radoslavralev commited on
Commit
b0acc25
·
verified ·
1 Parent(s): 2fd121f

Add new SentenceTransformer model

Browse files
1_Pooling/config.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
  "word_embedding_dimension": 768,
3
- "pooling_mode_cls_token": false,
4
- "pooling_mode_mean_tokens": true,
5
  "pooling_mode_max_tokens": false,
6
  "pooling_mode_mean_sqrt_len_tokens": false,
7
  "pooling_mode_weightedmean_tokens": false,
 
1
  {
2
  "word_embedding_dimension": 768,
3
+ "pooling_mode_cls_token": true,
4
+ "pooling_mode_mean_tokens": false,
5
  "pooling_mode_max_tokens": false,
6
  "pooling_mode_mean_sqrt_len_tokens": false,
7
  "pooling_mode_weightedmean_tokens": false,
README.md CHANGED
@@ -14,7 +14,7 @@ tags:
14
  - generated_from_trainer
15
  - dataset_size:9233417
16
  - loss:ArcFaceInBatchLoss
17
- base_model: answerdotai/ModernBERT-base
18
  widget:
19
  - source_sentence: Hayley Vaughan portrayed Ripa on the ABC daytime soap opera , ``
20
  All My Children `` , between 1990 and 2002 .
@@ -79,34 +79,34 @@ model-index:
79
  type: test
80
  metrics:
81
  - type: cosine_accuracy@1
82
- value: 0.4134460403309574
83
  name: Cosine Accuracy@1
84
  - type: cosine_precision@1
85
- value: 0.4134460403309574
86
  name: Cosine Precision@1
87
  - type: cosine_recall@1
88
- value: 0.39978666009460445
89
  name: Cosine Recall@1
90
  - type: cosine_ndcg@10
91
- value: 0.5951872317507402
92
  name: Cosine Ndcg@10
93
  - type: cosine_mrr@1
94
- value: 0.4134460403309574
95
  name: Cosine Mrr@1
96
  - type: cosine_map@100
97
- value: 0.5433812918329537
98
  name: Cosine Map@100
99
  ---
100
 
101
  # Redis fine-tuned BiEncoder model for semantic caching on LangCache
102
 
103
- This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base) on the [LangCache Sentence Pairs (all)](https://huggingface.co/datasets/redis/langcache-sentencepairs-v2) dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for sentence pair similarity.
104
 
105
  ## Model Details
106
 
107
  ### Model Description
108
  - **Model Type:** Sentence Transformer
109
- - **Base model:** [answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base) <!-- at revision 8949b909ec900327062f0ebf497f51aef5e6f0c8 -->
110
  - **Maximum Sequence Length:** 100 tokens
111
  - **Output Dimensionality:** 768 dimensions
112
  - **Similarity Function:** Cosine Similarity
@@ -126,7 +126,7 @@ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [a
126
  ```
127
  SentenceTransformer(
128
  (0): Transformer({'max_seq_length': 100, 'do_lower_case': False, 'architecture': 'ModernBertModel'})
129
- (1): Pooling({'word_embedding_dimension': 768, '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})
130
  )
131
  ```
132
 
