Skip to content

Tutorials

Explore the following tutorials for a better understanding of how to use the RBLN SDK.

RBLN Compiler

These tutorials demonstrate how to use the RBLN Python API (Compile and Runtime APIs) for PyTorch and TensorFlow models. The TensorFlow tutorials target TensorFlow 2.18 with tf_keras (the Keras 2 API); Keras 3 is not supported.

  • PyTorch PyTorch ResNet50 (Vision)
    Compiling a TorchVision ResNet50 model with the RBLN Compile API and running inference.

  • PyTorch PyTorch BERT (NLP)
    Compiling a PyTorch BERT-base model with the RBLN Compile API and running inference.

  • TensorFlow TensorFlow EfficientNetB0 (Vision)
    Compiling a TF Keras Applications EfficientNet-B0 model with the RBLN Compile API and running inference.

  • Concurrent_processing Concurrent Processing
    Running models asynchronously with the RBLN Runtime API (AsyncRuntime).

  • Bucketing Bucketing Bucketing
    Compiling a model that supports multiple input shapes with a single runtime.

RBLN C/C++ Runtime API

These tutorials demonstrate how to deploy precompiled models using the RBLN C/C++ Runtime API.

RBLN Optimum (HuggingFace Model Support)

These tutorials demonstrate how to compile and run inference on HuggingFace models with RBLN Optimum.

  • HuggingFace Llama3-8B (Chatbot)
    Compiling and deploying the Llama3-8B model on a single Rebel100™ NPU with RBLN Optimum.

  • HuggingFace Qwen3-VL-2B (VLM)
    Running a Vision-Language Model effectively on Rebel100™ with RBLN Optimum.

  • HuggingFace Gemma-4-31B (Bucketing)
    Compiling the dense Gemma4 model with bucketing, so one runtime handles several input shapes.

RBLN Profiler

These tutorials demonstrate how to profile and analyze models during inference using the RBLN Runtime.

RBLN PyTorch

These tutorials demonstrate how to run PyTorch code on RBLN NPUs with torch-rbln.

Model Serving

These tutorials demonstrate how to serve models with RBLN vLLM, NVIDIA Triton Inference Server, Ray Serve, and TorchServe.