v1.1.1

• Fix NaN in FP16 mode.

v1.1.0

• Add support for FP16 mode.
• Add deblur/demosaick/denoise/jpeg/sr models.
• Add cuda_graphs parameter.
• The table_index_mask buffers are now created with padded input dimensions at initialization for better performance.

v1.0.0

• Initial release.

model

Initial commit