Fix model card: citation title, line break, add results viewer link

README.md

@@ -13,7 +13,8 @@ library_name: mdiffae
 
 **mDiffAE v2** — **M**asked **Diff**usion **A**uto**E**ncoder v2.
 A fast, single-GPU-trainable diffusion autoencoder with a **96-channel**
-spatial bottleneck and a skip-concat decoder with optional PDG sharpening.
+spatial bottleneck and optional PDG sharpening.
+
 **This is the recommended version** — it offers substantially better
 reconstruction than [v1](https://huggingface.co/data-archetype/mdiffae-v1)
 (+1.7 dB mean PSNR) while maintaining the same or better convergence for
@@ -26,6 +27,7 @@ Bottleneck: **96 channels** at patch size 16
 ## Documentation
 
 - [Technical Report](technical_report_mdiffae_v2.md) — architecture, training changes from v1, and results
+- [Results — interactive viewer](https://huggingface.co/spaces/data-archetype/mdiffae-v2-results) — full-resolution side-by-side comparison
 - [mDiffAE v1](https://huggingface.co/data-archetype/mdiffae-v1) — previous version
 - [iRDiffAE Technical Report](https://huggingface.co/data-archetype/irdiffae-v1/blob/main/technical_report.md) — full background on VP diffusion, DiCo blocks, patchify encoder, AdaLN
 
@@ -117,7 +119,7 @@ recon = model.decode(latents, height=H, width=W, inference_config=cfg)
 
 ```bibtex
 @misc{mdiffae_v2,
-  title   = {mDiffAE v2: A Fast Masked Diffusion Autoencoder with Skip-Concat Decoder},
+  title   = {mDiffAE v2: A Fast Masked Diffusion Autoencoder},
   author  = {data-archetype},
   year    = {2026},
   month   = mar,