Collection returns empty in question answering engine.

[lydacious]

Hello,
I am following the guide on creating a question-answering engine. I have done everything the same as the guide, but the collection stays empty for some reason. Before pasting the code here, this is the list of what I've done so far:

• Printing row - prints them perfectly,
• Added counter for every insert_pipe(*row) - count also worked perfectly,
• Print(collection.is_empty) - returns True,
• print(collection.name) - returns question_answer
• Checked the port - can connect to it perfectly.
• Output: True. Total number of inserted data is 0.
  P.S: I am using VS Code. Have tried Jupyter Notebook as well. Same results.

Code:

import pandas as pd
import numpy as np
from pymilvus import connections, FieldSchema, CollectionSchema, DataType, Collection, utility
from towhee import pipe, ops
from towhee.datacollection import DataCollection



# Read data from CSV
df = pd.read_csv('question_answer.csv')
id_answer = df.set_index('id')['answer'].to_dict()

# Milvus connection
connections.connect(host='127.0.0.1', port='19530')

# Milvus collection creation function
def create_milvus_collection(collection_name, dim):
    if utility.has_collection(collection_name):
        utility.drop_collection(collection_name)
    
    fields = [
        FieldSchema(name='id', dtype=DataType.VARCHAR, description='ids', max_length=500, is_primary=True, auto_id=False),
        FieldSchema(name='embedding', dtype=DataType.FLOAT_VECTOR, description='embedding vectors', dim=dim)
    ]
    schema = CollectionSchema(fields=fields, description='reverse image search')
    collection = Collection(name=collection_name, schema=schema)

    # Create IVF_FLAT index for collection
    index_params = {
        'metric_type': 'L2',
        'index_type': 'IVF_FLAT',
        'params': {'nlist': 2048}
    }
    collection.create_index(field_name='embedding', index_params=index_params)
    return collection

# Create Milvus collection
collection = create_milvus_collection('question_answer', 768)

insert_pipe = (
    pipe.input('id', 'question', 'answer')
        .map('question', 'vec', ops.text_embedding.dpr(model_name='facebook/dpr-ctx_encoder-single-nq-base'))
        .map('vec', 'vec', lambda x: x / np.linalg.norm(x, axis=0))
        .map(('id', 'vec'), 'insert_status', ops.ann_insert.milvus_client(host='127.0.0.1', port='19530', collection_name='question_answer'))
        .output()
)

import csv
with open('question_answer.csv', encoding='utf-8') as f:
    reader = csv.reader(f)
    next(reader)
    for row in reader:
        insert_pipe(*row)


print(collection.is_empty)

print('Total number of inserted data is {}.'.format(collection.num_entities))```

[Vokinloksar]

same here, suspect related to version issue of certain package.

[junjiejiangjjj]

You need to call the collection.load interface after insert. https://milvus.io/docs/load_collection.md

[AneetaNiazi]

instead of inserting each row separately, I inserted all rows as a batch and this error was resolved:

%%time
from towhee import pipe, ops
import numpy as np
from towhee.datacollection import DataCollection

insert_pipe = (
 pipe.input('id', 'question', 'answer')
     .map('question', 'vec', ops.text_embedding.dpr(model_name='facebook/dpr-ctx_encoder-single-nq-base'))
     .map('vec', 'vec', lambda x: x / np.linalg.norm(x, axis=0))
     .map(('id', 'vec'), 'insert_status', ops.ann_insert.milvus_client(host='127.0.0.1', port='19530', collection_name='question_answer'))
     .output()
)

import csv
with open('question_answer.csv', encoding='utf-8') as f:
 reader = csv.reader(f)
 next(reader)
 allRows=[]
 for row in reader:
     allRows.append(row)
 res=insert_pipe.batch(allRows)