TruEra, a leading provider of tools to test, debug, and monitor machine language (ML) models, has launched TruLens – an open-source software that helps enterprises test their applications built on large language models (LLMs) like the GPT series. The software is free to use and it eliminates the chances of hallucination and bias in the production stage.
Developing applications based on LLMs can be a tiring process, which requires developers to manually test and review the app’s answers, adjust prompts, hyperparameters and models repeatedly until it yields the desired result. TruLens simplifies this experience by introducing a programmatic method of evaluation known as “feedback functions”. By analyzing the generated text from the LLM and metadata, the software provides enterprises with the ability to score the output of their app and evaluate its quality and efficacy. . TruLens can additionally be used to log pings to an LLM app, enabling users to monitor their usage costs.
TruEra was founded by Anupam Datta, who currently also serves as the President and Chief Scientist. With the launch of TruLens, the company has solved the commonly faced gap by entrepreneurs in developing applications using LLM, helping them create reliable and efficient applications much faster.
As the need for AI applications continues to increase, TruLens is positioned to be the go-to solution for many enterprises for testing and monitoring ML models, by offering a seamless and efficient experience. The tool will be highly beneficial to businesses as they increasingly rely on and utilize ML models for their operations.
