Skip to content

Framework for prototyping of LLM-based applications [under development]

License

Notifications You must be signed in to change notification settings

ITMO-NSS-team/ProtoLLM

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Acknowledgement to SAI Acknowledgement to ITMO

Intro

Proto-LLM is an open-source framework for fast protyping of LLM-based applications.

Proto LLM features (WIP)

  • Rapid prototyping of information retrieval systems based on BNM using RAG:

Implementations of architectural patterns for interacting with different databases and web service interfaces; Methods for optimising RAG pipelines to eliminate redundancy.

  • Development and integration of applications with BNM with connection of external services and models through plugin system:

Integration with AutoML solutions for predictive tasks; Providing structured output generation and validation;

  • Implementation of ensemble methods and multi-agent approaches to improve the efficiency of BNMs:

Possibility of combining arbitrary BNMs into ensembles to improve generation quality, automatic selection of ensemble composition; Work with model-agents and ensemble pipelines;

  • Generation of complex synthetic data for further training and improvement of BNM:

Generating examples from existing models and data sets; Evolutionary optimisation to increase the diversity of examples; Integration with Label Studio;

  • Providing interoperability with various LLM providers:

Support for native models (GigaChat, YandexGPT, vsegpt, etc.). Interaction with open-source models deployed locally.

Analogs

LLMware, LLMStach, LangChain

However, they are not direct competitors to the framework being created, as it is a higher-level tool that uses existing LLMOps solutions where possible and necessary, and provides compatibility with them for most tasks.

Project Structure

The latest stable release of ProtoLLM is in the master branch.

The repository includes the following directories:

  • Package core contains the main modules. It is the core of the ProtoLLM framework
  • Package examples includes several how-to-use-cases where you can start to discover how ProtoLLM works
  • All unit and integration tests can be observed in the test directory
  • The sources of the documentation are in the docs directory

Contribution Guide

  • The contribution guide is available in this repository.

Acknowledgments

We acknowledge the contributors for their important impact and the participants of the numerous scientific conferences and workshops for their valuable advice and suggestions.

Supported by

The study is supported by the Research Center Strong Artificial Intelligence in Industry of ITMO University as part of the plan of the center's program.

Contacts

About

Framework for prototyping of LLM-based applications [under development]

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Contributors 4

  •  
  •  
  •  
  •  

Languages