improve: Updated fastembed models with smaller models

improve: Updated vectors with smaller ones
qdrant · Mar 11, 2025 · 7cfedd1 · 7cfedd1
1 parent d8951c5
commit 7cfedd1
Show file tree

Hide file tree

Showing 10 changed files with 33 additions and 35 deletions.
diff --git a/tests/congruence_tests/test_common.py b/tests/congruence_tests/test_common.py
@@ -14,17 +14,17 @@
 COLLECTION_NAME = "congruence_test_collection"
 
 # dense vectors sizes
-text_vector_size = 50
-image_vector_size = 100
-code_vector_size = 80
+text_vector_size = 25
+image_vector_size = 50
+code_vector_size = 40
 
 # sparse vectors sizes
-sparse_text_vector_size = 100
-sparse_image_vector_size = 1_000
-sparse_code_vector_size = 10_000
+sparse_text_vector_size = 50
+sparse_image_vector_size = 500
+sparse_code_vector_size = 1_000
 
 # number of vectors to generate
-NUM_VECTORS = 1000
+NUM_VECTORS = 100
 
 dense_vectors_config = {
     "text": models.VectorParams(

diff --git a/tests/congruence_tests/test_count.py b/tests/congruence_tests/test_count.py
@@ -38,7 +38,7 @@ def test_simple_count():
 
     compare_client_results(local_client, remote_client, count_all)
 
-    for i in range(100):
+    for i in range(50):
         count_filter = one_random_filter_please()
         try:
             compare_client_results(
@@ -60,7 +60,7 @@ def test_simple_sparse_search():
 
     compare_client_results(local_client, remote_client, count_all)
 
-    for i in range(100):
+    for i in range(50):
         count_filter = one_random_filter_please()
         try:
             compare_client_results(

diff --git a/tests/congruence_tests/test_group_search.py b/tests/congruence_tests/test_group_search.py
@@ -303,7 +303,7 @@ def test_simple_group_search():
     compare_client_results(local_client, remote_client, searcher.group_search_image_select_vector)
     compare_client_results(local_client, remote_client, searcher.group_search_payload_exclude)
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_client_results(
@@ -336,7 +336,7 @@ def test_single_vector():
     for group_size in (1, 5):
         searcher.group_size = group_size
 
-        for i in range(100):
+        for i in range(50):
             query_filter = one_random_filter_please()
 
             try:

diff --git a/tests/congruence_tests/test_query.py b/tests/congruence_tests/test_query.py
@@ -859,7 +859,7 @@ def test_dense_query_filtered_prefetch():
 
     local_client, http_client, grpc_client = init_clients(fixture_points)
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_clients_results(
@@ -1001,7 +1001,7 @@ def test_dense_query():
     )
     compare_clients_results(local_client, http_client, grpc_client, searcher.dense_payload_exclude)
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_clients_results(
@@ -1131,7 +1131,7 @@ def test_simple_opt_vectors_query():
     )
     compare_clients_results(local_client, http_client, grpc_client, searcher.dense_payload_exclude)
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_clients_results(
@@ -1167,7 +1167,7 @@ def test_single_dense_vector():
         fixture_points, vectors_config=vectors_config
     )
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_clients_results(
@@ -1470,7 +1470,7 @@ def test_query_group():
         )
 
     searcher.group_by = "city.name"
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_clients_results(

diff --git a/tests/congruence_tests/test_search.py b/tests/congruence_tests/test_search.py
@@ -166,7 +166,7 @@ def test_simple_search():
     compare_client_results(local_client, remote_client, searcher.simple_search_image_select_vector)
     compare_client_results(local_client, remote_client, searcher.search_payload_exclude)
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_client_results(
@@ -198,7 +198,7 @@ def test_simple_opt_vectors_search():
     compare_client_results(local_client, remote_client, searcher.simple_search_image_select_vector)
     compare_client_results(local_client, remote_client, searcher.search_payload_exclude)
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_client_results(
@@ -225,7 +225,7 @@ def test_single_vector():
     remote_client = init_remote()
     init_client(remote_client, fixture_points, vectors_config=vectors_config)
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_client_results(

diff --git a/tests/congruence_tests/test_sparse_search.py b/tests/congruence_tests/test_sparse_search.py
@@ -161,7 +161,7 @@ def test_simple_search():
     compare_client_results(local_client, remote_client, searcher.simple_search_image_select_vector)
     compare_client_results(local_client, remote_client, searcher.search_payload_exclude)
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_client_results(
@@ -192,7 +192,7 @@ def test_simple_opt_vectors_search():
     compare_client_results(local_client, remote_client, searcher.simple_search_image_select_vector)
     compare_client_results(local_client, remote_client, searcher.search_payload_exclude)
 
