nodes/kijai-ComfyUI-WanVideoWrapper
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Use fp32 norms for other models too and other fixes

Browse Source
This commit is contained in:
kijai
2025-10-28 01:52:48 +02:00
parent e69e068b57
commit c80a488f70
3 changed files with 82 additions and 111 deletions
Download Patch File Download Diff File Expand all files Collapse all files
HuMo/audio_proj.py
+3 -3
View File
@@ -16,10 +16,10 @@ class WanRMSNorm(nn.Module):
Args: Args:
x(Tensor): Shape [B, L, C] x(Tensor): Shape [B, L, C]
""" """
return self._norm(x.float()).type_as(x) * self.weight return self._norm(x.to(self.weight.dtype)) * self.weight
def _norm(self, x): def _norm(self, x):
return x * torch.rsqrt(x.pow(2).mean(dim=-1, keepdim=True) + self.eps) return x * (torch.rsqrt(x.pow(2).mean(dim=-1, keepdim=True) + self.eps)).to(x.dtype)
class DummyAdapterLayer(nn.Module): class DummyAdapterLayer(nn.Module):
@@ -81,7 +81,7 @@ class AudioProjModel(nn.Module):
context_tokens = self.audio_proj_glob_3(audio_embeds).reshape(batch_size, self.context_tokens, self.output_dim) context_tokens = self.audio_proj_glob_3(audio_embeds).reshape(batch_size, self.context_tokens, self.output_dim)
context_tokens = self.audio_proj_glob_norm(context_tokens) context_tokens = self.audio_proj_glob_norm(context_tokens.to(self.audio_proj_glob_norm.layer.weight.dtype)).to(audio_embeds.dtype)
context_tokens = rearrange(context_tokens, "(bz f) m c -> bz f m c", f=video_length) context_tokens = rearrange(context_tokens, "(bz f) m c -> bz f m c", f=video_length)
return context_tokens return context_tokens
nodes_sampler.py
+1 -2
View File
@@ -995,12 +995,11 @@ class WanVideoSampler:
# Experimental args # Experimental args
use_cfg_zero_star = use_tangential = use_fresca = bidirectional_sampling = use_tsr = False use_cfg_zero_star = use_tangential = use_fresca = bidirectional_sampling = use_tsr = False
raag_alpha = 0.0 raag_alpha = 0.0
transformer.video_attention_split_steps = []
if experimental_args is not None: if experimental_args is not None:
video_attention_split_steps = experimental_args.get("video_attention_split_steps", []) video_attention_split_steps = experimental_args.get("video_attention_split_steps", [])
if video_attention_split_steps: if video_attention_split_steps:
transformer.video_attention_split_steps = [int(x.strip()) for x in video_attention_split_steps.split(",")] transformer.video_attention_split_steps = [int(x.strip()) for x in video_attention_split_steps.split(",")]
else:
transformer.video_attention_split_steps = []
use_zero_init = experimental_args.get("use_zero_init", True) use_zero_init = experimental_args.get("use_zero_init", True)
use_cfg_zero_star = experimental_args.get("cfg_zero_star", False) use_cfg_zero_star = experimental_args.get("cfg_zero_star", False)
wanvideo/modules/model.py
+66 -94
View File
@@ -345,10 +345,10 @@ class WanRMSNorm(nn.Module):
if use_chunked: if use_chunked:
return self.forward_chunked(x, num_chunks) return self.forward_chunked(x, num_chunks)
else: else:
return self._norm(x.float()).type_as(x) * self.weight return self._norm(x.to(self.weight.dtype)) * self.weight
def _norm(self, x): def _norm(self, x):
return x * torch.rsqrt(x.pow(2).mean(dim=-1, keepdim=True) + self.eps) return x * (torch.rsqrt(x.pow(2).mean(dim=-1, keepdim=True) + self.eps)).to(x.dtype)
def forward_chunked(self, x, num_chunks=4): def forward_chunked(self, x, num_chunks=4):
