St Greek Media Longformer 4096

🚀 Greek Media SLF (Sentence-Longformer)

This model is based on sentence-transformers and the Greek Media Longformer. It maps sentences and paragraphs to a 768-dimensional dense vector space and can be used for tasks such as clustering or semantic search.

📋 Model Information

🚀 Quick Start

📦 Installation

Using this model becomes easy when you have sentence-transformers installed:

pip install -U sentence-transformers

💻 Usage Examples

🔍 Basic Usage (Sentence-Transformers)

from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('dimitriz/st-greek-media-longformer-4096')
embeddings = model.encode(sentences)
print(embeddings)

⚙️ Advanced Usage (HuggingFace Transformers)

Without sentence-transformers, 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.

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 = ['This is an example sentence', 'Each sentence is converted']

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('dimitriz/st-greek-media-longformer-4096')
model = AutoModel.from_pretrained('dimitriz/st-greek-media-longformer-4096')

# 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

For an automated evaluation of this model, see the Sentence Embeddings Benchmark: https://seb.sbert.net

🔧 Training Details

The model was trained on a custom dataset containing triplets from the combined Greek 'internet', 'social-media' and 'press' domains, described in the paper DACL.

• The dataset was created by sampling triplets of sentences from the same domain, where the first two sentences are more similar than the third one.
• Training objective was to maximize the similarity between the first two sentences and minimize the similarity between the first and the third sentence.
• The model was trained for 3 epochs with a batch size of 2 and a maximum sequence length of 4096 tokens.
• The model was trained on a single NVIDIA RTX A6000 GPU with 48GB of memory.

⚙️ Training Parameters

DataLoader: torch.utils.data.dataloader.DataLoader of length 172897 with parameters:

{'batch_size': 1, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}

Loss: sentence_transformers.losses.TripletLoss.TripletLoss with parameters:

{'distance_metric': 'TripletDistanceMetric.EUCLIDEAN', 'triplet_margin': 5}

Parameters of the fit()-Method:

{
    "epochs": 3,
    "evaluation_steps": 1000,
    "evaluator": "sentence_transformers.evaluation.TripletEvaluator.TripletEvaluator",
    "max_grad_norm": 1,
    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
    "optimizer_params": {
        "lr": 2e-05
    },
    "scheduler": "WarmupLinear",
    "steps_per_epoch": null,
    "warmup_steps": 17290,
    "weight_decay": 0.01
}

📚 Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 4096, 'do_lower_case': False}) with Transformer model: LongformerModel
  (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})
)

📄 Citing & Authors

The model has been officially released with the article "From Pre-training to Meta-Learning: A journey in Low-Resource-Language Representation Learning". Dimitrios Zaikis and Ioannis Vlahavas. In: IEEE Access.

If you use the model, please cite the following:

@ARTICLE{10288436,
    author =  {Zaikis, Dimitrios and Vlahavas, Ioannis},
    journal = {IEEE Access},
    title =   {From Pre-training to Meta-Learning: A journey in Low-Resource-Language Representation Learning},
    year =    {2023},
    volume =  {},
    number =  {},
    pages =   {1-1},
    doi =     {10.1109/ACCESS.2023.3326337}
  }