initial commit

1_Pooling/config.json

@@ -0,0 +1,7 @@
+{
+  "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
+}

README.md

@@ -0,0 +1,106 @@
+---
+pipeline_tag: sentence-similarity
+language: fr
+tags:
+  - sentence-similarity
+  - transformers
+  - fr
+  - flaubert
+  - sentence-transformers
+  - feature-extraction
+  - xnli
+  - stsb_multi_mt
+datasets:
+  - xnli
+  - stsb_multi_mt
+---
+
+# inokufu/flaubert-base-uncased-xnli-sts
+
+This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.
+
+## Details
+
+This model is based on the French flaubert-base-uncased pre-trained model [1, 2].
+
+It was then fine-tuned on a natural language inference task (XNLI) [3]. This task consists in training the model to recognize relations between sentences (contradiction, neutral, implication).
+
+It was then fine-tuned on a text semantic similarity task (on STS-fr data) [4]. This task consists in training the model to estimate the similarity between two sentences.
+
+This fine-tuning process allows our model to have a semantic representation of words that is much better than the one proposed by the base model.
+
+## Usage (Sentence-Transformers)
+
+Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
+
+```
+pip install -U sentence-transformers
+```
+
+Then you can use the model like this:
+
+```python
+from sentence_transformers import SentenceTransformer
+sentences = ["Apprendre le python", "Devenir expert en comptabilité"]
+
+model = SentenceTransformer('inokufu/flaubert-base-uncased-xnli-sts')
+embeddings = model.encode(sentences)
+print(embeddings)
+```
+
+## Usage (HuggingFace Transformers)
+
+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.
+
+```python
+from transformers import AutoTokenizer, AutoModel
+import torch
+
+
+#Mean Pooling - Take attention mask into account for correct averaging
+def mean_pooling(model_output, attention_mask):
+    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 = ["Apprendre le python", "Devenir expert en comptabilité"]
+
+# Load model from HuggingFace Hub
+tokenizer = AutoTokenizer.from_pretrained('inokufu/flaubert-base-uncased-xnli-sts')
+model = AutoModel.from_pretrained('inokufu/flaubert-base-uncased-xnli-sts')
+
+# Tokenize sentences
+encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
+
+# Compute token embeddings
+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)
+```
+
+## Evaluation Results
+
+STS (fr) score: 83.07%
+
+## Model Architecture
+
+```
+SentenceTransformer(
+  (0): Transformer({'max_seq_length': 512, 'do_lower_case': True}) with Transformer model: FlaubertModel
+  (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})
+)
+```
+
+## References
+
+[1] https://hal.archives-ouvertes.fr/hal-02784776v3/document <br>
+[2] https://huggingface.co/flaubert/flaubert_base_uncased <br>
+[3] https://arxiv.org/abs/1809.05053 <br>
+[4] https://huggingface.co/datasets/stsb_multi_mt <br>

config.json

@@ -0,0 +1,80 @@
+{
+  "_name_or_path": "inokufu/flaubert-base-uncased-xnli-sts",
+  "amp": 1,
+  "architectures": [
+    "FlaubertModel"
+  ],
+  "asm": false,
+  "attention_dropout": 0.1,
+  "bos_index": 0,
+  "bos_token_id": 0,
+  "bptt": 512,
+  "causal": false,
+  "clip_grad_norm": 5,
+  "dropout": 0.1,
+  "emb_dim": 768,
+  "embed_init_std": 0.02209708691207961,
+  "encoder_only": true,
+  "end_n_top": 5,
+  "eos_index": 1,
+  "fp16": true,
+  "gelu_activation": true,
+  "group_by_size": true,
+  "id2lang": {
+    "0": "fr"
+  },
+  "init_std": 0.02,
+  "is_encoder": true,
+  "lang2id": {
+    "fr": 0
+  },
+  "lang_id": 0,
+  "langs": [
+    "fr"
+  ],
+  "layer_norm_eps": 1e-12,
+  "layerdrop": 0.0,
+  "lg_sampling_factor": -1,
+  "lgs": "fr",
+  "mask_index": 5,
+  "mask_token_id": 0,
+  "max_batch_size": 0,
+  "max_position_embeddings": 512,
+  "max_vocab": -1,
+  "mlm_steps": [
+    [
+      "fr",
+      null
+    ]
+  ],
+  "model_type": "flaubert",
+  "n_heads": 12,
+  "n_langs": 1,
+  "n_layers": 12,
+  "pad_index": 2,
+  "pad_token_id": 2,
+  "pre_norm": false,
+  "sample_alpha": 0,
+  "share_inout_emb": true,
+  "sinusoidal_embeddings": false,
+  "start_n_top": 5,
+  "summary_activation": null,
+  "summary_first_dropout": 0.1,
+  "summary_proj_to_labels": true,
+  "summary_type": "first",
+  "summary_use_proj": true,
+  "tokens_per_batch": -1,
+  "torch_dtype": "float32",
+  "transformers_version": "4.11.3",
+  "unk_index": 3,
+  "use_lang_emb": true,
+  "vocab_size": 67542,
+  "word_blank": 0,
+  "word_dropout": 0,
+  "word_keep": 0.1,
+  "word_mask": 0.8,
+  "word_mask_keep_rand": "0.8,0.1,0.1",
+  "word_pred": 0.15,
+  "word_rand": 0.1,
+  "word_shuffle": 0
+}

