File size: 16,418 Bytes
fb0011a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 | # Modules/diffusion/dit1d.py
# DiT1d z cross-attention, full multi-head K/V, AdaLNZero i opcjonalnym RoPE.
from typing import Optional
import math
import torch
import torch.nn as nn
from einops import rearrange, repeat
from torch import Tensor, einsum
# --- Helpers ----------------------------------------------------------------
def exists(x):
return x is not None
def default(val, d):
return val if exists(val) else (d() if callable(d) else d)
def rand_bool(shape, proba, device=None):
if proba == 1:
return torch.ones(shape, device=device, dtype=torch.bool)
if proba == 0:
return torch.zeros(shape, device=device, dtype=torch.bool)
return torch.bernoulli(torch.full(shape, proba, device=device)).to(torch.bool)
# --- Time / Fixed embeddings ------------------------------------------------
class LearnedPositionalEmbedding(nn.Module):
"""Continuous learned positional embedding for time (Fourier features)."""
def __init__(self, dim: int):
super().__init__()
assert (dim % 2) == 0, "dim must be even"
half_dim = dim // 2
self.weights = nn.Parameter(torch.randn(half_dim))
def forward(self, x: Tensor) -> Tensor:
# x: [B] (per-batch scalar time)
x = rearrange(x, "b -> b 1")
freqs = x * rearrange(self.weights, "d -> 1 d") * 2 * math.pi
fouriered = torch.cat((freqs.sin(), freqs.cos()), dim=-1)
fouriered = torch.cat((x, fouriered), dim=-1)
return fouriered
def TimePositionalEmbedding(dim: int, out_features: int) -> nn.Module:
return nn.Sequential(
LearnedPositionalEmbedding(dim),
nn.Linear(in_features=dim + 1, out_features=out_features),
)
class FixedEmbedding(nn.Module):
"""Fixed learned positional embeddings (used as fallback/masked embedding)."""
def __init__(self, max_length: int, features: int):
super().__init__()
self.max_length = max_length
self.embedding = nn.Embedding(max_length, features)
def forward(self, x: Tensor) -> Tensor:
# x: [B, L, E] (we only need L)
batch_size, length, device = x.shape[0], x.shape[1], x.device
assert length <= self.max_length, "Input sequence length must be <= max_length"
pos = torch.arange(length, device=device)
pe = self.embedding(pos) # [L, E]
pe = repeat(pe, "n d -> b n d", b=batch_size)
return pe
# --- AdaLNZero ---------------------------------------------------------------
class AdaLNZero(nn.Module):
"""
Adaptive LayerNorm with two separate linears producing gamma and beta.
x: [B, T, D], cond: [B, Dc]
"""
def __init__(self, dim: int, cond_dim: int):
super().__init__()
self.norm = nn.LayerNorm(dim)
self.gamma = nn.Linear(cond_dim, dim, bias=True)
self.beta = nn.Linear(cond_dim, dim, bias=True)
nn.init.zeros_(self.gamma.weight)
nn.init.zeros_(self.gamma.bias)
nn.init.zeros_(self.beta.weight)
nn.init.zeros_(self.beta.bias)
def forward(self, x: Tensor, c: Tensor) -> Tensor:
x = self.norm(x)
g = self.gamma(c).unsqueeze(1) # [B, 1, D]
b = self.beta(c).unsqueeze(1)
return x * (1 + g) + b
# --- Rotary embeddings (RoPE) ------------------------------------------------
class RotaryEmbedding(nn.Module):
"""
Rotary positional embeddings helper. Stores inverse freqs buffer.
"""
def __init__(self, dim: int, base: int = 10000):
super().__init__()
assert dim % 2 == 0, "rotary dim must be even"
self.dim = dim
inv_freq = 1.0 / (base ** (torch.arange(0, dim, 2).float() / dim))
self.register_buffer("inv_freq", inv_freq)
def get_sin_cos(self, seq_len: int, device: torch.device, dtype: torch.dtype):
t = torch.arange(seq_len, device=device, dtype=self.inv_freq.dtype)
sinusoid_inp = torch.einsum("i,j->ij", t, self.inv_freq) # [seq_len, dim//2]
sin = torch.sin(sinusoid_inp).to(dtype=dtype, device=device)
cos = torch.cos(sinusoid_inp).to(dtype=dtype, device=device)
return sin, cos
def apply_rotary_pos_emb(x: Tensor, sin: Tensor, cos: Tensor) -> Tensor:
"""
Apply rotary embedding to x.