output = torch.empty_like(x) output = torch.empty_like(x)
@@ -398,23 +398,23 @@ class WanLayerNorm(nn.LayerNorm):
def __init__(self, dim, eps=1e-6, elementwise_affine=False): def __init__(self, dim, eps=1e-6, elementwise_affine=False):
super().__init__(dim, elementwise_affine=elementwise_affine, eps=eps) super().__init__(dim, elementwise_affine=elementwise_affine, eps=eps)
def forward(self, inputs: torch.Tensor) -> torch.Tensor: # def forward(self, inputs: torch.Tensor) -> torch.Tensor:
origin_dtype = inputs.dtype # origin_dtype = inputs.dtype
out = F.layer_norm( # out = F.layer_norm(
inputs.float(), # inputs.float(),
self.normalized_shape, # self.normalized_shape,
None if self.weight is None else self.weight.float(), # None if self.weight is None else self.weight.float(),
None if self.bias is None else self.bias.float() , # None if self.bias is None else self.bias.float() ,
self.eps # self.eps
).to(origin_dtype) # ).to(origin_dtype)
return out # return out
# def forward(self, x): def forward(self, x):
# r""" r"""
# Args: Args:
# x(Tensor): Shape [B, L, C] x(Tensor): Shape [B, L, C]
# """ """
# return super().forward(x) return super().forward(x)
class WanSelfAttention(nn.Module): class WanSelfAttention(nn.Module):
@@ -465,8 +465,8 @@ class WanSelfAttention(nn.Module):
def qkv_fn(self, x): def qkv_fn(self, x):
b, s, n, d = *x.shape[:2], self.num_heads, self.head_dim b, s, n, d = *x.shape[:2], self.num_heads, self.head_dim
q = self.norm_q(self.q(x)).view(b, s, n, d) q = self.norm_q(self.q(x).to(self.norm_q.weight.dtype)).to(x.dtype).view(b, s, n, d)
k = self.norm_k(self.k(x)).view(b, s, n, d) k = self.norm_k(self.k(x).to(self.norm_k.weight.dtype)).to(x.dtype).view(b, s, n, d)
v = self.v(x).view(b, s, n, d) v = self.v(x).view(b, s, n, d)
return q, k, v return q, k, v
@@ -481,8 +481,8 @@ class WanSelfAttention(nn.Module):
def qkv_fn_ip(self, x): def qkv_fn_ip(self, x):
b, s, n, d = *x.shape[:2], self.num_heads, self.head_dim b, s, n, d = *x.shape[:2], self.num_heads, self.head_dim
q = self.norm_q(self.q(x) + self.q_loras(x)).view(b, s, n, d) q = self.norm_q(self.q(x) + self.q_loras(x).to(self.norm_q.weight.dtype)).to(x.dtype).view(b, s, n, d)
k = self.norm_k(self.k(x) + self.k_loras(x)).view(b, s, n, d) k = self.norm_k(self.k(x) + self.k_loras(x).to(self.norm_k.weight.dtype)).to(x.dtype).view(b, s, n, d)
v = (self.v(x) + self.v_loras(x)).view(b, s, n, d) v = (self.v(x) + self.v_loras(x)).view(b, s, n, d)
return q, k, v return q, k, v
@@ -549,7 +549,7 @@ class WanSelfAttention(nn.Module):
return x, x_ref_attn_map return x, x_ref_attn_map
def forward_split(self, q, k, v, seq_lens, grid_sizes, freqs, seq_chunks=1,current_step=0, video_attention_split_steps = []): def forward_split(self, q, k, v, seq_lens, grid_sizes, seq_chunks):
r""" r"""
Args: Args:
x(Tensor): Shape [B, L, num_heads, C / num_heads] x(Tensor): Shape [B, L, num_heads, C / num_heads]
@@ -559,62 +559,36 @@ class WanSelfAttention(nn.Module):
""" """
# Split by frames if multiple prompts are provided # Split by frames if multiple prompts are provided
if seq_chunks > 1 and current_step in video_attention_split_steps: frames, height, width = grid_sizes[0]
outputs = []
# Extract frame, height, width from grid_sizes
frames = grid_sizes[0][0]
height = grid_sizes[0][1]
width = grid_sizes[0][2]
tokens_per_frame = height * width tokens_per_frame = height * width
