Changelog for the AI Developer Hub

AI developer tutorials are available at the AI Developer Hub.

Version 6.0

Added

• New inference tutorials:

  • Accelerating DeepSeek-V3 inference using multi-token prediction in SGLang

  • Multi-agents with Google ADK and A2A protocol

Removed

• The following inference tutorials were removed:

  • Hugging Face TGI

  • Llama Stack

• The following fine-tuning tutorials were removed:

  • LLM with LoRA

  • LLM with QLoRA

• The following pretraining tutorial was removed:

  • OLMo model with PyTorch FSDP

Version 5.1

Added

• New inference tutorials:

  • PD disaggregation with SGLang

Changed

• Updated the pydantic_ai argument and Node version in the AI agent with MCPs using vLLM and PydanticAI tutorial.

Version 5.0

Added

• New inference tutorials:

  • ChatQnA vLLM deployment and performance evaluation

  • Text-to-video generation with ComfyUI

  • DeepSeek Janus Pro on CPU or GPU

  • DeepSeek-R1 with vLLM V1

• New GPU development and optimization tutorial:

  • MLA decoding kernel of AITER library

Version 4.0

Added

• New inference tutorial:

  • AI agent with MCPs using vLLM and PydanticAI

• New GPU development and optimization tutorials:

  • Kernel development and optimization with Triton

  • Profiling Llama-4 inference with vLLM

  • FP8 quantization with AMD Quark for vLLM

Changed

• DDIM pretraining tutorial renamed to Custom diffusion model with PyTorch with some minor changes.

Version 3.1

Added

• New fine-tuning tutorial:

  • GRPO with Unsloth

Version 3.0

Added

• New inference tutorials:

  • Speculative decoding with vLLM

  • Llama Stack

  • DeepSeek-R1 with SGLang

• New fine-tuning tutorial:

  • Llama-3.1 8B with Llama-Factory

• New pretraining tutorial:

  • DDIM

Version 2.0

Added

• New inference tutorials:

  • OCR with vision-language models with vLLM

  • Building AI pipelines for voice assistants

• New fine-tuning tutorial:

  • Llama-3.1 8B with torchtune

• New pretraining tutorial:

  • Llama-3.1 8B with torchtitan

Changed

• Updated the supported ROCm versions for some guides

• Additional minor changes

Version 1.0

Added

• New inference tutorials:

  • Hugging Face Transformers

  • Hugging Face TGI

  • Deploying with vLLM

  • From chatbot to rap bot with vLLM

  • RAG with LlamaIndex and Ollama

• New fine-tuning tutorials:

  • VLM with PEFT

  • LLM with LoRA

  • LLM with QLoRA

• New pretraining tutorials:

  • OLMo model with PyTorch FSDP

  • Training configuration with Megatron-LM

  • LLM with Megatron-LM