In the talk, Mr. Sahil Verma and Mr. Eby Kuriakose embarked on a comprehensive discussion abot working of an Large Language Models (LLMs), cutting-edge advancements in LLMs, Agenting AI, Tasks needed to be resolved/acheived for an Multi Agentic environment and career landscape in GenAI , NLP and Artificial Intelligence domain. They taught the students what goes in the background of Large Language Models like ChatGPT, Grok etc. The session started with an overview of the fundamentals of LLMs and then transitioned to the latest innovations in LLMs, behind the scenes of ChatGPT, Agentic AI and Multi Agentic AI.
They also gave the perspective of what kind of challenges they are dealing with in AWS to make Multi Agentic AI a reality.
They discussed the Carrier Impact GenAI and NLP can create. Also discussed open domains in GenAI and NLP which the students can solve and become a successful AI engineer/ researcher.
The event began with a short presentation by Dr. Richa Tengshe (Head CoE NLP) about the history, current trends and the future direction of NLP. After the presentation six student teams presented their posters and explained the idea behind their creation.
Innovative talk was conducted by CoE NLP in association with Department of AIML CMRIT. It was held on 13th Feb 2025 from 2 pm to 4 p.m. in AV HALL, the fourth floor of the D Block, CMRIT. The session started with an introduction session about the introductory workflow in NLP and need of error correction in Kannada text by Prof. Sushmitha R AP, AIML CMRIT. Her research areas included Kannada Textual error correction using T5. This was a part of her MTech thesis.At the end of session an open discussion was conducted for developing and improving competency of students and faculty of CMRIT in domain. A total of 23 including students and faculties from CMRIT attended the event.
The presentation began by providing an overview of LLMs and their capabilities in natural language understanding and generation. It then delved into the concept of fine-tuning, wherein pre-trained LLMs are adapted to domain-specific tasks by further training on task-specific data. In the context of healthcare, fine-tuning enables LLMs to understand medical terminology, comprehend clinical notes, and generate contextually relevant responses.
In his talk, Dr. Rajath embarked on a comprehensive journey through the evolution of Natural Language Processing (NLP), from foundational embedding techniques to the cutting-edge advancements in Large Language Models (LLMs). The session started with an overview of the fundamental concepts of NLP, exploring the development of embeddings such as Word2Vec, GloVe, and transformer-based embeddings like BERT. The talk then transitioned to the latest innovations in LLMs, focusing on the Mistral model, a state-of-the-art language model designed for complex NLP tasks.
A live demonstration of the the practical application of these concepts using Streamlit, showcasing the use of Mistral LLM for question answering in the domain of material science is done. The demo highlighted how Mistral can be integrated with Retrieval-Augmented Generation (RAG) techniques to enhance document matching. The talk also demonstrated how to retrieve relevant context from a corpus of saved documents and effectively use this context to craft prompts for the Mistral model, thereby improving the accuracy and relevance of the generated answers.
The session started with an introduction session about the objective and requirement of the Gen AI by Dr. Chandrika Senior Scientist at ABB Bengaluru. Exploring the potential of generative AI for prototyping and productivity environments can be both transformational and exciting. This Innovative Session unlocked new levels of efficiency, creativity, and innovation. Provides insights into the latest patent trends and applications on generative AI algorithms , AI education on tracking student progress and AI in Autonomous Vehicles etc. Important articles in the field were referred to in order to give an understanding of the research being done in generative AI. How to use generative AI for particular tasks by sharing domain-specific system designs.