@@ -159,9 +159,9 @@ print(embeddings.shape)
159
  # Get the similarity scores for the embeddings
160
  similarities = model.similarity(embeddings, embeddings)
161
  print(similarities)
162
- # tensor([[0.9961, 0.9922, 0.9922],
163
- # [0.9922, 1.0000, 1.0000],
164
- # [0.9922, 1.0000, 1.0000]], dtype=torch.bfloat16)
165
  ```
166
 
167
  <!--
@@ -197,14 +197,14 @@ You can finetune this model on your own dataset.
197
  * Dataset: `test`
198
  * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
199
 
200
- | Metric | Value |
201
- |:-------------------|:-----------|
202
- | cosine_accuracy@1 | 0.4134 |
203
- | cosine_precision@1 | 0.4134 |
204
- | cosine_recall@1 | 0.3998 |
205
- | **cosine_ndcg@10** | **0.5952** |
206
- | cosine_mrr@1 | 0.4134 |
207
- | cosine_map@100 | 0.5434 |
208
 
209
  <!--
210
  ## Bias, Risks and Limitations
@@ -277,7 +277,7 @@ You can finetune this model on your own dataset.
277
  ### Training Logs
278
  | Epoch | Step | test_cosine_ndcg@10 |
279
  |:-----:|:----:|:-------------------:|
280
- | -1 | -1 | 0.5952 |
281
 
282
 
283
  ### Framework Versions
 
14
  - generated_from_trainer
15
  - dataset_size:9233417
16
  - loss:ArcFaceInBatchLoss
17
+ base_model: Alibaba-NLP/gte-modernbert-base
18
  widget:
19
  - source_sentence: Hayley Vaughan portrayed Ripa on the ABC daytime soap opera , ``
20
  All My Children `` , between 1990 and 2002 .
 
79
  type: test
80
  metrics:
81
  - type: cosine_accuracy@1
82
+ value: 0.5861241448475948
83
  name: Cosine Accuracy@1
84
  - type: cosine_precision@1
85
+ value: 0.5861241448475948
86
  name: Cosine Precision@1
87
  - type: cosine_recall@1
88
+ value: 0.5679885764966713
89
  name: Cosine Recall@1
90
  - type: cosine_ndcg@10
91
+ value: 0.7729838064849864
92
  name: Cosine Ndcg@10
93
  - type: cosine_mrr@1
94
+ value: 0.5861241448475948
95
  name: Cosine Mrr@1
96
  - type: cosine_map@100
97
+ value: 0.7216697804426214
98
  name: Cosine Map@100
99
  ---
100
 
101
  # Redis fine-tuned BiEncoder model for semantic caching on LangCache
102
 
103
+ 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) on the [LangCache Sentence Pairs (all)](https://huggingface.co/datasets/redis/langcache-sentencepairs-v2) dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for sentence pair similarity.
104
 
105
  ## Model Details
106
 
107
  ### Model Description
108
  - **Model Type:** Sentence Transformer
109
+ - **Base model:** [Alibaba-NLP/gte-modernbert-base](https://huggingface.co/Alibaba-NLP/gte-modernbert-base) <!-- at revision e7f32e3c00f91d699e8c43b53106206bcc72bb22 -->
110
  - **Maximum Sequence Length:** 100 tokens
111
  - **Output Dimensionality:** 768 dimensions
112
  - **Similarity Function:** Cosine Similarity
 
126
  ```
127
  SentenceTransformer(
128
  (0): Transformer({'max_seq_length': 100, 'do_lower_case': False, 'architecture': 'ModernBertModel'})
129
+ (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})
130
  )
131
  ```
132
 
 
159
  # Get the similarity scores for the embeddings
160
  similarities = model.similarity(embeddings, embeddings)
161
  print(similarities)
162
+ # tensor([[1.0000, 0.9961, 0.9922],
163
+ # [0.9961, 1.0000, 0.9961],
164
+ # [0.9922, 0.9961, 0.9961]], dtype=torch.bfloat16)
165
  ```
166
 
167
  <!--
 
197
  * Dataset: `test`
198
  * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
199
 
200
+ | Metric | Value |
201
+ |:-------------------|:----------|
202
+ | cosine_accuracy@1 | 0.5861 |
203
+ | cosine_precision@1 | 0.5861 |
204
+ | cosine_recall@1 | 0.568 |
205
+ | **cosine_ndcg@10** | **0.773** |
206
+ | cosine_mrr@1 | 0.5861 |
207
+ | cosine_map@100 | 0.7217 |
208
 
209
  <!--
210
  ## Bias, Risks and Limitations
 
277
  ### Training Logs
278
  | Epoch | Step | test_cosine_ndcg@10 |
279
  |:-----:|:----:|:-------------------:|
280
+ | -1 | -1 | 0.7730 |
281
 
282
 
283
  ### Framework Versions
config_sentence_transformers.json CHANGED
@@ -1,5 +1,4 @@
1
  {
2
- "model_type": "SentenceTransformer",
3
  "__version__": {
4
  "sentence_transformers": "5.1.0",
5
  "transformers": "4.56.0",
@@ -10,5 +9,6 @@
10
  "document": ""
11
  },
12
  "default_prompt_name": null,
13
- "similarity_fn_name": "cosine"
 
14
  }
 
1
  {
 
2
  "__version__": {
3
  "sentence_transformers": "5.1.0",
4
  "transformers": "4.56.0",
 
9
  "document": ""
10
  },
11
  "default_prompt_name": null,
12
+ "similarity_fn_name": "cosine",
13
+ "model_type": "SentenceTransformer"
14
  }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:275fea20b9e3f51da926fce266f04e7d3defcc7a3143e947a76c09a5afaaa03c
3
  size 298041696
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95d02211c4cca89113f9f3e93ed91f5176bf50170faa2cb835f7bfea15bb9dd2
3
  size 298041696
tokenizer_config.json CHANGED
@@ -938,7 +938,7 @@
938
  "input_ids",
939
  "attention_mask"
940
  ],
941
- "model_max_length": 8192,
942
  "pad_token": "[PAD]",
943
  "sep_token": "[SEP]",
944
  "tokenizer_class": "PreTrainedTokenizerFast",
 
938
  "input_ids",
939
  "attention_mask"
940
  ],
941
+ "model_max_length": 1000000000000000019884624838656,
942
  "pad_token": "[PAD]",
943
  "sep_token": "[SEP]",
944
  "tokenizer_class": "PreTrainedTokenizerFast",