x: [B, H, N, D], sin/cos: [N, D//2]
returns x with same shape
"""
b, h, n, d = x.shape
assert d % 2 == 0
# sin, cos expected shape [N, D//2], expand to [1,1,N,D//2]
sin = sin[None, None, :, :].to(dtype=x.dtype, device=x.device)
cos = cos[None, None, :, :].to(dtype=x.dtype, device=x.device)
x_even = x[..., ::2]
x_odd = x[..., 1::2]
x_rot_even = x_even * cos - x_odd * sin
x_rot_odd = x_even * sin + x_odd * cos
x_rot = torch.stack((x_rot_even, x_rot_odd), dim=-1).reshape(b, h, n, d)
return x_rot
# --- Feed-forward & Attention base -----------------------------------------
def FeedForward(features: int, multiplier: int) -> nn.Module:
mid = features * multiplier
return nn.Sequential(
nn.Linear(in_features=features, out_features=mid),
nn.GELU(),
nn.Linear(in_features=mid, out_features=features),
)
class AttentionBase(nn.Module):
"""
Core attention (multi-head), optionally applies RoPE to q/k.
q/k/v expected pre-projected to shape [B, N, num_heads*head_features]
"""
def __init__(
self,
num_heads: int,
head_features: int,
out_features: Optional[int] = None,
use_rope: bool = True,
):
super().__init__()
self.num_heads = num_heads
self.head_features = head_features
self.scale = head_features ** -0.5
self.use_rope = use_rope
if use_rope:
assert head_features % 2 == 0, "head_features must be even for RoPE"
self.rotary = RotaryEmbedding(head_features)
mid = num_heads * head_features
if out_features is None:
out_features = mid
self.to_out = nn.Linear(in_features=mid, out_features=out_features)
def forward(self, q: Tensor, k: Tensor, v: Tensor) -> Tensor:
# shapes: q: [B, Nq, mid], k: [B, Nk, mid], v: [B, Nk, mid]
q = rearrange(q, "b n (h d) -> b h n d", h=self.num_heads)
k = rearrange(k, "b m (h d) -> b h m d", h=self.num_heads)
v = rearrange(v, "b m (h d) -> b h m d", h=self.num_heads)
if self.use_rope:
sin_q, cos_q = self.rotary.get_sin_cos(q.shape[2], device=q.device, dtype=q.dtype)
sin_k, cos_k = self.rotary.get_sin_cos(k.shape[2], device=k.device, dtype=k.dtype)
q = apply_rotary_pos_emb(q, sin_q, cos_q)
k = apply_rotary_pos_emb(k, sin_k, cos_k)
sim = einsum("b h n d, b h m d -> b h n m", q, k)
sim = sim * self.scale
attn = sim.softmax(dim=-1)
out = einsum("b h n m, b h m d -> b h n d", attn, v)
out = rearrange(out, "b h n d -> b n (h d)")
return self.to_out(out)
class Attention(nn.Module):
def __init__(
self,
features: int,
*,
head_features: int,
num_heads: int,
out_features: Optional[int] = None,
context_features: Optional[int] = None,
use_rope: bool = True,
):
super().__init__()
self.num_heads = num_heads
self.head_features = head_features
mid = head_features * num_heads
context_features = default(context_features, features)
self.norm = nn.Identity()
self.norm_context = nn.LayerNorm(context_features)
self.to_q = nn.Linear(in_features=features, out_features=mid, bias=False)
self.to_kv = nn.Linear(in_features=context_features, out_features=mid * 2, bias=False)
# -> WAŻNA ZMIANA: wymuszamy out_features == features
self.base = AttentionBase(
num_heads=num_heads,
head_features=head_features,
out_features=features,
use_rope=use_rope,
)
def forward(self, x: Tensor, *, context: Optional[Tensor] = None) -> Tensor:
context = default(context, x)
# Oczekujemy, że `x` zostało już znormalizowane przez AdaLNZero w
# DiTBlock. Nie re-normalizujemy queries, żeby nie tracić efektu
# modulacji (gamma/beta).
q_in = x # no extra LayerNorm on queries
# Dla kontekstu: normalizujemy tylko wtedy, gdy jest to inny tensor niż
# queries (czyli cross-attention). Jeśli context domyślnie == x (self-attn),
# to nie chcemy dodatkowej normalizacji.
if context is x:
context_norm = context
else:
context_norm = self.norm_context(context)
q = self.to_q(q_in)
k, v = torch.chunk(self.to_kv(context_norm), chunks=2, dim=-1)
out = self.base(q, k, v)
return out
# --- DiT blocks -------------------------------------------------------------
class DiTBlock(nn.Module):
"""
DiT-style block: pre-normalization via AdaLNZero before attention and FFN.
Supports self-attention + optional cross-attention (context).
