Update README.md

README.md

@@ -1,84 +1,130 @@
 ---
 pipeline_tag: sentence-similarity
 language: fr
-license: apache-2.0
+license: mit
 datasets:
 - unicamp-dl/mmarco
 metrics:
 - recall
 tags:
-- feature-extraction
-- sentence-similarity
+- passage-retrieval
 library_name: sentence-transformers
+model-index:
+- name: biencoder-mMiniLMv2-L6-mmarcoFR
+  results:
+    - task:
+        type: sentence-similarity
+        name: Passage Retrieval
+      dataset:
+        type: unicamp-dl/mmarco
+        name: mMARCO-fr
+        config: french
+        split: validation
+      metrics:
+        - type: recall_at_500
+          name: Recall@500
+          value: 79.83
+        - type: recall_at_100
+          name: Recall@100
+          value: 66.78
+        - type: recall_at_10
+          name: Recall@10
+          value: 41.25
+        - type: map_at_10
+          name: MAP@10
+          value: 21.80
+        - type: ndcg_at_10
+          name: nDCG@10
+          value: 26.57
+        - type: mrr_at_10
+          name: MRR@10
+          value: 22.29
 ---
 
 # biencoder-mMiniLMv2-L6-mmarcoFR
 
-This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 384 dimensional dense vector space and can be used for tasks like clustering or semantic search. The model was trained on the **French** portion of the [mMARCO](https://huggingface.co/datasets/unicamp-dl/mmarco) dataset.
+This is a dense single-vector bi-encoder model. It maps sentences and paragraphs to a 384 dimensional dense vector space and should be used for semantic search. The model was trained on the **French** portion of the [mMARCO](https://huggingface.co/datasets/unicamp-dl/mmarco) retrieval dataset.
 
 ## Usage
-***
-
-#### Sentence-Transformers
 
-Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
+Here are some examples for using the model with [Sentence-Transformers](#using-sentence-transformers), [FlagEmbedding](#using-flagembedding), or [Huggingface Transformers](#using-huggingface-transformers).
 
-```
-pip install -U sentence-transformers
-```
+#### Using Sentence-Transformers
 
-Then you can use the model like this:
+Start by installing the [library](https://www.SBERT.net): `pip install -U sentence-transformers`. Then, you can use the model like this:
 
 ```python
 from sentence_transformers import SentenceTransformer
-sentences = ["This is an example sentence", "Each sentence is converted"]
+
+queries = ["Ceci est un exemple de requête.", "Voici un second exemple."]
+passages = ["Ceci est un exemple de passage.", "Et voilà un deuxième exemple."]
 
 model = SentenceTransformer('antoinelouis/biencoder-mMiniLMv2-L6-mmarcoFR')
-embeddings = model.encode(sentences)
-print(embeddings)
+q_embeddings = model.encode(queries, normalize_embeddings=True)
+p_embeddings = model.encode(passages, normalize_embeddings=True)
+
+similarity = q_embeddings @ p_embeddings.T
+print(similarity)
 ```
 
-#### 🤗 Transformers
+#### Using FlagEmbedding
 
-Without [sentence-transformers](https://www.SBERT.net), you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings.
+Start by installing the [library](https://github.com/FlagOpen/FlagEmbedding/): `pip install -U FlagEmbedding`. Then, you can use the model like this:
 
 ```python
-from transformers import AutoTokenizer, AutoModel
-import torch
+from FlagEmbedding import FlagModel
+
+queries = ["Ceci est un exemple de requête.", "Voici un second exemple."]
+passages = ["Ceci est un exemple de passage.", "Et voilà un deuxième exemple."]
+
+model = FlagModel('antoinelouis/biencoder-mMiniLMv2-L6-mmarcoFR')
+q_embeddings = model.encode(queries, normalize_embeddings=True)
+p_embeddings = model.encode(passages, normalize_embeddings=True)
+
+similarity = q_embeddings @ p_embeddings.T
+print(similarity)
+```
+
+#### Using Transformers
+
+Start by installing the [library](https://huggingface.co/docs/transformers): `pip install -U transformers`. Then, you can use the model like this:
 
+```python
+from transformers import AutoTokenizer, AutoModel
+from torch.nn.functional import normalize
 
-#Mean Pooling - Take attention mask into account for correct averaging
 def mean_pooling(model_output, attention_mask):
+    """ Perform mean pooling on-top of the contextualized word embeddings, while ignoring mask tokens in the mean computation."""
     token_embeddings = model_output[0] #First element of model_output contains all token embeddings
     input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
     return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
 
 
-# Sentences we want sentence embeddings for
-sentences = ['This is an example sentence', 'Each sentence is converted']
+queries = ["Ceci est un exemple de requête.", "Voici un second exemple."]
+passages = ["Ceci est un exemple de passage.", "Et voilà un deuxième exemple."]
 
-# Load model from HuggingFace Hub
 tokenizer = AutoTokenizer.from_pretrained('antoinelouis/biencoder-mMiniLMv2-L6-mmarcoFR')
 model = AutoModel.from_pretrained('antoinelouis/biencoder-mMiniLMv2-L6-mmarcoFR')
 
-# Tokenize sentences
-encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
-
-# Compute token embeddings
+q_input = tokenizer(queries, padding=True, truncation=True, return_tensors='pt')
+p_input = tokenizer(passages, padding=True, truncation=True, return_tensors='pt')
 with torch.no_grad():
-    model_output = model(**encoded_input)
-
-# Perform pooling. In this case, mean pooling.
-sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
-
-print("Sentence embeddings:")
-print(sentence_embeddings)
+    q_output = model(**encoded_queries)
+    p_output = model(**encoded_passages)
+q_embeddings = mean_pooling(q_output, q_input['attention_mask'])
+q_embedddings = normalize(q_embeddings, p=2, dim=1)
+p_embeddings = mean_pooling(p_output, p_input['attention_mask'])
+p_embedddings = normalize(p_embeddings, p=2, dim=1)
+
+similarity = q_embeddings @ p_embeddings.T
+print(similarity)
 ```
 
