1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
import torch

from aeca import Channel, SentenceTransformerEncoder


channel = Channel("localhost", 10080)
encoder = SentenceTransformerEncoder(channel, "all-MiniLM-L6-v2")

input_sentence = "This is a happy person."
sample_sentences = [
    "This is a sad person.",
    "This person is very happy.",
    "Today is a nice day.",
    "I feel so happy today."
]
sentences = [input_sentence] + sample_sentences
embeddings = torch.tensor(encoder.encode(sentences))
similarities = torch.cosine_similarity(embeddings[0], embeddings)
sorted_sentences = sorted(zip(sentences, similarities.tolist()), key=lambda item: item[1], reverse=True)
for sentence, similarity in sorted_sentences[1:]:
    print(f"{sentence}: {similarity:.3f}")