milvus: [Bug]: Recall is only 0.384 for first search, which is expected to be 0.97~0.98, but for the late search , the recall can be 0.984
Is there an existing issue for this?
- I have searched the existing issues
Environment
- Milvus version:master-20230116-836773f1
- Deployment mode(standalone or cluster):cluster and standalone
- MQ type(rocksmq, pulsar or kafka):Kafka and pulsar
- SDK version(e.g. pymilvus v2.0.0rc2):
- OS(Ubuntu or CentOS):
- CPU/Memory:
- GPU:
- Others:
Current Behavior
[2023-01-17T10:11:37.991Z] Search...
[2023-01-17T10:11:37.991Z] search cost 3.9131 seconds
[2023-01-17T10:11:37.991Z] Traceback (most recent call last):
[2023-01-17T10:11:37.991Z] File "scripts/first_recall_test.py", line 136, in <module>
[2023-01-17T10:11:37.991Z] milvus_recall_test(host)
[2023-01-17T10:11:37.991Z] File "scripts/first_recall_test.py", line 117, in milvus_recall_test
[2023-01-17T10:11:37.991Z] assert recall >= 0.95
[2023-01-17T10:11:37.991Z] AssertionError
collection name: sift_128_euclidean
Expected Behavior
[2023-01-15T09:28:45.814Z] Insert 1000000 vectors cost 32.9921 seconds
[2023-01-15T09:28:45.814Z]
[2023-01-15T09:28:45.814Z] Get collection entities...
[2023-01-15T09:28:45.814Z] 1000000
[2023-01-15T09:28:45.814Z]
[2023-01-15T09:28:45.814Z] Get collection entities cost 2.5327 seconds
[2023-01-15T09:28:45.814Z]
[2023-01-15T09:28:45.814Z] Create index...
[2023-01-15T09:28:45.814Z]
[2023-01-15T09:28:45.814Z] Create index cost 3.5314 seconds
[2023-01-15T09:28:45.814Z]
[2023-01-15T09:28:45.814Z] load collection...
[2023-01-15T09:28:45.814Z]
[2023-01-15T09:28:45.814Z] load collection cost 6.5466 seconds
[2023-01-15T09:28:45.814Z]
[2023-01-15T09:28:45.814Z] Search...
[2023-01-15T09:28:45.814Z] search cost 4.1557 seconds
[2023-01-15T09:28:45.814Z] recall=0.983
[2023-01-15T09:28:45.814Z] {'float': 2.0, 'int64': 2}
[2023-01-15T09:28:45.814Z] {'float': 4.0, 'int64': 4}
[2023-01-15T09:28:45.814Z] {'float': 6.0, 'int64': 6}
[2023-01-15T09:28:45.814Z] {'float': 8.0, 'int64': 8}
Steps To Reproduce
No response
Milvus Log
failed job: https://qa-jenkins.milvus.io/blue/organizations/jenkins/deploy_test_cron/detail/deploy_test_cron/225/pipeline log: artifacts-pulsar-cluster-reinstall-225-server-logs.tar.gz artifacts-pulsar-cluster-reinstall-225-pytest-logs.tar.gz
Anything else?
No response
About this issue
- Original URL
- State: closed
- Created a year ago
- Comments: 16 (16 by maintainers)
/assign @xige-16 pls take a look into it