"""
def __init__(
self,
features: int,
num_heads: int,
head_features: int,
multiplier: int,
cond_dim: int,
context_features: Optional[int] = None,
use_rope: bool = True,
):
super().__init__()
self.use_cross_attention = exists(context_features) and context_features > 0
# adaptive layer norms
self.attn_mod = AdaLNZero(dim=features, cond_dim=cond_dim)
self.ffn_mod = AdaLNZero(dim=features, cond_dim=cond_dim)
# self-attention and optional cross-attention (full-KV)
self.self_attn = Attention(
features=features,
head_features=head_features,
num_heads=num_heads,
use_rope=use_rope,
)
if self.use_cross_attention:
self.cross_attn = Attention(
features=features,
head_features=head_features,
num_heads=num_heads,
context_features=context_features,
use_rope=use_rope,
)
self.feed_forward = FeedForward(features=features, multiplier=multiplier)
def forward(self, x: Tensor, cond: Tensor, context: Optional[Tensor] = None) -> Tensor:
# Self-attention (pre-norm & conditioning)
attn_in = self.attn_mod(x, cond)
x = self.self_attn(attn_in) + x
# Cross-attention (if context provided)
if self.use_cross_attention and exists(context):
cross_in = self.attn_mod(x, cond)
x = self.cross_attn(cross_in, context=context) + x
# Feed-forward
ffn_in = self.ffn_mod(x, cond)
x = self.feed_forward(ffn_in) + x
return x
# --- Main DiT1d model ------------------------------------------------------
class _BaseDiT1d(nn.Module):
def __init__(
self,
num_layers: int,
channels: int,
num_heads: int,
head_features: int,
multiplier: int,
context_embedding_features: int,
embedding_max_length: int = 512,
use_style_conditioning: bool = False,
style_features: Optional[int] = None,
use_rope: bool = True,
):
super().__init__()
assert exists(context_embedding_features), "context_embedding_features must be provided"
self.use_style_conditioning = use_style_conditioning
token_dim = channels # token dimensionality (we use cross-attention to context)
cond_dim = token_dim
# time positional -> cond_dim
self.to_time = nn.Sequential(TimePositionalEmbedding(dim=channels, out_features=cond_dim), nn.GELU())
if use_style_conditioning:
assert exists(style_features), "style_features must be provided when use_style_conditioning=True"
self.to_style = nn.Sequential(nn.Linear(in_features=style_features, out_features=cond_dim), nn.GELU())
self.mapping = nn.Sequential(
nn.Linear(cond_dim, cond_dim),
nn.GELU(),
nn.Linear(cond_dim, cond_dim),
nn.GELU(),
)
# build blocks with cross-attention to embeddings
self.blocks = nn.ModuleList(
[
DiTBlock(
features=token_dim,
num_heads=num_heads,
head_features=head_features,
multiplier=multiplier,
cond_dim=cond_dim,
context_features=context_embedding_features,
use_rope=use_rope,
)
for _ in range(num_layers)
]
)
self.fixed_embedding = FixedEmbedding(max_length=embedding_max_length, features=context_embedding_features)
self.out_norm = nn.LayerNorm(token_dim)
self.to_out = nn.Linear(in_features=token_dim, out_features=channels)
def get_conditioning(self, time: Tensor, features: Optional[Tensor]):
cond = self.to_time(time)
if self.use_style_conditioning:
assert exists(features), "style features must be provided"
cond = cond + self.to_style(features)
return self.mapping(cond)
def run(self, x: Tensor, time: Tensor, embedding: Tensor, features: Optional[Tensor]):
# x: [B, 1, C], embedding: [B, L, E]
L = embedding.shape[1]
x_tokens = x.expand(-1, L, -1) # [B, L, C] - queries
tokens = x_tokens # we use cross-attention to conditioning embeddings
cond = self.get_conditioning(time, features) # [B, cond_dim]
for block in self.blocks:
tokens = block(tokens, cond=cond, context=embedding)
tokens = self.out_norm(tokens)
tokens = tokens.mean(dim=1) # avg pool over seq length -> [B, C]
out = self.to_out(tokens) # [B, C]
return out.unsqueeze(1) # [B, 1, C]
def forward(
self,
x: Tensor,
time: Tensor,
embedding_mask_proba: float = 0.0,
embedding: Optional[Tensor] = None,
features: Optional[Tensor] = None,
embedding_scale: float = 1.0,
) -> Tensor:
b, device = x.shape[0], x.device
assert exists(embedding), "context embedding must be provided"
fixed_embedding = self.fixed_embedding(embedding)
if embedding_mask_proba > 0.0:
batch_mask = rand_bool((b, 1, 1), proba=embedding_mask_proba, device=device)
embedding = torch.where(batch_mask, fixed_embedding, embedding)
if embedding_scale != 1.0:
out = self.run(x, time, embedding=embedding, features=features)
out_masked = self.run(x, time, embedding=fixed_embedding, features=features)
return out_masked + (out - out_masked) * embedding_scale
else:
return self.run(x, time, embedding=embedding, features=features)
# --- Public wrapper --------------------------------------------------------
class Transformer1d(_BaseDiT1d):
"""API-compatible DiT1d (single unified version with cross-attention & RoPE)."""
def __init__(
self,
num_layers: int,
channels: int,
num_heads: int,
head_features: int,
multiplier: int,
context_embedding_features: Optional[int] = None,
embedding_max_length: int = 512,
use_rope: bool = True,
**kwargs,
):
assert exists(context_embedding_features)
super().__init__(
num_layers=num_layers,
channels=channels,
num_heads=num_heads,
head_features=head_features,
multiplier=multiplier,
context_embedding_features=context_embedding_features,
embedding_max_length=embedding_max_length,
use_style_conditioning=False,
style_features=None,
use_rope=use_rope,
)
|