|
|
--- |
|
|
emoji: 🔥 |
|
|
colorFrom: gray |
|
|
colorTo: indigo |
|
|
sdk: static |
|
|
pinned: false |
|
|
title: "DSTI — AI Community" |
|
|
type: "organization" |
|
|
tags: |
|
|
- community |
|
|
- education |
|
|
- machine-learning |
|
|
- deep-learning |
|
|
- nlp |
|
|
- computer-vision |
|
|
- generative-ai |
|
|
- open-source |
|
|
- teaching |
|
|
- research |
|
|
library_name: |
|
|
- transformers |
|
|
- diffusers |
|
|
- datasets |
|
|
- peft |
|
|
--- |
|
|
|
|
|
# DSTI — AI Community |
|
|
|
|
|
**DSTI** is a university community focused on teaching, researching, and building with **AI**. |
|
|
We share open-source models, datasets, and learning resources across **NLP**, **Computer Vision**, and **Generative AI**, with a strong emphasis on **reproducibility** and **knowledge sharing**. |
|
|
|
|
|
> **Interests:** Deep Learning (NLP & CV), Open Source, Reproducible Research, Knowledge Sharing, MLOps, Responsible AI. |
|
|
|
|
|
--- |
|
|
|
|
|
## 🎯 Mission & Values |
|
|
|
|
|
- **Educate**: Provide hands-on, industry-relevant ML/AI education. |
|
|
- **Open**: Build in the open—code, weights, datasets, and docs. |
|
|
- **Reproduce**: Promote rigorous, repeatable experiments and clear reporting. |
|
|
- **Uplift**: Help students and practitioners learn, contribute, and grow together. |
|
|
|
|
|
--- |
|
|
|
|
|
## 🧭 Focus Areas |
|
|
|
|
|
- **NLP**: instruction-tuned LMs, evaluation, tokenization, prompt design. |
|
|
- **Computer Vision**: image classification, detection, segmentation, multimodal. |
|
|
- **Generative AI**: diffusion models, LLM fine-tuning/LoRA, evaluation. |
|
|
- **MLOps & Tooling**: training recipes, experiment tracking, deployment tips. |
|
|
- **Responsible AI**: bias, safety, documentation, and model cards. |
|
|
|
|
|
--- |
|
|
|
|
|
## 📦 What You’ll Find Here |
|
|
|
|
|
- **Models**: Student projects, course baselines, research prototypes. |
|
|
- **Datasets**: Curated, class-safe datasets with clear licenses and splits. |
|
|
- **Spaces**: Demos, teaching assistants, benchmarking dashboards. |
|
|
- **Learning Resources**: Labs, notebooks, reading lists, and cheatsheets. |
|
|
|
|
|
> All repos aim to include a **README/model card**, **license**, **training details**, and **repro steps**. |
|
|
|
|
|
|
|
|
|
