[Bug]: [int8_vector] The int8_vector is not shown in the results when specifying the int8_vector field as output_fields for query

[binbinlv]

Is there an existing issue for this?

• I have searched the existing issues

Environment

- Milvus version: master-20250211-85c9f92f-amd64
- Deployment mode(standalone or cluster):both
- MQ type(rocksmq, pulsar or kafka):    all
- SDK version(e.g. pymilvus v2.0.0rc2): 2.6.0rc71
- OS(Ubuntu or CentOS): 
- CPU/Memory: 
- GPU: 
- Others:

Current Behavior

The int8_vector is not shown in the results when specifying the int8_vector field as output_fields for query

>>> client.query(collection_name="customized_setup_1", filter="my_id>=0", output_fields=["my_vector"])
data: ["{'my_id': 0}"]

Expected Behavior

The int8_vector is shown in the results when specifying the int8_vector field as output_fields for query

just like float_vector:

>>> client.query(collection_name="customized_setup_2", filter="my_id>=0", output_fields=["my_vector"])
data: ["{'my_id': 0, 'my_vector': [0.48306835, 0.9815398, 0.13763158, 0.0997467, 0.14570269, 0.83381623, 0.2533495, 0.5141766, 3.248233e-06, 0.7560254, 0.78803337, 0.20656978, 0.5998218, 0.92462415, 0.5946385, 0.6319435, 0.35751677, 0.13305469, 0.35023752, 0.2134315, 0.7266131, 0.45943576, 0.4115885, 0.094741896, 0.42435962, 0.36004803, 0.3514786, 0.5841294, 0.71644586, 0.17080733, 0.014636853, 0.12820345, 0.8857452, 0.63410044, 0.6873833, 0.18846478, 0.22275887, 0.33786368, 0.40385348, 0.48239967, 0.0026390124, 0.90537494, 0.019583173, 0.14822645, 0.120368354, 0.5146042, 0.2832205, 0.78208506, 0.6385497, 0.12974623, 0.86614126, 0.32375425, 0.94300115, 0.29979256, 0.9403251, 0.35166037, 0.38923737, 0.046944916, 0.16529597, 0.66263044, 0.70765626, 0.9447244, 0.9664119, 0.82592314, 0.29692733, 0.55993456, 0.88830036, 0.7149717, 0.13324727, 0.24039853, 0.5394017, 0.3482095, 0.6519117, 0.9961026, 0.09185784, 0.7966022, 0.37718624, 0.37350395, 0.07291514, 0.5765629, 0.9380147, 0.8548175, 0.16825835, 0.047109794, 0.02843495, 0.58276457, 0.8786268, 0.6065395, 0.24458265, 0.3322088, 0.53052264, 0.26629698, 0.30301484, 0.7512133, 0.7805258, 0.47281718, 0.2468269, 0.27081817, 0.6599332, 0.20857278, 0.586381, 0.57643646, 0.75740004, 0.8954807, 0.5434921, 0.18676259, 0.50763893, 0.4595938, 0.90946364, 0.036301173, 0.8530404, 0.80743694, 0.16855665, 0.18533601, 0.8363561, 0.2781933, 0.5601252, 0.62934434, 0.73877335, 0.9165499, 0.96782345, 0.07416456, 0.15385732, 0.4842282, 0.011274599, 0.0977744, 0.345789, 0.38579237]}"]

Steps To Reproduce

from pymilvus import MilvusClient, DataType
import numpy as np
import random


client = MilvusClient()
dim=128
schema = client.create_schema()
schema.add_field(field_name="my_id", datatype=DataType.INT64, is_primary=True)
schema.add_field(field_name="my_vector", datatype=DataType.INT8_VECTOR, dim=128)
schema.add_field(field_name="my_varchar", datatype=DataType.VARCHAR, max_length=512)

index_params = client.prepare_index_params()
index_params.add_index(field_name="my_vector", index_type="AUTOINDEX", metric_type="COSINE")
client.create_collection(collection_name="customized_setup_1", schema=schema, index_params=index_params)

def gen_int8_vectors(num, dim):
     raw_vectors = []
     int8_vectors = []
     for _ in range(num):
        raw_vector = [random.randint(-128, 127) for _ in range(dim)]
        raw_vectors.append(raw_vector)
        int8_vector = np.array(raw_vector, dtype=np.int8)
        int8_vectors.append(int8_vector)
     return raw_vectors, int8_vectors

