Model card: the half-size figures per backend, and what the GFPGAN file is

README.md

@@ -78,8 +78,10 @@ float32. Nothing else in the graph changes. onnxruntime folds the casts into
 ordinary float32 weights when it builds the session, so all arithmetic is
 still float32; what is lost is the rounding of each weight to 11 significant
 bits. Measured against the float32 files over ten real photographs, the
-outputs agree to 59–69 dB PSNR, and no pixel differs by more than 5 of 255
-levels. The download is half the size.
+outputs agree to 59–69 dB PSNR on onnxruntime's CPU backend, with no pixel
+more than 3 of 255 levels apart, and to 59–67 dB in Chromium on
+onnxruntime-web's WebGPU and WASM backends, with at most 5 of 255. The
+download is half the size.
 
 ## Where each file comes from
 
@@ -116,9 +118,11 @@ source code in the GFPGAN repository, and two of them carry non-commercial
 terms: StyleGAN2 code, modified from rosinality's stylegan2-pytorch, whose
 upstream (NVlabs/stylegan2) is under the NVIDIA licence reproduced there,
 including its section 3.3 use limitation; and code "largely modified from"
-DFDNet, under CC BY-NC-SA 4.0. These terms attach to that source code:
-`gfpgan_1.4_w16.onnx` is the trained weights and an ONNX graph, and contains
-none of GFPGAN's source code. GFPGAN's README separately offers a "clean"
+DFDNet, under CC BY-NC-SA 4.0. `gfpgan_1.4_w16.onnx` contains no source
+code: it is the trained GFPGAN v1.4 weights and an ONNX graph of the network
+they belong to, `GFPGANv1Clean`, whose generator follows the StyleGAN2
+architecture that the listed StyleGAN2 code implements. Read the full terms
+before relying on this file for commercial use. GFPGAN's README separately offers a "clean"
 version of the architecture that does not require its customised CUDA
 extensions. The full text is included so the terms can be read as they are
 written rather than as summarised here.