Adjust sample timestep sigmas to be model evals for h3 for 1 extra step

extensions_built_in/diffusion_models/minimax_h3/src/packing.py

@@ -431,9 +431,10 @@ def remap_sigma(
 def build_sigma_schedule(
     num_inference_steps: int, shift: float = VIDEO_SIGMA_SHIFT
 ) -> torch.Tensor:
-    """The released sampling grid: linspace(1, 0, steps) through the
-    exponential shift, consecutive duplicates collapsed — the terminal 0 is
-    part of the count, so `steps` yields `steps - 1` model evaluations."""
-    base = torch.linspace(1.0, 0.0, num_inference_steps, dtype=torch.float32)
+    """The released sampling grid: linspace(1, 0, steps + 1) through the
+    exponential shift, consecutive duplicates collapsed — `steps` yields
+    `steps` model evaluations (the released repo counts the terminal 0 in
+    `steps`; we don't, so sample_steps means model evals)."""
+    base = torch.linspace(1.0, 0.0, num_inference_steps + 1, dtype=torch.float32)
     sigmas = shift_sigma(base, shift)
     return torch.unique_consecutive(sigmas)

extensions_built_in/diffusion_models/minimax_h3/src/pipeline.py

@@ -9,9 +9,9 @@ exactly one transformer forward per step. ``unconditional_embeds`` and
 ``guidance_scale`` are accepted for harness compatibility and ignored.
 
 Scheduler (the released math, not diffusers'):
-  - sigma grid: ``linspace(1, 0, steps)`` through the exponential shift
-    (video 12, audio 3), consecutive duplicates collapsed; the terminal 0 is
-    part of the count so ``steps`` yields ``steps - 1`` model evaluations
+  - sigma grid: ``linspace(1, 0, steps + 1)`` through the exponential shift
+    (video 12, audio 3), consecutive duplicates collapsed; ``steps`` yields
+    ``steps`` model evaluations (steps = 1 is one full 1 -> 0 step)
   - the model consumes ``t = 1 - sigma`` (t = 1 means clean) and predicts the
     data-ward velocity ``clean - noise``: ``denoised = x + sigma * v``
   - Euler update ``x_next = r * x + (1 - r) * denoised`` with