Initial commit: Sentence-Transformers mean-pooling wrapper

README.md

@@ -1,144 +1,80 @@
 ---
+license: apache-2.0
+language:
+- fr
+- en
+library_name: sentence-transformers
+pipeline_tag: sentence-similarity
 tags:
 - sentence-transformers
-- sentence-similarity
+- embeddings
+- eurobert
+- multilingual
 - feature-extraction
-- dense
-base_model: Mathlesage/euroBertV10
-pipeline_tag: sentence-similarity
-library_name: sentence-transformers
+base_model: EuroBERT/EuroBERT-210m
 ---
 
-# SentenceTransformer based on Mathlesage/euroBertV10
-
-This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [Mathlesage/euroBertV10](https://huggingface.co/Mathlesage/euroBertV10). 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:** [Mathlesage/euroBertV10](https://huggingface.co/Mathlesage/euroBertV10) <!-- at revision 6056a08488ad1c7d39822e6306e086ce83b4a6f0 -->
-- **Maximum Sequence Length:** 8196 tokens
-- **Output Dimensionality:** 768 dimensions
-- **Similarity Function:** Cosine Similarity
-<!-- - **Training Dataset:** Unknown -->
-<!-- - **Language:** Unknown -->
-<!-- - **License:** Unknown -->
+# OrdalieTech/Solon-embeddings-mini-beta-1.1
 
-### Model Sources
+Le modèle d'origine a été créé à partir de `EuroBERT/EuroBERT-210m`, puis entraîné avec la technique **InfoNCE** sur des **paires de très haute qualité générées par LLM**
 
-- **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
-- **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
-- **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
 
-### Full Model Architecture
+## Points clés
 
-```
-SentenceTransformer(
-  (0): Transformer({'max_seq_length': 8196, 'do_lower_case': False, 'architecture': 'EuroBertModel'})
-  (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})
-)
-```
+- **Backbone** : `EuroBERT/EuroBERT-210m`  
+- **Pooling** : moyenne des tokens (CLS désactivé, max désactivé)  
+- **Dimensions** : 768  
+- **Langues** : multilingue dont le français et l'anglais
 
-## Usage
+## Exemples d'usage
 
-### Direct Usage (Sentence Transformers)
+### Avec `sentence-transformers`
 
-First install the Sentence Transformers library:
-
-```bash
+```python
 pip install -U sentence-transformers
 ```
 
-Then you can load this model and run inference.
 ```python
 from sentence_transformers import SentenceTransformer
 
-# Download from the 🤗 Hub
-model = SentenceTransformer("sentence_transformers_model_id")
-# Run inference
-sentences = [
-    'The weather is lovely today.',
-    "It's so sunny outside!",
-    'He drove to the stadium.',
-]
-embeddings = model.encode(sentences)
-print(embeddings.shape)
-# [3, 768]
-
-# Get the similarity scores for the embeddings
-similarities = model.similarity(embeddings, embeddings)
-print(similarities)
-# tensor([[1.0000, 0.5651, 0.0976],
-#         [0.5651, 1.0000, 0.2057],
-#         [0.0976, 0.2057, 1.0000]])
+model = SentenceTransformer("OrdalieTech/Solon-embeddings-mini-beta-1.1")
+sentences = ["Ceci est une phrase d'exemple", "Chaque phrase est convertie en vecteur"]
+embeddings = model.encode(sentences, convert_to_tensor=False, normalize_embeddings=True)
+print(embeddings[0].shape)  # (768,)
 ```
 
-<!--
-### Direct Usage (Transformers)
-
-<details><summary>Click to see the direct usage in Transformers</summary>
-
-</details>
--->
-
-<!--
-### Downstream Usage (Sentence Transformers)
+### Avec `transformers` (feature extraction)
 
-You can finetune this model on your own dataset.
-
-<details><summary>Click to expand</summary>
-
-</details>
--->
-
-<!--
-### Out-of-Scope Use
-
-*List how the model may foreseeably be misused and address what users ought not to do with the model.*
--->
-
-<!--
-## Bias, Risks and Limitations
-
-*What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
--->
-
-<!--
-### Recommendations
-
-*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
--->
-
-## Training Details
-
-### Framework Versions
-- Python: 3.9.6
-- Sentence Transformers: 5.1.0
-- Transformers: 4.55.2
-- PyTorch: 2.8.0
-- Accelerate: 
-- Datasets: 2.21.0
-- Tokenizers: 0.21.4
+```python
+pip install -U transformers torch
+```
 
-## Citation
+```python
+from transformers import AutoTokenizer, AutoModel
+import torch
 
-### BibTeX
+tok = AutoTokenizer.from_pretrained("EuroBERT/EuroBERT-210m", trust_remote_code=True)
+enc = AutoModel.from_pretrained("EuroBERT/EuroBERT-210m", trust_remote_code=True)
 
-<!--
-## Glossary
+inputs = tok(["Ceci est une phrase d'exemple"], padding=True, truncation=True, return_tensors="pt")
+with torch.no_grad():
+    out = enc(**inputs).last_hidden_state  # (batch, seq, 768)
 
-*Clearly define terms in order to be accessible across audiences.*
--->
+mask = inputs["attention_mask"].unsqueeze(-1)  # (batch, seq, 1)
+mean_emb = (out * mask).sum(dim=1) / mask.sum(dim=1).clamp(min=1)
+```
 
-<!--
-## Model Card Authors
+## Cas d'usage
 
-*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
--->
+- Recherche sémantique
+- Reranking
+- Similarité sémantique de phrases (STS)
+- Recommandation de contenu
+- Classification basée sur des embeddings
 
-<!--
-## Model Card Contact
+## Crédit et licence
 
-*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
--->
+- Modèle de base : [`EuroBERT/EuroBERT-210m`](https://huggingface.co/EuroBERT/EuroBERT-210m) • licence Apache-2.0  
+- Cette publication reprend la licence Apache-2.0 et respecte les conditions de redistribution du modèle de base
+- Merci aux auteurs d'EuroBERT pour leur travail et l'ouverture du modèle
+- Création : @matheoqtb