-    for i in range(100):
+    for i in range(50):
         query_filter = one_random_filter_please()
         try:
             compare_client_results(

diff --git a/tests/embed_tests/test_local_inference.py b/tests/embed_tests/test_local_inference.py
@@ -14,13 +14,13 @@
 
 
 COLLECTION_NAME = "inference_collection"
-DENSE_MODEL_NAME = "sentence-transformers/all-MiniLM-L6-v2"
+DENSE_MODEL_NAME = "BAAI/bge-small-en-v1.5"
 DENSE_DIM = 384
 SPARSE_MODEL_NAME = "Qdrant/bm42-all-minilm-l6-v2-attentions"
-COLBERT_MODEL_NAME = "colbert-ir/colbertv2.0"
-COLBERT_DIM = 128
-DENSE_IMAGE_MODEL_NAME = "Qdrant/resnet50-onnx"
-DENSE_IMAGE_DIM = 2048
+COLBERT_MODEL_NAME = "answerdotai/answerai-colbert-small-v1"
+COLBERT_DIM = 96
+DENSE_IMAGE_MODEL_NAME = "Qdrant/clip-ViT-B-32-vision"
+DENSE_IMAGE_DIM = 512
 
 TEST_IMAGE_PATH = Path(__file__).parent / "misc" / "image.jpeg"
 

diff --git a/tests/test_async_qdrant_client.py b/tests/test_async_qdrant_client.py
@@ -72,7 +72,7 @@ async def test_async_qdrant_client(prefer_grpc):
                 vector=np.random.rand(10).tolist(),
                 payload={"random_dig": random.randint(1, 100)},
             )
-            for i in range(100)
+            for i in range(50)
         ],
     )
     assert (await client.count(COLLECTION_NAME)).count == 100
@@ -314,7 +314,7 @@ async def test_async_qdrant_client_local():
                 vector=np.random.rand(10).tolist(),
                 payload={"random_dig": random.randint(1, 100)},
             )
-            for i in range(100)
+            for i in range(50)
         ],
     )
     assert (await client.count(COLLECTION_NAME)).count == 100

diff --git a/tests/test_local_persistence.py b/tests/test_local_persistence.py
@@ -37,7 +37,7 @@ def ingest_dense_vector_data(
         collection_name=collection_name,
         points=construct(
             rest.Batch,
-            ids=random.sample(range(100), len(lines)),
+            ids=random.sample(range(50), len(lines)),
             vectors=embeddings,
         ),
     )
@@ -67,7 +67,7 @@ def ingest_sparse_vector_data(
 
     batch = construct(
         rest.Batch,
-        ids=random.sample(range(100), vector_count),
+        ids=random.sample(range(50), vector_count),
         vectors={"text": sparse_vectors},
     )
 

diff --git a/tests/test_migrate.py b/tests/test_migrate.py
@@ -16,7 +16,7 @@
 from tests.utils import read_version
 
 
-VECTOR_NUMBER = 1000
+VECTOR_NUMBER = 100
 
 
 @pytest.fixture
@@ -478,9 +478,7 @@ def test_recreate_collection(remote_client: QdrantClient):
             compression=models.CompressionRatio.X4, always_ram=False
         )
     )
-    strict_mode_config = models.StrictModeConfig(
-        read_rate_limit=1000000
-    )
+    strict_mode_config = models.StrictModeConfig(read_rate_limit=1000000)
     remote_client.create_collection(
         collection_name,
         vectors_config=vectors_config,
@@ -492,7 +490,7 @@ def test_recreate_collection(remote_client: QdrantClient):
         optimizers_config=optimizers_config,
         wal_config=wal_config,
         quantization_config=general_quantization_config,
-        strict_mode_config=strict_mode_config
+        strict_mode_config=strict_mode_config,
     )
 
     remote_client.create_payload_index(