{"id":995493,"date":"2024-01-05T08:01:42","date_gmt":"2024-01-05T16:01:42","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-project&p=995493"},"modified":"2024-10-23T12:11:56","modified_gmt":"2024-10-23T19:11:56","slug":"afmr-model-advancement","status":"publish","type":"msr-project","link":"https:\/\/www.microsoft.com\/en-us\/research\/project\/afmr-model-advancement\/","title":{"rendered":"AFMR: Model Advancement"},"content":{"rendered":"
\n\t
\n\t\t
\n\t\t\t\"white\t\t<\/div>\n\t\t\n\t\t
\n\t\t\t\n\t\t\t
\n\t\t\t\t\n\t\t\t\t
\n\t\t\t\t\t\n\t\t\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\t<\/span>\n\t\t\t\t\t\t\t\t\tAccelerating Foundation Models Research\t\t\t\t\t\t\t\t<\/a>\n\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\n\n

Model Advancement<\/h1>\n\n\n\n

<\/p>\n\n\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t<\/div>\n\t\t<\/div>\n\t<\/div>\n<\/section>\n\n\n\n\n\n

\n

Academic research plays such an important role in advancing science, technology, culture, and society. This grant program helps ensure this community has access to the latest and leading AI models.<\/em><\/strong><\/p>\nBrad Smith, Vice Chair and President<\/cite><\/blockquote>\n\n\n\n

\n
<\/div>\n\n\n\n
\n
\"green<\/figure>\n\n\n\n

AFMR Goal: Align AI with shared human goals, values, and preferences via research on models<\/h2>\n\n\n\n

which enhances safety, robustness, sustainability, responsibility, and transparency, while ensuring rapid progress can be measured via new evaluation methods<\/p>\n<\/div>\n\n\n\n

<\/div>\n<\/div>\n\n\n\n
\n\t\n\t
\n\t\t
\n\t\t\t
<\/div>\t\t<\/div>\n\t<\/div>\n\n\t<\/div>\n\n\n\n

A common theme among these research projects revolves around improving the LLM’s alignment with human goals, addressing challenges like hallucinations, unfaithful information generation, lack of control, and improving their robustness, interpretability, and generalizability. Several proposals also emphasize enhancement of specific reasoning capabilities, like logical, commonsense, syntactic, inductive, abductive reasoning, and multi-document reasoning. Other specific advancements include enabling LLMs to reason about time-series data, collaborate amongst themselves, simulate public responses to projected AI actions, interact with external environments, etc. In terms of techniques, reinforcement learning, human feedback, retrieval-based methods, fine-tuning, model compression, task-oriented dialogue, and sequence decision-making is being explored for improving LLM’s performance and utility.<\/p>\n\n\n\n

<\/div>\n\n\n\n\n\n

IIT Kharagpur<\/strong>: Pawan Goyal (PI)<\/p>\n\n\n\n

The research proposes a novel automated evaluation framework for natural language generation via large language models. The framework aims to overcome the limitations of existing evaluation metrics that fail to fully capture the nuances of LLM-generated content. It also seeks to mitigate biases such as positional, length, and self-enhancement biases that are often present in such models. The framework will undergo rigorous testing across diverse tasks, including summarization and mathematical reasoning. The research also aims to explore the performance of various LLMs and develop new metrics for evaluating their outputs.<\/p>\n\n\n\n\n\n

Harvard University<\/strong>: Stefano Maria Iacus (PI)<\/p>\n\n\n\n

The proposal aims to explore the capabilities of Foundation Models to facilitate the discovery of research and research data. The aim is to enrich metadata and build semantic knowledge graphs to accelerate knowledge acquisition. OpenAI models and fine-tuned open source models such as Llama-2, Falcon or MPT will be utilized.<\/p>\n\n\n\n\n\n

University of California, Riverside<\/strong>: Amr Magdy (PI)<\/p>\n\n\n\n

This project will investigate how to generate linguistic summaries in natural language from real-time millimeter-wave and sub-terahertz radar data. Beamforming and range\/speed tracking capabilities of radar systems will be leveraged to enable human and environmental context perception. Radar datasets will be pre-processed, and useful features will be utilized as input to large-scale language models. The ultimate objective of this project is to enable the fusion of wireless sensing with natural language, revealing concealed patterns within unstructured radar signatures. Radars have already been used as sensing modality for vital signs detection and human behavior recognition. However, existing research on radar-based vital signs detection and human activity monitoring focuses on traditional machine learning models tailored to the application use case, which may require considerable expertise from the end user and cannot adapt to the constant changes in realistic scenarios. To tackle these challenges, large language models will be utilized to assist radar signal analysis. In this study, commercially available radar platforms will be employed to obtain data associated with typical human activities. Algorithms for signal processing will be developed to combine radar data with Generative Pre-Trained Transformer models, and insights into the strengths and weaknesses of existing foundation models will be garnered.<\/p>\n\n\n\n\n\n

Carnegie Mellon University<\/strong>: Carlee Joe-Wong (PI)<\/p>\n\n\n\n

This proposal presents a novel approach for improving alignment of Foundation Language Models (LMs) to human goals. The focus is on creating an ensemble of LMs incorporating cost constraints, human feedback, and strategic utilization of different LMs. The team plans to employ online learning mechanisms, particularly reinforcement learning, to optimize this process. The approach will be validated using various datasets.<\/p>\n\n\n\n\n\n

Tokyo Institute of Technology<\/strong>: Naoaki Okazaki (PI)<\/p>\n\n\n\n

Naoaki Okazaki proposes research to examine whether large language models (LLMs) can benefit from a more dynamic, interactive, and bidirectional learning process that mirrors human cognition. This approach intends to improve how LLMs understand and generate language, addressing limitations observed in current LLMs’ performances in complex reasoning tasks. The research introduces a paradigm of using real-time, adaptive discussions between two LLMs for training, with one LLM serving as the ‘learner’ while the other as a ‘discussion partner’.<\/p>\n\n\n\n

Related papers:<\/strong><\/p>\n\n\n\n