actual_chunks = torch.min(torch.tensor(seq_chunks, device=frames.device), frames) seq_chunks_tensor = torch.tensor(seq_chunks, device=q.device, dtype=frames.dtype)
actual_chunks = torch.minimum(seq_chunks_tensor, frames)
base_frames_per_chunk = frames // actual_chunks base_frames_per_chunk = frames // actual_chunks
extra_frames = frames % actual_chunks extra_frames = frames % actual_chunks
# Calculate all chunk boundaries chunk_indices = torch.arange(actual_chunks, device=q.device)
chunk_indices = torch.arange(actual_chunks, device=frames.device) chunk_sizes = base_frames_per_chunk + (chunk_indices < extra_frames)
chunk_sizes = base_frames_per_chunk + (chunk_indices < extra_frames).long() chunk_starts = torch.cumsum(torch.cat([torch.zeros(1, device=q.device, dtype=torch.long), chunk_sizes[:-1]]), dim=0)
chunk_starts = torch.cumsum(torch.cat([torch.zeros(1, device=frames.device), chunk_sizes[:-1]]), dim=0).long()
chunk_ends = chunk_starts + chunk_sizes chunk_ends = chunk_starts + chunk_sizes
# Process each chunk using tensor indexing outputs = []
for i in range(actual_chunks.item()): for i in chunk_indices:
start_frame = chunk_starts[i] start_idx = chunk_starts[i] * tokens_per_frame
end_frame = chunk_ends[i] end_idx = chunk_ends[i] * tokens_per_frame
# Convert to token indices using tensor operations
start_idx = start_frame * tokens_per_frame
end_idx = end_frame * tokens_per_frame
chunk_q = q[:, start_idx:end_idx, :, :]
chunk_k = k[:, start_idx:end_idx, :, :]
chunk_v = v[:, start_idx:end_idx, :, :]
chunk_out = attention( chunk_out = attention(
q=chunk_q, q[:, start_idx:end_idx, :, :],
k=chunk_k, k[:, start_idx:end_idx, :, :],
v=chunk_v, v[:, start_idx:end_idx, :, :],
k_lens=seq_lens, k_lens=seq_lens,
attention_mode=self.attention_mode) attention_mode=self.attention_mode
)
outputs.append(chunk_out) outputs.append(chunk_out)
# Concatenate outputs along the sequence dimension
x = torch.cat(outputs, dim=1) x = torch.cat(outputs, dim=1)
else:
# Original attention computation
x = attention(
q=q,
k=k,
v=v,
k_lens=seq_lens,
attention_mode=self.attention_mode)
# output # output
x = x.flatten(2) return self.o(x.flatten(2))
x = self.o(x)
return x
def normalized_attention_guidance(self, b, n, d, q, context, nag_context=None, nag_params={}): def normalized_attention_guidance(self, b, n, d, q, context, nag_context=None, nag_params={}):
# NAG text attention # NAG text attention
@@ -624,9 +598,9 @@ class WanSelfAttention(nn.Module):
nag_alpha = nag_params['nag_alpha'] nag_alpha = nag_params['nag_alpha']
nag_tau = nag_params['nag_tau'] nag_tau = nag_params['nag_tau']
k_positive = self.norm_k(self.k(context_positive)).view(b, -1, n, d) k_positive = self.norm_k(self.k(context_positive).to(self.norm_k.weight.dtype)).view(b, -1, n, d).to(q.dtype)
v_positive = self.v(context_positive).view(b, -1, n, d) v_positive = self.v(context_positive).view(b, -1, n, d)
k_negative = self.norm_k(self.k(context_negative)).view(b, -1, n, d) k_negative = self.norm_k(self.k(context_negative).to(self.norm_k.weight.dtype)).view(b, -1, n, d).to(q.dtype)
v_negative = self.v(context_negative).view(b, -1, n, d) v_negative = self.v(context_negative).view(b, -1, n, d)
x_positive = attention(q, k_positive, v_positive, attention_mode=self.attention_mode) x_positive = attention(q, k_positive, v_positive, attention_mode=self.attention_mode)
@@ -703,15 +677,15 @@ class WanT2VCrossAttention(WanSelfAttention):
x = x[:, num_cond_latents_thw:] x = x[:, num_cond_latents_thw:]
