Research Papers by GenAI360 Team

In order to keep abreast of the ever changing world of Gen AI, team at eClerx conducts and publishes research papers in international conferences on regular basis. Our latest research in the area of vector databases and LLMs is mentioned below:

  1. Vector databases and vector embedding - REVIEW

    A comprehensive survey of vector databases and vector embedding techniques was conducted. A concise overview of the evolution, architecture, advantages and challenges of vector databases were presented in this paper. Conversion of unstructured data into vectors, various embedding techniques with their in-depth technical description were also surveyed in this research paper. The existing vector databases' characteristics and features are described to select an appropriate vector database.

    This research help us bolster our objective towards building enterprise grade RAG applications.

    The paper was published and presented in IEEE approved International Workshop on Artificial Intelligence and Image Processing (IWAIP 2023) Indonesia.

    Know More

  2. Literature survey on large language models

    As the usage of AI algorithms changes the way businesses, scientific community and industry works, it is imperative to review the advances of LLMs. This paper aimed to conduct a literature survey on open source large language models with focus on major aspects of open source LLMs - pre-training covering data collection and pre-processing, model architecture and training. The paper also covered the importance and impact of open source large language models in shaping various businesses. Further,

    this research empowers us to select the best and most apt LLMs for our customers as per their growing business needs, considering scale and data privacy.

    The paper is presented in 7th International Conference on Computers in Management and Business (ICCMB 2024), Singapore and shall be published in ACM journal March 2024.

  3. Performance evaluation of vector embeddings with retrieval-augmented generation

    The Retrieval Augmented Generation(RAG) plays a crucial role in NLP now a days. This research is intended to evaluate the performance of vector embedding techniques with RAG. investigations focus on the impact of alternative embedding approaches on the overall performance of context generation across the NCERT books dataset using a systematic evaluation.

    The capabilities of ChatGPT and Llama2 are employed, for evaluating the performance of embedding models. The variety in vector embedding approaches is exhibited by the experimental results.

    The paper is submitted to reputed IEEE conferences. The first level of review is accepted.