config_sentence_transformers.json

@@ -0,0 +1,7 @@
+{
+  "__version__": {
+    "sentence_transformers": "2.1.0",
+    "transformers": "4.11.3",
+    "pytorch": "1.9.0+cu102"
+  }
+}

eval/similarity_evaluation_sts-dev_results.csv

@@ -0,0 +1,5 @@
+epoch,steps,cosine_pearson,cosine_spearman,euclidean_pearson,euclidean_spearman,manhattan_pearson,manhattan_spearman,dot_pearson,dot_spearman
+0,-1,0.8476134716760887,0.8468186524889956,0.8374628393139093,0.8393587213401801,0.8361178664176142,0.8379331088259698,0.822123471306284,0.8200800390948763
+1,-1,0.8585223492907658,0.8570807869134168,0.8424447467205926,0.8452738848553554,0.8416316010899966,0.8442137608694786,0.8304583702559676,0.8305974975814222
+2,-1,0.8571486998231743,0.8559970703038707,0.8426665991938748,0.8453606609929885,0.8418315419031606,0.8446965654748465,0.829168847143357,0.8290776387285839
+3,-1,0.8578583320115634,0.8563828855356073,0.8415682855261146,0.8444170090920187,0.8408005059135651,0.8439627134028199,0.8303650556789606,0.8306674833948446

modules.json

@@ -0,0 +1,14 @@
+[
+  {
+    "idx": 0,
+    "name": "0",
+    "path": "",
+    "type": "sentence_transformers.models.Transformer"
+  },
+  {
+    "idx": 1,
+    "name": "1",
+    "path": "1_Pooling",
+    "type": "sentence_transformers.models.Pooling"
+  }
+]

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e833a8a240f04e8e3b19bacad9b0b6ff09768429d47c116e701b35a70a0c40b
+size 549350969

sentence_bert_config.json

@@ -0,0 +1,4 @@
+{
+  "max_seq_length": 512,
+  "do_lower_case": true
+}

similarity_evaluation_sts-test_results.csv

@@ -0,0 +1,3 @@
+epoch,steps,cosine_pearson,cosine_spearman,euclidean_pearson,euclidean_spearman,manhattan_pearson,manhattan_spearman,dot_pearson,dot_spearman
+-1,-1,0.8281639363790627,0.8306544280206849,0.8144472895806177,0.8190524312029783,0.8146705390507208,0.8188915767762216,0.7682952962055747,0.7565634657936158
+-1,-1,0.8281638924716349,0.830654716456512,0.8144472832412396,0.8190596125240516,0.8146705315250665,0.8188990444980802,0.7682951830461527,0.7565642629626018

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "</s>", "mask_token": "<special1>", "additional_special_tokens": ["<special0>", "<special1>", "<special2>", "<special3>", "<special4>", "<special5>", "<special6>", "<special7>", "<special8>", "<special9>"]}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "<unk>", "bos_token": "<s>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "</s>", "mask_token": "<special1>", "additional_special_tokens": ["<special0>", "<special1>", "<special2>", "<special3>", "<special4>", "<special5>", "<special6>", "<special7>", "<special8>", "<special9>"], "lang2id": null, "id2lang": null, "do_lowercase_and_remove_accent": true, "do_lower_case": true, "model_max_length": 512, "special_tokens_map_file": null, "tokenizer_file": null, "name_or_path": "inokufu/flaubert-base-uncased-xnli-sts", "tokenizer_class": "FlaubertTokenizer"}