gubartz/st_scibert_abstruct

1---2pipeline_tag: sentence-similarity3tags:4- sentence-transformers5- feature-extraction6- sentence-similarity7- transformers8 9---10 11# {MODEL_NAME}12 13This 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.14 15<!--- Describe your model here -->16 17## Usage (Sentence-Transformers)18 19Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:20 21```22pip install -U sentence-transformers23```24 25Then you can use the model like this:26 27```python28from sentence_transformers import SentenceTransformer29sentences = ["This is an example sentence", "Each sentence is converted"]30 31model = SentenceTransformer('{MODEL_NAME}')32embeddings = model.encode(sentences)33print(embeddings)34```35 36 37 38## Usage (HuggingFace Transformers)39Without [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.40 41```python42from transformers import AutoTokenizer, AutoModel43import torch44 45 46#Mean Pooling - Take attention mask into account for correct averaging47def mean_pooling(model_output, attention_mask):48    token_embeddings = model_output[0] #First element of model_output contains all token embeddings49    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()50    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)51 52 53# Sentences we want sentence embeddings for54sentences = ['This is an example sentence', 'Each sentence is converted']55 56# Load model from HuggingFace Hub57tokenizer = AutoTokenizer.from_pretrained('{MODEL_NAME}')58model = AutoModel.from_pretrained('{MODEL_NAME}')59 60# Tokenize sentences61encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')62 63# Compute token embeddings64with torch.no_grad():65    model_output = model(**encoded_input)66 67# Perform pooling. In this case, mean pooling.68sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])69 70print("Sentence embeddings:")71print(sentence_embeddings)72```73 74 75 76## Evaluation Results77 78<!--- Describe how your model was evaluated -->79 80For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})81 82 83## Training84The model was trained with the parameters:85 86**DataLoader**:87 88`torch.utils.data.dataloader.DataLoader` of length 352 with parameters:89```90{'batch_size': 32, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}91```92 93**Loss**:94 95`sentence_transformers.losses.BatchAllTripletLoss.BatchAllTripletLoss` 96 97Parameters of the fit()-Method:98```99{100    "epochs": 20,101    "evaluation_steps": 0,102    "evaluator": "sentence_transformers.evaluation.TripletEvaluator.TripletEvaluator",103    "max_grad_norm": 1,104    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",105    "optimizer_params": {106        "lr": 2e-05107    },108    "scheduler": "WarmupLinear",109    "steps_per_epoch": null,110    "warmup_steps": 704,111    "weight_decay": 0.01112}113```114 115 116## Full Model Architecture117```118SentenceTransformer(119  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 120  (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})121)122```123 124## Citing & Authors125 126<!--- Describe where people can find more information -->