Use AutoScheduler for Template-Free Scheduling¶

The TVM AutoScheduler offers a template-free way to tune models. These how-tos demonstrate how to tune a variety of different models to target a number of common platforms.

../../_images/sphx_glr_tune_conv2d_layer_cuda_thumb.png

Auto-scheduling a Convolution Layer for GPU¶

../../_images/sphx_glr_tune_network_x86_thumb.png

Auto-scheduling a Neural Network for x86 CPU¶

../../_images/sphx_glr_tune_network_cuda_thumb.png

Auto-scheduling a Neural Network for NVIDIA GPU¶

../../_images/sphx_glr_tune_network_arm_thumb.png

Auto-scheduling a Neural Network for ARM CPU¶

../../_images/sphx_glr_tune_network_mali_thumb.png

Auto-scheduling a Neural Network for mali GPU¶

../../_images/sphx_glr_tune_sparse_x86_thumb.png

Auto-scheduling Sparse Matrix Multiplication on CPU with Custom Sketch Rule¶

Gallery generated by Sphinx-Gallery