8e-05
-- `weight_decay`: 0.005
+- `learning_rate`: 0.0001
+- `weight_decay`: 0.001
 - `adam_beta1`: 0.9
 - `adam_beta2`: 0.999
 - `adam_epsilon`: 1e-08
 - `max_grad_norm`: 1.0
 - `num_train_epochs`: 3.0
-- `max_steps`: 1687
+- `max_steps`: 5062
 - `lr_scheduler_type`: linear
 - `lr_scheduler_kwargs`: {}
 - `warmup_ratio`: 0.1
@@ -628,17 +628,30 @@ 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       | -             | 4.0678          | 0.6259                     | 0.6583                | 0.6421                       |
-| 0.2874     | 250     | 1.7031        | 0.8455          | 0.5349                     | 0.5337                | 0.5343                       |
-| **0.5747** | **500** | **0.949**     | **0.8059**      | **0.5292**                 | **0.5546**            | **0.5419**                   |
-| 0.8621     | 750     | 0.9183        | 0.7856          | 0.5325                     | 0.5433                | 0.5379                       |
-| 1.1494     | 1000    | 0.8561        | 0.7834          | 0.5132                     | 0.5408                | 0.5270                       |
-| 1.4368     | 1250    | 0.8156        | 0.7782          | 0.5150                     | 0.5353                | 0.5252                       |
-| 1.7241     | 1500    | 0.8064        | 0.7715          | 0.5369                     | 0.5407                | 0.5388                       |
-
-* The bold row denotes the saved checkpoint.
+| Epoch  | Step | Training Loss | Validation Loss | NanoMSMARCO_cosine_ndcg@10 | NanoNQ_cosine_ndcg@10 | NanoBEIR_mean_cosine_ndcg@10 |
+|:------:|:----:|:-------------:|:---------------:|:--------------------------:|:---------------------:|:----------------------------:|
+| 0      | 0    | -             | 1.1445          | 0.5540                     | 0.5931                | 0.5735                       |
+| 0.2874 | 250  | 1.1025        | 0.8649          | 0.4839                     | 0.5173                | 0.5006                       |
+| 0.5747 | 500  | 0.9965        | 0.8468          | 0.5015                     | 0.4853                | 0.4934                       |
+| 0.8621 | 750  | 0.9723        | 0.8249          | 0.5063                     | 0.4415                | 0.4739                       |
+| 1.1494 | 1000 | 0.9091        | 0.8153          | 0.4996                     | 0.4265                | 0.4630                       |
+| 1.4368 | 1250 | 0.868         | 0.8118          | 0.5418                     | 0.4201                | 0.4809                       |
+| 1.7241 | 1500 | 0.863         | 0.8032          | 0.5073                     | 0.4010                | 0.4542                       |
+| 2.0115 | 1750 | 0.8557        | 0.8096          | 0.5121                     | 0.3922                | 0.4521                       |
+| 2.2989 | 2000 | 0.7687        | 0.8067          | 0.4885                     | 0.3905                | 0.4395                       |
+| 2.5862 | 2250 | 0.7718        | 0.8011          | 0.4848                     | 0.3960                | 0.4404                       |
+| 2.8736 | 2500 | 0.7648        | 0.8022          | 0.4765                     | 0.4119                | 0.4442                       |
+| 3.1609 | 2750 | 0.7339        | 0.8176          | 0.4813                     | 0.3885                | 0.4349                       |
+| 3.4483 | 3000 | 0.7055        | 0.8101          | 0.4753                     | 0.3991                | 0.4372                       |
+| 3.7356 | 3250 | 0.7065        | 0.8195          | 0.5022                     | 0.3715                | 0.4368                       |
+| 4.0230 | 3500 | 0.7014        | 0.8258          | 0.5272                     | 0.3856                | 0.4564                       |
+| 4.3103 | 3750 | 0.6601        | 0.8191          | 0.4957                     | 0.3766                | 0.4361                       |
+| 4.5977 | 4000 | 0.6632        | 0.8264          | 0.4649                     | 0.3741                | 0.4195                       |
+| 4.8851 | 4250 | 0.664         | 0.8191          | 0.4954                     | 0.3662                | 0.4308                       |
+| 5.1724 | 4500 | 0.6422        | 0.8277          | 0.4851                     | 0.3749                | 0.4300                       |
+| 5.4598 | 4750 | 0.6336        | 0.8296          | 0.4855                     | 0.3725                | 0.4290                       |
+| 5.7471 | 5000 | 0.6316        | 0.8279          | 0.4874                     | 0.3745                | 0.4310                       |
+
 
 ### Framework Versions
 - Python: 3.10.18

config_sentence_transformers.json

@@ -1,10 +1,10 @@
 {
-  "model_type": "SentenceTransformer",
   "__version__": {
     "sentence_transformers": "5.2.0",
     "transformers": "4.57.3",
     "pytorch": "2.9.1+cu128"
   },
+  "model_type": "SentenceTransformer",
   "prompts": {
     "query": "",
     "document": ""