Pawan Kumar Rajpoot, Senior Data Scientist at DataX, received the top prize from the prestigious NeurIPS Large Language Model Efficiency Challenge: 1 LLM + 1GPU + 1Day, held by AI at Meta and Microsoft Research in New Orleans, USA, in December 2023. This event attracted 75 teams comprising 187 participants globally, who worked rigorously over a 4-month competition period.

The NeurIPS Large Language Model Efficiency Challenge focused on enabling access to Large Language Models (LLMs) so that everyone could equally access and utilize these models. This challenge is a significant event in the AI, Machine Learning, and Deep Learning sphere, renowned as a prime platform for experts and enthusiasts in AI to gather annually and discuss the latest advancements and research in the field.

Mr. Pawan Kumar Rajpoot, Senior Data Scientist at DataX, was a member of the winning team 'Upaya,' which triumphed in this competition. This is a source of pride for the company, showcasing a highly capable AI team on a global level. The challenge was to fine-tune a base open-source LLM on a single GPU (RTX 4090 or A100 40GB) for 24 hours. The fine-tuned models were evaluated on a subset of the HELM benchmark along with holdout tasks prepared by organizers. The winning model ????????????????????????-????????-???????? is based on 16-hours fine-tuning of the Mistral AI Mistral-7B model using a single RTX 4090 on the curated dataset.

Additionally, representatives from DataX, namely Mr.Nut Chukamphaeng and Dr.Pakhapoom Sarapat also received recognition, securing the 4th position in the competition. This further emphasizes that DataX is an organization filled with knowledgeable experts in AI and Machine Learning.