wlsaidhi commited on
Commit
b06b101
·
verified ·
1 Parent(s): cf2917a

[docs]: document the FastH3 VSA kernel requirement

Browse files

State prominently that this checkpoint requires FastVideo VSA-H3 and the Video Sparse Attention kernel.

Files changed (1) hide show
  1. README.md +4 -0
README.md CHANGED
@@ -29,6 +29,10 @@ synchronized video and audio with four transformer forwards and uses VSA-H3 at
29
  See the complete [FastH3 4-Step Preview v1
30
  collection](https://huggingface.co/collections/FastVideo/fastvideo-fasth3).
31
 
 
 
 
 
32
  ## Checkpoint identity
33
 
34
  - Training run: `20260823T132231Z-v10p5-datafree-sp4`
 
29
  See the complete [FastH3 4-Step Preview v1
30
  collection](https://huggingface.co/collections/FastVideo/fastvideo-fasth3).
31
 
32
+ > **VSA requirement:** This checkpoint must be run with FastVideo's Video
33
+ > Sparse Attention (VSA-H3) backend and kernel enabled at 90% sparsity with
34
+ > 64-token tiles. Dense attention is not a supported substitute.
35
+
36
  ## Checkpoint identity
37
 
38
  - Training run: `20260823T132231Z-v10p5-datafree-sp4`