q = self.norm_q(self.q(x).view(b, -1, n, d)) q = self.norm_q(self.q(x).view(b, -1, n, d))
else: else:
q = self.norm_q(self.q(x),num_chunks=2 if rope_func == "comfy_chunked" else 1).view(b, -1, n, d) q = self.norm_q(self.q(x).to(self.norm_q.weight.dtype),num_chunks=2 if rope_func == "comfy_chunked" else 1).to(x.dtype).view(b, -1, n, d)
if nag_context is not None and not is_uncond: if nag_context is not None and not is_uncond:
x = self.normalized_attention_guidance(b, n, d, q, context, nag_context, nag_params) x = self.normalized_attention_guidance(b, n, d, q, context, nag_context, nag_params)
else: else:
if is_longcat: if is_longcat:
k = self.norm_k(self.k(context).view(b, -1, n, d)) k = self.norm_k(self.k(context).to(self.norm_k.weight.dtype)).to(x.dtype).view(b, -1, n, d)
else: else:
k = self.norm_k(self.k(context)).view(b, -1, n, d) k = self.norm_k(self.k(context).to(self.norm_k.weight.dtype)).to(x.dtype).view(b, -1, n, d)
v = self.v(context).view(b, -1, n, d) v = self.v(context).view(b, -1, n, d)
@@ -794,19 +768,19 @@ class WanI2VCrossAttention(WanSelfAttention):
""" """
b, n, d = x.size(0), self.num_heads, self.head_dim b, n, d = x.size(0), self.num_heads, self.head_dim
# compute query # compute query
q = self.norm_q(self.q(x),num_chunks=2 if rope_func == "comfy_chunked" else 1).view(b, -1, n, d) q = self.norm_q(self.q(x).to(self.norm_q.weight.dtype),num_chunks=2 if rope_func == "comfy_chunked" else 1).view(b, -1, n, d).to(x.dtype)
if nag_context is not None and not is_uncond: if nag_context is not None and not is_uncond:
x_text = self.normalized_attention_guidance(b, n, d, q, context, nag_context, nag_params) x_text = self.normalized_attention_guidance(b, n, d, q, context, nag_context, nag_params)
else: else:
# text attention # text attention
k = self.norm_k(self.k(context)).view(b, -1, n, d) k = self.norm_k(self.k(context).to(self.norm_k.weight.dtype)).view(b, -1, n, d).to(x.dtype)
v = self.v(context).view(b, -1, n, d) v = self.v(context).view(b, -1, n, d)
x_text = attention(q, k, v, attention_mode=self.attention_mode).flatten(2) x_text = attention(q, k, v, attention_mode=self.attention_mode).flatten(2)
#img attention #img attention
if clip_embed is not None: if clip_embed is not None:
k_img = self.norm_k_img(self.k_img(clip_embed)).view(b, -1, n, d) k_img = self.norm_k_img(self.k_img(clip_embed).to(self.norm_k_img.weight.dtype)).view(b, -1, n, d).to(x.dtype)
v_img = self.v_img(clip_embed).view(b, -1, n, d) v_img = self.v_img(clip_embed).view(b, -1, n, d)
img_x = attention(q, k_img, v_img, attention_mode=self.attention_mode).flatten(2) img_x = attention(q, k_img, v_img, attention_mode=self.attention_mode).flatten(2)
x = x_text + img_x x = x_text + img_x
@@ -1068,8 +1042,9 @@ class WanAttentionBlock(nn.Module):
def forward( def forward(
self, x, e, seq_lens, grid_sizes, freqs, context, current_step, self, x, e, seq_lens, grid_sizes, freqs, context, current_step,
last_step=False, last_step=False,
video_attention_split_steps=[],
clip_embed=None, clip_embed=None,
seq_chunks=0, #comfy chunked cross-attn
chunked_self_attention=False,
camera_embed=None, #ReCamMaster camera_embed=None, #ReCamMaster
audio_proj=None, audio_scale=1.0, #fantasytalking audio_proj=None, audio_scale=1.0, #fantasytalking
num_latent_frames=21, num_latent_frames=21,
@@ -1111,9 +1086,9 @@ class WanAttentionBlock(nn.Module):
T = num_latent_frames T = num_latent_frames
is_longcat = C == 4096 is_longcat = C == 4096
