Skip to content

Release notes

0.3.0

- initial version of cache of llm classification performed during llm filtering step

- integrating ollama and openai llm handlers for llm filter feature

- upgrading older version of attrs to attrsx

0.2.7

- minor fixes for initial llm filter

- separate async search_database_async method

0.2.6

- cosine similarity with torch

- adding missing inputs for remote mocker in search_database method

0.2.5

- simple cli for initializing optional config

- wiring mocker connector into main mocker class

- a single interface to interact with local and remote mocker

- packaging existing mocker-db api routes with latest search options

0.2.4

- making hnswlib optional dependency

0.2.3

- providing flag to disable embedder initialization

- disabling old cli intefrace

- removing sentence_transformers from the list of default requirements so that SentenceTransformer needs to be provided externally if needed

0.2.2

- option to get embedded field

- initial llm filter

0.2.1

- ability to return unique hash key for each record, previously inaccesible

0.2.0

- precise keywords match with cutoff 1 and fuzzy match with < 1 through filters

- keywords search with difflib

0.1.3

- option to add embedding and distance to output list

- option to remove output key while outputting the rest of keys

0.1.2

- initital cli interface that allows to clone code from api version of mocker and run it

0.1.1

- initial MockerConnect for using MockerDB API

0.0.12

-  bugfix for similarity search through partly embedded data

0.0.11

- more advanced filtering

0.0.10

- fix for search without embeddings

0.0.6

- fix for embedding storage

0.0.5

- initial implementation of separate caching store for embeddings

0.0.4

- updating hnswlib 0.7.0 -> 0.8.0 to fix vulnerabilities issue

- fixing a bug with resetting mocker inner state properly after search

0.0.3

- slightly improving logic of embedding with batches in parallel for sentence transformer embedder (default embedder)

- updating desciption

0.0.2

- better error handling in situations when data was not found with applied filters

0.0.1

- initial version of MockerDB package that evolved from mock classes from redis into a standalone solution