_, vectors = gen_int8_vectors(1, dim)

data = [{"my_id": 0, "my_vector": vectors[0], "my_varchar": "varchar"}]
client.insert(collection_name="customized_setup_1", data=data)

client.flush(collection_name="customized_setup_1")
client.query(collection_name="customized_setup_1", filter="my_id>=0", output_fields=["my_vector"])

Milvus Log

https://grafana-4am.zilliz.cc/explore?orgId=1&panes=%7B%22Fjq%22:%7B%22datasource%22:%22vhI6Vw67k%22,%22queries%22:%5B%7B%22refId%22:%22A%22,%22expr%22:%22%7Bcluster%3D%5C%22devops%5C%22,namespace%3D%5C%22chaos-testing%5C%22,pod%3D~%5C%22master-latest-lpbqj.%2A%5C%22%7D%22,%22datasource%22:%7B%22type%22:%22loki%22,%22uid%22:%22vhI6Vw67k%22%7D%7D%5D,%22range%22:%7B%22from%22:%22now-1h%22,%22to%22:%22now%22%7D%7D%7D&schemaVersion=1

Anything else?

collection name: customized_setup_1
timeline: about 2025-02-13 14:38

[binbinlv]

/assign @cydrain

[binbinlv]

And the default result for query is different between float_vector and int8_vector too:

float_vector: (vector is included)

>>> client.query(collection_name="customized_setup_2", filter="my_id>=0")
data: ["{'my_id': 0, 'my_vector': [0.48306835, 0.9815398, 0.13763158, 0.0997467, 0.14570269, 0.83381623, 0.2533495, 0.5141766, 3.248233e-06, 0.7560254, 0.78803337, 0.20656978, 0.5998218, 0.92462415, 0.5946385, 0.6319435, 0.35751677, 0.13305469, 0.35023752, 0.2134315, 0.7266131, 0.45943576, 0.4115885, 0.094741896, 0.42435962, 0.36004803, 0.3514786, 0.5841294, 0.71644586, 0.17080733, 0.014636853, 0.12820345, 0.8857452, 0.63410044, 0.6873833, 0.18846478, 0.22275887, 0.33786368, 0.40385348, 0.48239967, 0.0026390124, 0.90537494, 0.019583173, 0.14822645, 0.120368354, 0.5146042, 0.2832205, 0.78208506, 0.6385497, 0.12974623, 0.86614126, 0.32375425, 0.94300115, 0.29979256, 0.9403251, 0.35166037, 0.38923737, 0.046944916, 0.16529597, 0.66263044, 0.70765626, 0.9447244, 0.9664119, 0.82592314, 0.29692733, 0.55993456, 0.88830036, 0.7149717, 0.13324727, 0.24039853, 0.5394017, 0.3482095, 0.6519117, 0.9961026, 0.09185784, 0.7966022, 0.37718624, 0.37350395, 0.07291514, 0.5765629, 0.9380147, 0.8548175, 0.16825835, 0.047109794, 0.02843495, 0.58276457, 0.8786268, 0.6065395, 0.24458265, 0.3322088, 0.53052264, 0.26629698, 0.30301484, 0.7512133, 0.7805258, 0.47281718, 0.2468269, 0.27081817, 0.6599332, 0.20857278, 0.586381, 0.57643646, 0.75740004, 0.8954807, 0.5434921, 0.18676259, 0.50763893, 0.4595938, 0.90946364, 0.036301173, 0.8530404, 0.80743694, 0.16855665, 0.18533601, 0.8363561, 0.2781933, 0.5601252, 0.62934434, 0.73877335, 0.9165499, 0.96782345, 0.07416456, 0.15385732, 0.4842282, 0.011274599, 0.0977744, 0.345789, 0.38579237], 'my_varchar': 'varchar'}"]

int8_vector (vector is not included)

>>> client.query(collection_name="customized_setup_1", filter="my_id>=0")
data: ["{'my_id': 0, 'my_varchar': 'varchar'}"]

[cydrain]