if is_longcat: if is_longcat:
input_x = self.modulate(self.norm1(x.view(B, T, -1, C).float()), shift_msa, scale_msa, seg_idx=self.seg_idx).to(input_dtype).view(B, N, C) input_x = self.modulate(self.norm1(x.view(B, T, -1, C).to(shift_msa.dtype)), shift_msa, scale_msa, seg_idx=self.seg_idx).to(input_dtype).view(B, N, C)
else: else:
input_x = self.modulate(self.norm1(x), shift_msa, scale_msa, seg_idx=self.seg_idx) input_x = self.modulate(self.norm1(x.to(shift_msa.dtype)), shift_msa, scale_msa, seg_idx=self.seg_idx).to(input_dtype)
del shift_msa, scale_msa del shift_msa, scale_msa
@@ -1199,14 +1174,8 @@ class WanAttentionBlock(nn.Module):
and inner_t is None and inner_t is None
and x_ip is None # Don't split when using IP-Adapter and x_ip is None # Don't split when using IP-Adapter
) )
if split_attn: if split_attn and chunked_self_attention:
y = self.self_attn.forward_split( y = self.self_attn.forward_split(q, k, v, seq_lens, grid_sizes, seq_chunks)
q, k, v,
seq_lens, grid_sizes, freqs,
seq_chunks=max(context.shape[0], clip_embed.shape[0] if clip_embed is not None else 0),
current_step=current_step,
video_attention_split_steps=video_attention_split_steps
)
elif ref_target_masks is not None: #multi/infinite talk elif ref_target_masks is not None: #multi/infinite talk
y, x_ref_attn_map = self.self_attn.forward_multitalk(q, k, v, seq_lens, grid_sizes, ref_target_masks) y, x_ref_attn_map = self.self_attn.forward_multitalk(q, k, v, seq_lens, grid_sizes, ref_target_masks)
elif self.attention_mode == "radial_sage_attention": elif self.attention_mode == "radial_sage_attention":
@@ -1301,14 +1270,13 @@ class WanAttentionBlock(nn.Module):
target_seq_lens=seq_lens_ovi, target_seq_lens=seq_lens_ovi,
target_grid_sizes=grid_sizes_ovi, target_grid_sizes=grid_sizes_ovi,
target_freqs=freqs_ovi) target_freqs=freqs_ovi)
y = self.ffn(torch.addcmul(shift_mlp, self.norm2(x), 1 + scale_mlp))
x = x.addcmul(y, gate_mlp)
elif split_attn: elif split_attn:
if nag_context is not None: if nag_context is not None:
raise NotImplementedError("nag_context is not supported in split_cross_attn_ffn") raise NotImplementedError("nag_context is not supported in split_cross_attn_ffn")
x = self.split_cross_attn_ffn(x, context, shift_mlp, scale_mlp, gate_mlp, clip_embed, grid_sizes) x = self.split_cross_attn_ffn(x, context, shift_mlp, scale_mlp, gate_mlp, clip_embed, grid_sizes)
return x, x_ip, lynx_ref_feature, x_ovi
else: else:
x = x + self.cross_attn(self.norm3(x), context, grid_sizes, clip_embed=clip_embed, audio_proj=audio_proj, audio_scale=audio_scale, x = x + self.cross_attn(self.norm3(x.to(self.norm3.weight.dtype)).to(input_dtype), context, grid_sizes, clip_embed=clip_embed, audio_proj=audio_proj, audio_scale=audio_scale,
num_latent_frames=num_latent_frames, nag_params=nag_params, nag_context=nag_context, is_uncond=is_uncond, num_latent_frames=num_latent_frames, nag_params=nag_params, nag_context=nag_context, is_uncond=is_uncond,
rope_func=self.rope_func, inner_t=inner_t, inner_c=inner_c, cross_freqs=cross_freqs, rope_func=self.rope_func, inner_t=inner_t, inner_c=inner_c, cross_freqs=cross_freqs,
adapter_proj=adapter_proj, ip_scale=ip_scale, orig_seq_len=original_seq_len, lynx_x_ip=lynx_x_ip, lynx_ip_scale=lynx_ip_scale, num_cond_latents=num_cond_latents) adapter_proj=adapter_proj, ip_scale=ip_scale, orig_seq_len=original_seq_len, lynx_x_ip=lynx_x_ip, lynx_ip_scale=lynx_ip_scale, num_cond_latents=num_cond_latents)
