Den4ikAI/rubert-tiny2-retriever

1---2pipeline_tag: sentence-similarity3tags:4- sentence-transformers5- feature-extraction6- sentence-similarity7- transformers8license: mit9language:10- ru11widget:12  - source_sentence: "query: Когда родился Пушкин?"13 14    sentences:15      - "passage: Алекса́ндр Серге́евич Пу́шкин (26 мая [6 июня] 1799, Москва — 29 января [10 февраля] 1837, Санкт-Петербург) — русский поэт, драматург и прозаик, заложивший основы русского реалистического направления[2], литературный критик[3] и теоретик литературы, историк[3], публицист, журналист[3]."16      - "passage: Пушкин ловил кайф со своими друзьями"17      - "passage: Пушкин из самых авторитетных литературных деятелей первой трети XIX века. Ещё при жизни Пушкина сложилась его репутация величайшего национального русского поэта[4][5]. Пушкин рассматривается как основоположник современного русского литературного языка[~ 2]."18---19 20# {MODEL_NAME}21 22This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 312 dimensional dense vector space and can be used for tasks like clustering or semantic search.23 24<!--- Describe your model here -->25 26## Usage (Sentence-Transformers)27 28Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:29 30```31pip install -U sentence-transformers32```33 34Then you can use the model like this:35 36```python37from sentence_transformers import SentenceTransformer38sentences = ["This is an example sentence", "Each sentence is converted"]39 40model = SentenceTransformer('{MODEL_NAME}')41embeddings = model.encode(sentences)42print(embeddings)43```44 45 46 47## Usage (HuggingFace Transformers)48Without [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.49 50```python51from transformers import AutoTokenizer, AutoModel52import torch53 54 55#Mean Pooling - Take attention mask into account for correct averaging56def mean_pooling(model_output, attention_mask):57    token_embeddings = model_output[0] #First element of model_output contains all token embeddings58    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()59    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)60 61 62# Sentences we want sentence embeddings for63sentences = ['This is an example sentence', 'Each sentence is converted']64 65# Load model from HuggingFace Hub66tokenizer = AutoTokenizer.from_pretrained('{MODEL_NAME}')67model = AutoModel.from_pretrained('{MODEL_NAME}')68 69# Tokenize sentences70encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')71 72# Compute token embeddings73with torch.no_grad():74    model_output = model(**encoded_input)75 76# Perform pooling. In this case, mean pooling.77sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])78 79print("Sentence embeddings:")80print(sentence_embeddings)81```82 83 84 85## Evaluation Results86 87<!--- Describe how your model was evaluated -->88 89For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})90 91 92## Training93The model was trained with the parameters:94 95**DataLoader**:96 97`torch.utils.data.dataloader.DataLoader` of length 966 with parameters:98```99{'batch_size': 10, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}100```101 102**Loss**:103 104`sentence_transformers.losses.ContrastiveLoss.ContrastiveLoss` with parameters:105  ```106  {'distance_metric': 'SiameseDistanceMetric.COSINE_DISTANCE', 'margin': 0.5, 'size_average': True}107  ```108 109Parameters of the fit()-Method:110```111{112    "epochs": 10,113    "evaluation_steps": 500,114    "evaluator": "sentence_transformers.evaluation.BinaryClassificationEvaluator.BinaryClassificationEvaluator",115    "max_grad_norm": 1,116    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",117    "optimizer_params": {118        "lr": 2e-05119    },120    "scheduler": "WarmupLinear",121    "steps_per_epoch": null,122    "warmup_steps": 966,123    "weight_decay": 1e-05124}125```126 127 128## Full Model Architecture129```130SentenceTransformer(131  (0): Transformer({'max_seq_length': 2048, 'do_lower_case': False}) with Transformer model: BertModel 132  (1): Pooling({'word_embedding_dimension': 312, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})133)134```135 136## Citing & Authors137 138<!--- Describe where people can find more information -->