tql-ANN 2021.2.5.19.19.33

description

:rocket: ANN :facepunch:

---

Install

pip install fast-ann

Usage

from ann import ANN
import numpy as np

data = np.random.random((1000, 128)).astype('float32')

ann = ANN()
ann.train(data, index_factory='IVF4000, Flat', noramlize=True)

dis, idx = ann.search(data[:10])

print(dis)
print(idx)

---

milvus镜像

yum -y install python3
# ln -sf /usr/bin/python3 /usr/bin/python
# ln -sf /usr/bin/pip3 /usr/bin/pip
pip3 install -U --no-cache-dir -i https://mirror.baidu.com/pypi/simple pip meutils pymilvus

rm -rf /tmp/*
rm -rf /root/.cache/pip*

---

• faiss不同量级对应的训练时间及内存测试

• 压缩方式测试

• 四种组合：默认是查向量返回 distance与index

  • id => id/vector
  • vector => id/vector
  • push场景需要 docid => title_vector => docid

• 线上服务

  • id2word
  • id2vector