Skip to main content

21 docs tagged with "Foundation-AI"

View all tags

Forte Overview

The Forte library provides robust out-of-distribution (OOD) detection capabilities through the ForteOODDetector class. The core algorithm is built on the principle of Finding Outliers using Representation Typicality Estimation, which:

How-To Guides

These are required setup steps to use PEFT methods in our codebase.

PEFT Vision

Lessons and Insights from a Unifying Study of Parameter-Efficient Fine-Tuning (PEFT) in Visual Recognition

Region2vec

Region2vec: Community Detection on Spatial Networks Using Graph Embedding with Node Attributes and Spatial Interactions

Science Agent

Language agents for data-driven scientific discovery tasks. The agent features the self-debug mechanism and shows the best performance on ScienceAgentBench.

ScienceAgent Interface

ScienceAgent Interface provides a web interface for conducting data-driven scientific tasks using ScienceAgent. The interface connects to a Python backend which allows users to execute generated programs in an isolated Docker environment and view the results.

Species Classification using Multimodal Heterogeneous Context

We present a species classification model that utilizes heterogeneous image contexts organized in a multimodal knowledge graph. The multimodal knowledge graph may include diverse forms of heterogeneous contexts that pertain to different modalities, such as numerical information for locations and time, categorical data for species/taxon IDs, and visual content such as images.