Pinned Repositories
clowder
A data management system that allows users to share, annotate, organize and analyze large collections of datasets. It provides support for extensible metadata annotation using JSON-LD and a distribute analytics event bus for automatic curation of uploaded data.
ai-devs-only
ai-ta-frontend
Chat with your documents. Upload anything, get answers.
LLM-Distributed-Quantization
Accelerating multi-node Large Language Model training with per-layer selective quantization (FP32 -> FP16) of the transformer architecture.
llm-serving-frontend
Frontend GUI for LLM Serving at NCSA
pdg_ray_workflow
Ray PDG's Viz-staging, Viz-Raster and Viz-3D in parallel using Ray Core and Ray Workflows.
toolbox
Useful ML engineering libraries and scripts
video-pretrained-transformer
Multi-model video-to-text by combining embeddings from Flan-T5 + CLIP + Whisper + SceneGraph. The 'backbone LLM' is pre-trained from scratch on YouTube (YT-1B dataset).
FakeBananas
Fake news detection web app. Fake Bananas finished top 10 at HackMIT 2017 and won awards for 'best use of data' and 'best use of machine learning for the common good.' (Project is no longer maintained)
ai-teaching-assistant-uiuc
The UX & driver code for the multi-model TA_Chatbot
KastanDay's Repositories
KastanDay/video-pretrained-transformer
Multi-model video-to-text by combining embeddings from Flan-T5 + CLIP + Whisper + SceneGraph. The 'backbone LLM' is pre-trained from scratch on YouTube (YT-1B dataset).
KastanDay/ai-ta-frontend
Chat with your documents. Upload anything, get answers.
KastanDay/toolbox
Useful ML engineering libraries and scripts
KastanDay/LLM-Distributed-Quantization
Accelerating multi-node Large Language Model training with per-layer selective quantization (FP32 -> FP16) of the transformer architecture.
KastanDay/llm-serving-frontend
Frontend GUI for LLM Serving at NCSA
KastanDay/pdg_ray_workflow
Ray PDG's Viz-staging, Viz-Raster and Viz-3D in parallel using Ray Core and Ray Workflows.
KastanDay/ai-devs-only
KastanDay/MAPLE-NGC
Running MAPLE code in a Nvidia NGC Singularity/Apptainer container for HPC.
KastanDay/notes_enhancer
Mission statement: 1. Find "unlinked mentions" between pages via semantic search. As a page is written, the rest of your notes are 'fuzzy' searched for synonymous (and opposite) terms, people. 2. Use content generation to "finish your thought" and extend your notes to make them more rich and searchable. 3. Search the web (Wikipedia, Google knowledge graph) for topics & people mentioned in your notes. Append that content to the bottom of your notes to make notes richer, and easily searchable. This is a positive feedback loop that helps make more connections between notes, as well.
KastanDay/ML4Bio
LLMs to execute Bioinformatics workflows, esp. RNA-seq
KastanDay/ACERetro
Asynchronous chemoenzymatic retrosynthesis
KastanDay/changelog
A constant update of how I spend my time at work.
KastanDay/cheese
Used for adaptive human in the loop evaluation of language and embedding models.
KastanDay/ColossalAI-quantized
Colossal-AI: A Unified Deep Learning System for Big Model Era
KastanDay/dotfiles
KastanDay/elk-centralized-logging
My config files for monitoring docker containers across many different hosts.
KastanDay/faster_OpenPSG
Benchmarking Panoptic Scene Graph Generation (PSG), ECCV'22
KastanDay/fastest-filesystem-and-databse-for-multimodal-ML
With fast GPUs and networking, file servers become the bottleneck, especially for multi-modal ML where small image and audio files wrecks havoc on traditional filesystems like Lustre, exa6, GPFS, Ceph, and even Hadoop (HDFS).
KastanDay/idefics-9b-instruct-8bit
KastanDay/langchain-improved-agents
⚡ Building applications with LLMs through composability ⚡
KastanDay/leerob.io
✨ My portfolio built with Next.js, Tailwind, Prisma, and Vercel.
KastanDay/lhotse_faster_whisper
Tools for handling speech data in machine learning projects.
KastanDay/llm-serving
Serve LLMs on NCSA hardware
KastanDay/ML4Bio-v2
LLMs to execute Bioinformatics workflows, esp. RNA-seq
KastanDay/ngc_utilities
Nice starting points for using Nvidia NGC images on Ubunut (just install a Nvidia the right Nvidia utilities firstt)
KastanDay/react-pdf-highlighter
Set of React components for PDF annotation
KastanDay/store
The home for importable Guidebooks
KastanDay/transformers-VPT
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
KastanDay/UIUC-NLP-Course-Projects
NLP Projects I made for Professor Heng Ji's Advanced NLP course at UIUC (CS 546).
KastanDay/uiuc-ta
AI TA for UIUC courses