MiniMax H3 cfg-distilled full denoise loop.
Per step, the positive presentation is forwarded exactly once. Video and audio target rows chain through the Euler-eta0 update while visual and audio condition rows stay pinned to their noised step-0 anchors.
Static per-branch state: packed layout + fixed forward kwargs.
packed is a minimax_h3_packed_sequence(...) result (or equivalent layout dict); text_embeddings is the branch's [text_len, 5120] hidden states; token_tags must already carry any fl2va vision-span overrides.
instance-attribute ¶
instance-attribute ¶
instance-attribute ¶
audio_x_base = torch.zeros(
1,
seq_len,
MINIMAX_H3_AUDIO_ROW_WIDTH,
dtype=torch.float32,
device=device,
)
instance-attribute ¶
instance-attribute ¶
static_kwargs: dict[str, Any] = {
"img_position_ids": self.img_position_ids_dev[None],
"update_mask": self.update_mask_dev,
"token_tags": self.token_tags_dev,
"skip_mask_out_condition": False,
"prompt_embeds": self.text_embeddings_dev,
"img_pos_info": {"position_ids": self.img_pos_dev},
"audio_pos_info": {"position_ids": self.audio_pos_dev},
"text_pos_info": {"position_ids": self.text_pos_dev},
"img_pos_for_infer_output_info": {
"position_ids": self.img_pos_dev
},
"packed_seq_params": {
"cu_seqlens_q": cu.to(device),
"max_seqlen_q": self.used_len,
"num_requests": 1,
},
"refiner_packed_seq_params": {
"cu_seqlens_q": torch.tensor(
[0, text_len, text_len],
dtype=torch.int32,
device=device,
),
"max_seqlen_q": text_len,
},
}
instance-attribute ¶
instance-attribute ¶
instance-attribute ¶
x_base = torch.zeros(
1,
seq_len,
MINIMAX_H3_VIDEO_ROW_WIDTH,
dtype=torch.float32,
device=device,
)
fill_timesteps(
timesteps: Tensor,
*,
t_video: float,
t_audio: float,
imgvid_cond_timestep: float,
audio_ref_cond_timestep: float,
video_target_timesteps: Tensor | None = None,
audio_target_timesteps: Tensor | None = None,
) -> None
Fill one request's packed row timesteps in an existing tensor.
forward_kwargs(
*,
video_rows: Tensor,
audio_rows: Tensor,
t_video: float,
t_audio: float,
imgvid_cond_timestep: float,
audio_ref_cond_timestep: float,
video_target_timesteps: Tensor | None = None,
audio_target_timesteps: Tensor | None = None,
) -> dict[str, Any]
prepare_rope_table(model: Any) -> None
Materialize the branch-local DiT RoPE table once per denoise run.
img_position_ids is immutable for this branch while latents and timesteps change every scheduler step. Keeping the table in static_kwargs makes every model call reuse the exact same BF16 tensor without extending its lifetime beyond this request branch.
minimax_h3_denoise_loop(
*,
model: Any,
positive: MiniMaxH3DenoiseBranch,
initial_video_rows: Tensor,
initial_audio_rows: Tensor,
keyframe_cond_rows: Tensor | None,
audio_ref_rows: Tensor | None = None,
video_edit: MiniMaxH3LatentEdit | None = None,
audio_edit: MiniMaxH3LatentEdit | None = None,
sigmas_video: list[float],
sigmas_audio: list[float],
device: device,
imgvid_cond_noise_aug_for_inference: float = MINIMAX_H3_IMGVID_COND_TIMESTEP,
audio_cond_noise_aug_for_inference: float = MINIMAX_H3_AUDIO_REF_COND_TIMESTEP,
on_step: Callable[[int, Tensor, Tensor], None]
| None = None,
step_profiler: Callable[[int], AbstractContextManager]
| None = None,
) -> tuple[Tensor, Tensor]
Run the full denoise loop; returns final (video_rows, audio_rows).
initial_video_rows covers all image rows of the positive layout. For a conditional task, pass keyframe_cond_rows and/or audio_ref_rows to pin those rows across every step. The model's raw positive velocity is the update signal; MiniMax H3 only supports cfg-distilled checkpoints.
minimax_h3_prepare_denoise_rows(
*,
positive: MiniMaxH3DenoiseBranch,
initial_video_rows: Tensor,
initial_audio_rows: Tensor,
keyframe_cond_rows: Tensor | None,
audio_ref_rows: Tensor | None,
device: device,
) -> tuple[Tensor, Tensor, Tensor | None, Tensor | None]
Validate the initial rows against the layout and pin the condition rows.
Returns (video_rows, audio_rows, cond_anchor, audio_anchor) with the anchors already written into their rows, which is the state both the request-mode loop and step-mode prepare_encode() start from.
minimax_h3_publish_denoise_progress(
step: int | None,
sigma_video: float | None,
total_steps: int | None = None,
) -> None
Publish denoise progress for step-gated attention features.
Both execution modes must publish the same trio: the step index drives the dense warmup of RAINFUSION_ATTN, the normalized descending timestep drives the TRTLLM_ATTN skip gate (which stays dense while it is unset), and the total step count enables the end_step tail fallback.