|
| 1 | +from colpali_engine.models import ColQwen2, ColQwen2Processor |
| 2 | +from colpali_engine.utils.torch_utils import get_torch_device |
| 3 | +from datasets import load_dataset |
| 4 | +from pgvector.psycopg import register_vector, Bit |
| 5 | +import psycopg |
| 6 | +import torch |
| 7 | + |
| 8 | +conn = psycopg.connect(dbname='pgvector_example', autocommit=True) |
| 9 | + |
| 10 | +conn.execute('CREATE EXTENSION IF NOT EXISTS vector') |
| 11 | +register_vector(conn) |
| 12 | + |
| 13 | +conn.execute('DROP TABLE IF EXISTS documents') |
| 14 | +conn.execute('CREATE TABLE documents (id bigserial PRIMARY KEY, embeddings bit(128)[])') |
| 15 | +conn.execute(""" |
| 16 | +CREATE OR REPLACE FUNCTION max_sim(document bit[], query bit[]) RETURNS double precision AS $$ |
| 17 | + WITH queries AS ( |
| 18 | + SELECT row_number() OVER () AS query_number, * FROM (SELECT unnest(query) AS query) |
| 19 | + ), |
| 20 | + documents AS ( |
| 21 | + SELECT unnest(document) AS document |
| 22 | + ), |
| 23 | + similarities AS ( |
| 24 | + SELECT query_number, 1 - ((document <~> query) / bit_length(query)) AS similarity FROM queries CROSS JOIN documents |
| 25 | + ), |
| 26 | + max_similarities AS ( |
| 27 | + SELECT MAX(similarity) AS max_similarity FROM similarities GROUP BY query_number |
| 28 | + ) |
| 29 | + SELECT SUM(max_similarity) FROM max_similarities |
| 30 | +$$ LANGUAGE SQL |
| 31 | +""") |
| 32 | + |
| 33 | +device = get_torch_device('auto') |
| 34 | +model = ColQwen2.from_pretrained('vidore/colqwen2-v1.0', torch_dtype=torch.bfloat16, device_map=device).eval() |
| 35 | +processor = ColQwen2Processor.from_pretrained('vidore/colqwen2-v1.0') |
| 36 | + |
| 37 | + |
| 38 | +def generate_embeddings(processed): |
| 39 | + with torch.no_grad(): |
| 40 | + return model(**processed.to(model.device)).to(torch.float32).numpy(force=True) |
| 41 | + |
| 42 | + |
| 43 | +def binary_quantize(embedding): |
| 44 | + return Bit(embedding > 0) |
| 45 | + |
| 46 | + |
| 47 | +input = load_dataset('vidore/docvqa_test_subsampled', split='test[:3]')['image'] |
| 48 | +for content in input: |
| 49 | + embeddings = [binary_quantize(e) for e in generate_embeddings(processor.process_images([content]))[0]] |
| 50 | + conn.execute('INSERT INTO documents (embeddings) VALUES (%s)', (embeddings,)) |
| 51 | + |
| 52 | +query = 'dividend' |
| 53 | +query_embeddings = [binary_quantize(e) for e in generate_embeddings(processor.process_queries([query]))[0]] |
| 54 | +result = conn.execute('SELECT id, max_sim(embeddings, %s) AS max_sim FROM documents ORDER BY max_sim DESC LIMIT 5', (query_embeddings,)).fetchall() |
| 55 | +for row in result: |
| 56 | + print(row) |
0 commit comments