-## Evaluation
 ***
 
-We evaluated our model on the smaller development set of mMARCO-fr, which consists of 6,980 queries for a corpus of 8.8M candidate passages. Below, we compared the model performance with other biencoder models fine-tuned on the same dataset. We report the mean reciprocal rank (MRR), normalized discounted cumulative gainand (NDCG), mean average precision (MAP), and recall at various cut-offs (R@k).
+## Evaluation
+
+We evaluate the model on the smaller development set of [mMARCO-fr](https://ir-datasets.com/mmarco.html#mmarco/v2/fr/), which consists of 6,980 queries for a corpus of 8.8M candidate passages. Below, we compare the model performance with other biencoder models fine-tuned on the same dataset. We report the mean reciprocal rank (MRR), normalized discounted cumulative gainand (NDCG), mean average precision (MAP), and recall at various cut-offs (R@k).
 
 |    | model                                                                                                                   | Vocab. | #Param. |  Size |   MRR@10 |   NDCG@10 |   MAP@10 |   R@10 |   R@100(↑) |   R@500 |
 |---:|:------------------------------------------------------------------------------------------------------------------------|:-------|--------:|------:|---------:|----------:|---------:|-------:|-----------:|--------:|
@@ -91,25 +137,22 @@ We evaluated our model on the smaller development set of mMARCO-fr, which consis
 |  7 | [biencoder-electra-base-french-mmarcoFR](https://huggingface.co/antoinelouis/biencoder-electra-base-french-mmarcoFR)    |     🇫🇷 |    110M | 440MB |    23.38 |     27.97 |    22.91 |  43.50 |      68.96 |   81.61 |
 |  8 | **biencoder-mMiniLMv2-L6-mmarcoFR**                                                                                     | 🇫🇷,99+ |    107M | 428MB |    22.29 |     26.57 |    21.80 |  41.25 |      66.78 |   79.83 |
 
-## Training
 ***
 
-#### Background
+## Training
 
-We used the [nreimers/mMiniLMv2-L6-H384-distilled-from-XLMR-Large](https://huggingface.co/nreimers/mMiniLMv2-L6-H384-distilled-from-XLMR-Large) model and fine-tuned it on a 500K sentence pairs dataset in French. We used a contrastive learning objective: given a sentence from the pair, the model should predict which out of a set of randomly sampled other sentences, was actually paired with it in our dataset. Formally, we compute the cos similarity from each possible sentence pairs from the batch. We then apply the cross entropy loss with a temperature of 0.05 by comparing with true pairs.
+#### Data
 
-#### Hyperparameters
+We use the French training samples from the [mMARCO](https://huggingface.co/datasets/unicamp-dl/mmarco) dataset, a multilingual machine-translated version of MS MARCO that contains 8.8M passages and 539K training queries. We do not employ the BM25 netaives provided by the official dataset but instead sample harder negatives mined from 12 distinct dense retrievers, using the [msmarco-hard-negatives](https://huggingface.co/datasets/sentence-transformers/msmarco-hard-negatives) distillation dataset.
 
-We trained the model on a single Tesla V100 GPU with 32GBs of memory during 20 epochs (i.e., 65.7k steps) using a batch size of 152. We used the AdamW optimizer with an initial learning rate of 2e-05, weight decay of 0.01, learning rate warmup over the first 500 steps, and linear decay of the learning rate. The sequence length was limited to 128 tokens.
+#### Implementation
 
-#### Data
+The model is initialized from the [nreimers/mMiniLMv2-L6-H384-distilled-from-XLMR-Large](https://huggingface.co/nreimers/mMiniLMv2-L6-H384-distilled-from-XLMR-Large) 
+checkpoint and optimized via the cross-entropy loss (as in [DPR](https://doi.org/10.48550/arXiv.2004.04906)) with a temperature of 0.05. It is fine-tuned on one 32GB 
+NVIDIA V100 GPU for 20 epochs (i.e., 65.7k steps) using the AdamW optimizer with a batch size of 152, a peak learning rate of 2e-5 with warm up along the first 500 steps 
+and linear scheduling. We set the maximum sequence lengths for both the questions and passages to 128 tokens. We use the cosine similarity to compute relevance scores.
 
-We used the French version of the [mMARCO](https://huggingface.co/datasets/unicamp-dl/mmarco) dataset to fine-tune our model. mMARCO is a multi-lingual machine-translated version of the MS MARCO dataset, a large-scale IR dataset comprising:
-- a corpus of 8.8M passages;
-- a training set of ~533k queries (with at least one relevant passage);
-- a development set of ~101k queries;
-- a smaller dev set of 6,980 queries (which is actually used for evaluation in most published works).
-Link: [https://ir-datasets.com/mmarco.html#mmarco/v2/fr/](https://ir-datasets.com/mmarco.html#mmarco/v2/fr/)
+***
 
 ## Citation