@@ -1343,10 +1311,10 @@ class WanAttentionBlock(nn.Module):
x_ffn = self.ffn(norm2_x) x_ffn = self.ffn(norm2_x)
else: else:
if not is_longcat: if not is_longcat:
mod_x = torch.addcmul(shift_mlp, self.norm2(x), 1 + scale_mlp) mod_x = torch.addcmul(shift_mlp, self.norm2(x.to(shift_mlp.dtype)), 1 + scale_mlp)
else: else:
mod_x = torch.addcmul(shift_mlp, self.norm2(x.view(B, -1, N//T, C).float()), 1 + scale_mlp).view(B, -1, C).to(input_dtype) mod_x = torch.addcmul(shift_mlp, self.norm2(x.view(B, -1, N//T, C).float()), 1 + scale_mlp).view(B, -1, C)
x_ffn = self.ffn(mod_x) x_ffn = self.ffn(mod_x.to(input_dtype))
del shift_mlp, scale_mlp del shift_mlp, scale_mlp
# gate_mlp # gate_mlp
@@ -1358,7 +1326,7 @@ class WanAttentionBlock(nn.Module):
x = x.add(x_ffn) x = x.add(x_ffn)
else: else:
if not is_longcat: if not is_longcat:
x = x.addcmul(x_ffn, gate_mlp) x = x.addcmul(x_ffn.to(gate_mlp.dtype), gate_mlp).to(input_dtype)
else: else:
x = x + (gate_mlp * x_ffn.view(B, -1, N//T, C).float()).to(input_dtype).view(B, -1, C) x = x + (gate_mlp * x_ffn.view(B, -1, N//T, C).float()).to(input_dtype).view(B, -1, C)
del gate_mlp del gate_mlp
@@ -1407,7 +1375,7 @@ class WanAttentionBlock(nn.Module):
segment_clip_embed = clip_embed[clip_idx:clip_idx+1] segment_clip_embed = clip_embed[clip_idx:clip_idx+1]
# Get tensor segment # Get tensor segment
x_segment = x[:, segment_indices, :] x_segment = x[:, segment_indices, :].to(self.norm3.weight.dtype)
# Process segment with its prompt and clip embedding # Process segment with its prompt and clip embedding
processed_segment = self.cross_attn(self.norm3(x_segment), segment_context, clip_embed=segment_clip_embed) processed_segment = self.cross_attn(self.norm3(x_segment), segment_context, clip_embed=segment_clip_embed)
@@ -1520,7 +1488,7 @@ class Head(nn.Module):
""" """
e = self.get_mod(e.to(x.device)) e = self.get_mod(e.to(x.device))
x = self.head(self.norm(x).mul_(1 + e[1]).add_(e[0])) x = self.head(self.norm(x.float()).to(x.dtype).mul_(1 + e[1]).add_(e[0]))
return x return x
class Head_adaLN(nn.Module): class Head_adaLN(nn.Module):
@@ -2844,6 +2812,9 @@ class WanModel(torch.nn.Module):
if unianim_data['start_percent'] <= current_step_percentage <= unianim_data['end_percent']: if unianim_data['start_percent'] <= current_step_percentage <= unianim_data['end_percent']:
dwpose_emb = rearrange(unianim_data['dwpose'], 'b c f h w -> b (f h w) c').contiguous() dwpose_emb = rearrange(unianim_data['dwpose'], 'b c f h w -> b (f h w) c').contiguous()
x.add_(dwpose_emb, alpha=unianim_data['strength']) x.add_(dwpose_emb, alpha=unianim_data['strength'])
seq_chunks = max(context.shape[0], clip_embed.shape[0] if clip_embed is not None else 0)
chunked_self_attention = seq_chunks > 1 and current_step in self.video_attention_split_steps
# arguments # arguments
kwargs = dict( kwargs = dict(
e=e0, e=e0,
@@ -2854,7 +2825,8 @@ class WanModel(torch.nn.Module):
clip_embed=clip_embed, clip_embed=clip_embed,
current_step=torch.tensor(current_step), current_step=torch.tensor(current_step),
last_step=torch.tensor(last_step, dtype=torch.bool), last_step=torch.tensor(last_step, dtype=torch.bool),
video_attention_split_steps=self.video_attention_split_steps, chunked_self_attention=chunked_self_attention,
seq_chunks=seq_chunks,
camera_embed=camera_embed, camera_embed=camera_embed,
audio_proj=audio_proj, audio_proj=audio_proj,
num_latent_frames = F, num_latent_frames = F,