Record H3 VAE pixel gate

This commit is contained in:
Daniel Maddern 2026-08-13 16:00:57 +07:00
parent 15ac9b4c9c
commit 55e2073784

View file

@ -187,7 +187,7 @@ gate and is exact.
| H3 final layer | Yes | Bit-exact final AdaLN, target-row modulation, and video/audio patch rows; Comfy materializes the AdaLN and output-head biases through BF16 |
| H3 DiT | Yes | Strict all-block numeric parity not achieved |
| Beta/RES sampler | Yes | H3 replay is bit-exact for all 12 captured sampler calls and RES update is bit-exact from captured denoised; direct raw-output to model-wrapper denoised conversion retains small BF16/state boundary deltas |
| Video VAE decoder | Yes | No direct-versus-Comfy same-latent pixel comparison yet |
| Video VAE decoder | Yes | Same final sampled latent decoded through direct VAE versus Comfy PNG frames: mean abs pixel delta `0.0833898`, max `6` |
| Audio VAE/decode/mux | No | Preview intentionally produces video only |
| Full Ref2VA | No | References, vision conditioning, VAE encode, audio, and muxing are unimplemented |
@ -198,12 +198,9 @@ unless the checkpoint, Comfy version, prompt, or backend changes.
1. **Final video latent layout.** Compare direct `unpatchify_video` output
with Comfy before entering the sampler.
2. **VAE pixel gate.** Decode the identical captured final video latent in
direct and Comfy and compare pixels before ffmpeg. This decides whether any
residual grid comes from decoder behavior.
3. **End-to-end FL2VA preview.** Generate the reference dragon only after
2. **End-to-end FL2VA preview.** Generate the reference dragon only after
gates 1-8 pass. Compare its raw final latent first, then video.
4. **Feature/performance work.** Only then add audio, Ref2VA/reference paths,
3. **Feature/performance work.** Only then add audio, Ref2VA/reference paths,
Sage3, CUDA graphs, and multi-GPU execution.
## Existing Tools And Their Intended Gate