daming-w Goto Github PK

cs-notes

:books: 技术面试必备基础知识、Leetcode、计算机操作系统、计算机网络、系统设计

data-juicer

A one-stop data processing system to make data higher-quality, juicier, and more digestible for LLMs! 🍎 🍋 🌽 ➡️ ➡️🍸 🍹 🍷为大语言模型提供更高质量、更丰富、更易”消化“的数据！

datacomp

DataComp: In search of the next generation of multimodal datasets

data_scoring_model

A demo implement of Quality Comparison Model. This Model aims to evalute image-text precomputed embeddings and sample better data for feeding LLM/MLLM training.

deeplearning

深度学习入门教程, 优秀文章, Deep Learning Tutorial

gd-based-ensemble

A Demo for implementing gradient-descent-based voting ensemble method which is specific designed for data curation work.

hide-and-seek

This research presents probabilistic machine learning methods to optimize basic image ranking models. And make it suitable for visual navigation and exploration tasks in the real complex world.

image-retrieval-from-text

This research project has constructed a two-stage clustering-based retrieval framework, as well as a deep learning-based retrieval algorithm using the CLIP model, which demonstrates zero-shot ability in the multimodal domain.

lavis

LAVIS - A One-stop Library for Language-Vision Intelligence

llava_su

Scenario Understanding with Visual-Question-Answering Base on Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.

pcd_3d_visualization

A quick script for visualizing point-cloud pseudo labeling results.

recurrent-grad-cam

ui-img-txt-contrastive-learning

AI4SE UI Multimodal Framework This framework introduces multimodal pretraining for software UIs in AI4SE, enabling tasks like UI captioning, retrieval, and localization. It uniquely combines Grad-CAM with text-image models for precise text-driven UI localization.