nemesis-gte-tiny

{nemesis-gte-tiny}

This is a sentence-transformers model: It maps sentences & paragraphs to a 384 dimensional dense vector space and can be used for tasks like clustering or semantic search.

It is fine-tuned from TaylorAI/gte-tiny on public documents processed through a Nemesis pipeline. The ~2500 documents were chunked into 512 tokens and submitted to Gemini for Question/Answer generation. Each query generated 2 questions, and the entire process was executed twice, resulting in ~10k questions generated for context chunks.

The positive chunks were linked to each qusetion and 5 random text chunks from documents other than the source were used as the negative training samples. We followed the guide from FlagEmbedding for fine tuning. The fine tuned model was merged back with the TaylorAI/gte-tiny base using LM_Cocktail as the guide described.

Usage (Sentence-Transformers)

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

pip install -U sentence-transformers

Then you can use the model like this:

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

model = SentenceTransformer('Harmj0y/nemesis-gte-tiny')
embeddings = model.encode(sentences)
print(embeddings)

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('Harmj0y/nemesis-gte-tiny')
model = AutoModel.from_pretrained('Harmj0y/nemesis-gte-tiny')

# 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

<!--- Describe how your model was evaluated -->

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

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
)

Citing & Authors

Fine tuned from TaylorAI/gte-tiny using FlagEmbedding's embedding fine-tuning guide.