Method: Vecsearch::GTETiny#encode

Defined in:
lib/vecsearch/gte_tiny.rb

#encode(sentence, n_threads: 1) ⇒ Object 



45
46
47
48
49
50
51
52
53
54
55
 
# File 'lib/vecsearch/gte_tiny.rb', line 45

def encode(sentence, n_threads: 1)
  # Encode the sentence into token embeddings
  token_embeddings = encode_batch([sentence], n_threads: 1)

  # Pool the token embeddings into a sentence embedding
  # For simplicity, we'll use an attention mask of all ones
  attention_mask = Array.new(token_embeddings.first.length, 1)
  # sentence_embedding = mean_pooling(token_embeddings, attention_mask)
  # sentence_embedding
  token_embeddings
end