DLTK

A hands-on tutorial for efficiently developing and deploying deep learning models.

• Deep learning model development: model design, model training & evaluation.
  • LLM development: data curation, model pre-training, instruction fine-tuning, reinforcement-based alignment.
  • Diffusion development: data curation, model pre-training, efficient fine-tuning.
• Deep learning model deployment: model optimization, inference server.
  • LLM optimizations: vLLM, SGLang, TensorRT-LLM.
  • Diffusion optimizations: diffusers, OneDiff.
• Model optimizations: hardware-aware optimizations, LLM-specific optimizations, Diffusion-specific optimizations.
• An interactive demo for deep learning models: Gradio, Streamlit, React.
• Multimodal understanding: multimodal models, basic vision tasks, advanced vision tasks.
• LLM applications: video understanding, recommendation systems, best practices.

Useful links

If you wish to publish your work on open-source community, the following resources are much helpful.

1. Lightning-AI. "Deep Learning Project Template (Code)". Github repo.
2. eliahuhorwitz. "Academic Project Page Template (Website)". Github repo.
3. mintlify. "The starter kit for your Mintlify docs (Product Spec)". Github repo.

Further readings

For insights into efficient on-device AIGC algorithms and systems, check out my blogs.