Document compressed HQ Core ML release

README.md

@@ -18,24 +18,27 @@ The iOS 18 batch-one U-Net accepts `[1, 9, 64, 64]`, runs classifier-free
 guidance as two serial passes, and exposes 144 mutable Core ML states for exact
 runtime composition of up to three Clover styles.
 
-The HQ U-Net is under `hq-v3/Unet.mlmodelc`. Shared Clover resources remain at
-the repository root so app upgrades only download files whose checksums
-changed. `hq-v3/adapter-schema.json` maps each small `.safetensors` style into
-one of three independent state slots.
+The production U-Net is under `hq-v4-int8/Unet.mlmodelc`. Its immutable
+weights use per-channel symmetric int8 compression, reducing the complete
+inpainting add-on from about 1.79 GB to **931 MB** while leaving mutable LoRA
+state in FP16. Shared tokenizer, text encoder, and VAE decoder resources are
+hard-linked from the required main Clover installation instead of downloaded
+again. `hq-v4-int8/adapter-schema.json` maps each small `.safetensors` style
+into one of three independent state slots.
 
 ## Validation
 
 PyTorch/Core ML parity on deterministic inputs:
 
-| Configuration | PSNR |
-|---|---:|
-| Base HQ inpainting U-Net | **78.56 dB** |
-| Monet 0.70 + Pointillism 0.45 + Watercolor Anime 1.10 | **78.47 dB** |
+| Configuration | FP16 | Int8 production |
+|---|---:|---:|
+| Base HQ inpainting U-Net | 78.56 dB | **59.32 dB** |
+| Monet 0.70 + Pointillism 0.45 + Watercolor Anime 1.10 | 78.47 dB | **59.30 dB** |
 
 The release gate is 35 dB. The three-style result validates the exact
 block-concatenated LoRA sum rather than a UI-only approximation.
 
-Held-out 24-case inpainting quality relative to the previous release:
+Held-out 24-case inpainting quality relative to the previous Diffusers release:
 
 | Metric | Previous | HQ |
 |---|---:|---:|