[docs]: document the FastH3 VSA kernel requirement

State prominently that VSA checkpoints and adapters require FastVideo VSA-H3 modules and the Video Sparse Attention kernel.

README.md

@@ -24,6 +24,12 @@ family](https://huggingface.co/collections/FastVideo/fastvideo-fasth3).
 Each adapter reconstructs one corresponding distilled transformer from the
 MiniMax H3 base model.
 
+> **VSA requirement:** The three VSA adapters must be loaded and run with
+> FastVideo's Video Sparse Attention (VSA-H3) backend and kernel enabled. VSA
+> must be enabled before adapter loading so the compression-gate modules exist
+> for the adapter's `.set_weight` tensors. Dense attention is not a supported
+> substitute. The Dense / Data-Free adapter does not require VSA.
+
 ## Variants
 
 | Variant | Adapter path | Full checkpoint | Attention | Training source | Step |