@binbinlv
can see vector data now

data: ["{'my_vector': [b'c\\xcd\\x07\\x05\\xae\\x91\\xb6\\x9f4Q\\xb1<S\\xd9\\x11\\xed}\\x99\\xcc\\xc3\\xd3\\x12+\\xb5.\\xd2$*\\xb3\\x05\\xea\\xc6\\x91\\x18\\x18F\\x84\\xe8X*(%\\x83azr\\xc8\\x1ai\\xd1{\\x81\\x8c\\xd0\\xa4\\nA#\\t\\x00\\xba\\x05\\xa9H\\xac\\xa3A?\\xa0\\x9a\\x93hX\\x10\\xbbknX\\xbdOZ\\x04\\xa7`\\x8f\\x83\\x12D\\xef\\xc4;\\xee\\xd0\\x80o\\xbd\\x17\\\\\\x91\\xc1\\x08>\\xb84\\x93\\x14dG\\xbb\\x92,\\x85\\xa7\\x13\\xc3.\\x17\\x0c\\xff\\xf0\\xc8\\xe6\\xa3\\xc4q\\xba\\xaf\\xc3'], 'my_id': 0}"]

[binbinlv]

Ok, will verify it when the fix pr is merged.

[cydrain]

@binbinlv fix merged

[binbinlv]

Verified and fixed:

pymilvus: 2.6.0rc72
milvus: master-20250211-85c9f92f-amd64

results:

>>> client.query(collection_name="customized_setup_1", filter="my_id>=0", output_fields=["my_vector"])
data: ['{\'my_vector\': [b\'\\x9dc<\\x84\\xaa2\\x97:\\xfb\\xba\\xac\\xe8\\x87\\xa2\\xee\\n\\xd2^\\x8d\\x07\\xfb\\xe3\\x1d\\x14\\x87\\x97.db\\xfb\\xe7\\x04\\x05\\xed>\\xc9\\xee\\xd3bn\\xa7\\x80/\\xa3\\xf6\\x87Cg\\x04\\xc6\\x07\\xf1\\xec+\\xa9B%|\\xc2\\x9d\\x16\\xbe\\x91\\xa1\\xe89|\\x01l5W\\x9c\\x03\\xea3\\xa5="\\r\\xb0\\x83\\x91\\xdeo\\xe9\\x8e\\xcf\\x8a\\x84 1\\x94\\xaeJ\\xc1\\xcb1x\\xd2\\x19\\x96"~r\\xfe5\\xedbe\\x837J\\xd0\\x04\\x82\\xe4\\xc6d\\xd6\\x98&\\xd8\\xfdG\\xa1\\xd7O\\x95\'], \'my_id\': 0}', '{\'my_vector\': [b\'_I &\\x1b\\xc7^\\x8b\\xb6\\x9d\\x95\\xb6\\xc0\\xeb\\x85\\xc1\\xa7\\xd4O\\xcc|^x@y~\\xc30\\x92q"\\xb6\\x14\\xa8Z\\x00q\\xea\\x8b\\x88"\\x8ec\\xb7gq\\xd8\\xdf\\xf3\\x0e\\xc4\\x01\\xc4\\x9eR\\xdb5SW\\xa7()Ffz4\\x80\\xf7h>\\xd9\\xcd<\\x94\\x83\\xa3\\xb7\\x16*\\x86`d9\\xd0\\x88#<\\xac\\n\\xef\\xb8\\x9d\\x00"\\xc0\\x148\\xd5\\x87`\\x93\\xa9}\\x0c\\xce\\xaf\\xd7p\\xd9XbA\\x8e\\x06\\xef\\xa6\\xedo\\x97\\xd8\\xec\\xe9C\\x8d=N\\xc9r\'], \'my_id\': 1}', "{'my_vector': [b')`6\\xbf3\\xc9\\x80\\x94\\x89\\x88;\\xb0*\\x9b\\xeax\\x9f\\xcd\\x81\\x0f\\xb1\\xe6\\xb3s\\xe2\\xdbV\\xf9\\x91\\xd9\\x88\\x93\\xd2\\x19\\xa4\\xb0J\\xf7#W\\x8f\\x97\\x83\\xaa#\\xe6@\\x12a[\\xdd\\xdc\\xac>\\x02\\x88\\x02\\xc0\\xdd\\xcb\\xe9vb\\xdaX\\x8a\\xa8\\x8e\\xd5|U\\xfd\\x1aTU\\x07\\xb3T\\xed\\xc1\\x17\\x0b[\\x1dy8}I\\x89\\xe7h2\\x1e\\x16N3\\xff\\xa9#\\x14\\xa3\\x0f\\x83\\xbf\\x93\\x95a\\x8cj\\xe0\\xf5.\\xb6F6/\\x91\\x0e\\xa9\\xdeT\\xb3\\xedf|*Z\\x9e'], 'my_id': 2}", '{\'my_vector\': [b\'\\xb2\\xf9\\xf5/|;\\xf5%w\\x8b\\xee\\x0e\\xc9\\xfaC\\xeb\\x0e\\x88\\x96^\\x08Z\\x99\\x92\\xfdw\\x17f\\xc97\\x10\\x99&?\\x0c\\x03\\x83\\x84\\x8dOM\\xc0\\xde\\x8ct\\xdbA\\xcf*[V:\\x1c\\xbfW\\x0c]\\xa3\\x96Jy\\xd0\\xb8\\x98/\\xda\\xear\\x1c\\x0f\\xc4\\x07:w\\xe4m\\x07\\xab\\xf56J\\xda\\x10\\xcf~\\xda7(\\n\\x11/}\\xee\\xb7K\\xfe\\xccJb\\xf2\\x0f\\xde|\\xa0\\xd5\\xfb\\x972m\\xfc\\x87u\\n\\xc2\\xdf\\xd9\\x1c\\xf5\\xa4m-\\xf8\\xd6"X\\x10\\x0e\\xf4\'], \'my_id\': 3}', "{'my_vector': [b',\\xbe#\\xcf\\xce\\xe7\\xc1\\xf5\\x8b\\x07iD\\xba}\\xd8$\\xe7\\xe9p\\xb5\\xb0\\xde\\xd1\\xae\\x00\\xc1\\xf9\\x17\\xe3\\nm$\\x82X\\x18\\xd0M\\x85\\x10_\\xfa\\xd5\\x88m|\\xbdR\\xc4Q\\x8b\\x00\\x03\\xf2\\xf8[\\x9f$\\xec\\x10v\\xc0\\xa4\\xab\\xd2N/\\tu\\xb0\\x7fqx\\xc8\\xf5\\xad?%K\\xd7\\x85\\x85\\xe6b\\xa0dC\\xaa7_\\xc7&Z\\x10\\xe1\\x15\\xc5ZCc\\x82\\xfe\\xefw\\x0e\\xb6\\xe4\\x9ab\\xcf\\x8d\\xd2\\xf5K\\xa1\\xad\\xecu\\x1d\\xb08\\x00\\x8c\\x0fz\\xc0\\xf7_\\x98'], 'my_id': 4}", '{\'my_vector\': [b"\\xbb\\x08\\xc0}%\\x03\\xea\\x84\\xf3\\xcc\\x8a\\x01\\xa6\\x99\\xe0\\x8bUqTt\'\\x11\\x90\\xa1=P&\\t\\x8c\\x0c\\xe7K\\xde\\x86\\tzl\\xe1\\xfdu\\x0f}\\xd4\\xa4\\xb5\\x80-Z\\xac$X\\x07\\x0f\\x9c\\xdd>@\\x88\\x9a\\xeeM\\xb3w=\\xec\\x04C\\xcb\\xa6j\\xc6\\x89D4S\\xbe\\x88\\x06 \\xc1\\x7f\\xa7WP\\xb2e/\\x8d\\xb3)\\x95X\\xcfV\\nOO&U\\xaf9\\xee\\x1c+\\xae\\x88\\xa0\\x10\\x15\\xee\\xce\\xb2\\x82v\\xf5\\xc1\\xbd\\x0e\\xff\\xffz/V=uy\\xa4\\x02"], \'my_id\': 5}', '{\'my_vector\': [b\'\\xfe\\x9e\\xacM\\xba\\xdc\\x00o\\xfc6\\xc6\\x98\\xd3\\xc9r>\\xac+#\\xda\\xe0}(B\\xde<\\xb2\\xc57\\xc4\\x84\\xf2)0\\x83\\x10~\\x06\\r\\xec\\x93\\x18j\\xfb\\xca\\x00R\\x87"\\xb3;^\\x06l\\xd5Q\\xe4\\x9f)\\x1d\\xa4Q:V\\x1a\\x18N\\xff\\x8c\\xc8\\xd9\\xa1\\x04\\x929\\x08\\xc2\\x07\\xc8A\\n\\x98\\xb1oKi\\x19\\xf0RK\\xa8$j\\xca?2\\x9c\\xcd\\x11E\\xa3\\xc1A\\x8e\\x8c\\x1f\\xddAat0\\x1e\\x9e\\xd4\\n\\x020\\x01\\x05\\x91\\x9a!=\\t\\xaf\\x985\\xfa\'], \'my_id\': 6}', "{'my_vector': [b'n22\\xc3\\x8eX\\xd9\\x80\\xaaa-\\xcd\\xfc}3\\x18u#Q\\x1c\\xcd\\xc5|}\\xa2a\\xbe\\x8b\\x05\\x98|\\xdc\\t\\xf4y|\\x82!\\x87\\x8e\\xe7\\x80\\xdf\\xb5\\xe3\\x9d\\xe7\\xce\\x87\\xcc#\\xe8\\xb1I\\xac\\xd0\\xdcl\\xac\\x8f\\xa1\\xdfs\\x14\\xc9\\xff\\x1b$\\t\\x82aS\\x08i\\x85d\\x9c\\x9a\\x0e2\\xc85F\\xc7\\x97\\x1eC\\\\\\xdc\\n\\xfa\\xaf\\xf6\\xcb,\\x15\\xf8\\x9ddw\\x90@\\x93o\\xc9\\xdfY=\\x16\\xb8#\\xd0\\xe1%,\\x03\\xff\\xe83\\xe2\\xf1\\x1e\\xf2U\\xabifU'], 'my_id': 7}", '{\'my_vector\': [b"{6{\\xb1X2L\\xf8@o\\xe0\\xa1\\xb8>\\xef;PM\\n\\x1c\\xd6\\x1d[\\xda\\xc5K]\\x84$X\\xf4D\\x0bJL\\xe4\\x9bc\\xf7q\\x1cD\\x147\\x1bg\\xa4T\\x13\\xca\\x08\\x15S\\xe0\\x9f\\x01O\\x8d\\xb61B\\x18Z\\xdc\\x88\\xd3d\\x8d\\xc9\\xb6\\xeb:\\x04\\xd1\\xfe\\xa7\\xff\\xb5\\x97~\'\\x90\\xa9\\xd2\\xb6\\x9b\\x9d\\xa0I}\\x9a\\xb6\\x93q\\x89\\xc3\\xd3\\xd0\\xc1\\xac\\xff\\xf3\\x92\\xe0G\\xcb\\xd4d%\\x87\\x14\\xe8\\x0cI\\x9agO\\x97\\xdf\\x94/\\x02u\\xaf\\xb3p@!"], \'my_id\': 8}', '{\'my_vector\': [b\'w\\x07n\\\\"\\xfcY\\xf2jQ{E\\x0b\\\\\\xaey@\\x1d\\xba\\xc7\\\\d\\x82q\\x96\\x97\\xe5Y\\xdd\\xc3\\xda\\xdb\\xf5\\x07SS{N\\x1b\\xdb\\x8a?\\x01\\x86\\xa0g\\x9e\\xd6\\xf1\\x9c#h\\x83\\xcb\\xc1\\n\\xea\\x9aas\\x13\\\\\\xe3)A\\x16?\\xde*\\x07\\x17X\\xa4S,\\xeb8]\\xa0\\xbf\\r\\x83\\xecE1W[k\\xff\\x96\\x1f;\\xfe\\xe8\\xae\\x1b\\x8b\\xa4`\\xf4f-\\xb1\\x192\\x9d\\x89Y5\\xa4\\xd6\\xb7\\xaa$\\x8f\\r\\xb5\\xfc\\x12T\\x047\\xcf\\xff\\xd6\\x9f\\xc9+\'], \'my_id\': 9}'] ...
>>>
>>>
>>> client.query(collection_name="customized_setup_1", filter="my_id>=0", limit=1)
data: ['{\'my_vector\': [b\'\\x9dc<\\x84\\xaa2\\x97:\\xfb\\xba\\xac\\xe8\\x87\\xa2\\xee\\n\\xd2^\\x8d\\x07\\xfb\\xe3\\x1d\\x14\\x87\\x97.db\\xfb\\xe7\\x04\\x05\\xed>\\xc9\\xee\\xd3bn\\xa7\\x80/\\xa3\\xf6\\x87Cg\\x04\\xc6\\x07\\xf1\\xec+\\xa9B%|\\xc2\\x9d\\x16\\xbe\\x91\\xa1\\xe89|\\x01l5W\\x9c\\x03\\xea3\\xa5="\\r\\xb0\\x83\\x91\\xdeo\\xe9\\x8e\\xcf\\x8a\\x84 1\\x94\\xaeJ\\xc1\\xcb1x\\xd2\\x19\\x96"~r\\xfe5\\xedbe\\x837J\\xd0\\x04\\x82\\xe4\\xc6d\\xd6\\x98&\\xd8\\xfdG\\xa1\\xd7O\\x95\'], \'my_varchar\': \'varchar\', \'my_id\': 0}']