import itertools import torch import torch.nn as nn import torch.nn.functional as F from models.arch_util import ResBlock from models.audio.music.music_quantizer2 import MusicQuantizer2 from models.diffusion.nn import timestep_embedding, normalization, zero_module, conv_nd, linear from models.diffusion.unet_diffusion import TimestepBlock from models.lucidrains.x_transformers import Encoder, Attention, RMSScaleShiftNorm, RotaryEmbedding, \ FeedForward from trainer.networks import register_model from utils.util import checkpoint, print_network def is_latent(t): return t.dtype == torch.float def is_sequence(t): return t.dtype == torch.long class MultiGroupEmbedding(nn.Module): def __init__(self, tokens, groups, dim): super().__init__() self.m = nn.ModuleList([nn.Embedding(tokens, dim // groups) for _ in range(groups)]) def forward(self, x): h = [embedding(x[:, :, i]) for i, embedding in enumerate(self.m)] return torch.cat(h, dim=-1) class TimestepRotaryEmbedSequential(nn.Sequential, TimestepBlock): def forward(self, x, emb, rotary_emb): for layer in self: if isinstance(layer, TimestepBlock): x = layer(x, emb, rotary_emb) else: x = layer(x, rotary_emb) return x class SubBlock(nn.Module): def __init__(self, inp_dim, contraction_dim, heads, dropout): super().__init__() self.attn = Attention(inp_dim, out_dim=contraction_dim, heads=heads, dim_head=contraction_dim//heads, causal=False, dropout=dropout) self.attnorm = nn.LayerNorm(contraction_dim) self.ff = nn.Conv1d(inp_dim+contraction_dim, contraction_dim, kernel_size=3, padding=1) self.ffnorm = nn.LayerNorm(contraction_dim) def forward(self, x, rotary_emb): ah, _, _, _ = checkpoint(self.attn, x, None, None, None, None, None, rotary_emb) ah = F.gelu(self.attnorm(ah)) h = torch.cat([ah, x], dim=-1) hf = checkpoint(self.ff, h.permute(0,2,1)).permute(0,2,1) hf = F.gelu(self.ffnorm(hf)) h = torch.cat([h, hf], dim=-1) return h class ConcatAttentionBlock(TimestepBlock): def __init__(self, trunk_dim, contraction_dim, time_embed_dim, heads, dropout): super().__init__() self.prenorm = RMSScaleShiftNorm(trunk_dim, embed_dim=time_embed_dim, bias=False) self.block1 = SubBlock(trunk_dim, contraction_dim, heads, dropout) self.block2 = SubBlock(trunk_dim+contraction_dim*2, contraction_dim, heads, dropout) self.out = nn.Linear(contraction_dim*4, trunk_dim, bias=False) self.out.weight.data.zero_() def forward(self, x, timestep_emb, rotary_emb): h = self.prenorm(x, norm_scale_shift_inp=timestep_emb) h = self.block1(h, rotary_emb) h = self.block2(h, rotary_emb) h = self.out(h[:,:,x.shape[-1]:]) return h + x class TransformerDiffusion(nn.Module): """ A diffusion model composed entirely of stacks of transformer layers. Why would you do it any other way? """ def __init__( self, prenet_channels=1024, prenet_layers=3, time_embed_dim=256, model_channels=1024, contraction_dim=256, num_layers=8, in_channels=256, rotary_emb_dim=32, input_vec_dim=1024, out_channels=512, # mean and variance num_heads=4, dropout=0, use_fp16=False, ar_prior=False, # Parameters for regularization. unconditioned_percentage=.1, # This implements a mechanism similar to what is used in classifier-free training. ): super().__init__() self.in_channels = in_channels self.model_channels = model_channels self.prenet_channels = prenet_channels self.time_embed_dim = time_embed_dim self.out_channels = out_channels self.dropout = dropout self.unconditioned_percentage = unconditioned_percentage self.enable_fp16 = use_fp16 self.inp_block = conv_nd(1, in_channels, prenet_channels, 3, 1, 1) self.time_embed = nn.Sequential( linear(time_embed_dim, time_embed_dim), nn.SiLU(), linear(time_embed_dim, time_embed_dim), ) self.ar_prior = ar_prior prenet_heads = prenet_channels//64 if ar_prior: self.ar_input = nn.Linear(input_vec_dim, prenet_channels) self.ar_prior_intg = Encoder( dim=prenet_channels, depth=prenet_layers, heads=prenet_heads, ff_dropout=dropout, attn_dropout=dropout, use_rmsnorm=True, ff_glu=True, rotary_pos_emb=True, zero_init_branch_output=True, ff_mult=1, ) else: self.input_converter = nn.Linear(input_vec_dim, prenet_channels) self.code_converter = Encoder( dim=prenet_channels, depth=prenet_layers, heads=prenet_heads, ff_dropout=dropout, attn_dropout=dropout, use_rmsnorm=True, ff_glu=True, rotary_pos_emb=True, zero_init_branch_output=True, ff_mult=1, ) self.unconditioned_embedding = nn.Parameter(torch.randn(1,1,prenet_channels)) self.rotary_embeddings = RotaryEmbedding(rotary_emb_dim) self.intg = nn.Linear(prenet_channels*2, model_channels) self.layers = TimestepRotaryEmbedSequential(*[ConcatAttentionBlock(model_channels, contraction_dim, time_embed_dim, num_heads, dropout) for _ in range(num_layers)]) self.out = nn.Sequential( normalization(model_channels), nn.SiLU(), zero_module(conv_nd(1, model_channels, out_channels, 3, padding=1)), ) self.debug_codes = {} def get_grad_norm_parameter_groups(self): groups = { 'layers': list(self.layers.parameters()) + list(self.inp_block.parameters()), 'code_converters': list(self.input_converter.parameters()) + list(self.code_converter.parameters()), 'time_embed': list(self.time_embed.parameters()), } return groups def timestep_independent(self, prior, expected_seq_len): code_emb = self.ar_input(prior) if self.ar_prior else self.input_converter(prior) code_emb = self.ar_prior_intg(code_emb) if self.ar_prior else self.code_converter(code_emb) # Mask out the conditioning branch for whole batch elements, implementing something similar to classifier-free guidance. if self.training and self.unconditioned_percentage > 0: unconditioned_batches = torch.rand((code_emb.shape[0], 1, 1), device=code_emb.device) < self.unconditioned_percentage code_emb = torch.where(unconditioned_batches, self.unconditioned_embedding.repeat(prior.shape[0], 1, 1), code_emb) expanded_code_emb = F.interpolate(code_emb.permute(0,2,1), size=expected_seq_len, mode='nearest').permute(0,2,1) return expanded_code_emb def forward(self, x, timesteps, codes=None, conditioning_input=None, precomputed_code_embeddings=None, conditioning_free=False): if precomputed_code_embeddings is not None: assert codes is None and conditioning_input is None, "Do not provide precomputed embeddings and the other parameters. It is unclear what you want me to do here." unused_params = [] if conditioning_free: code_emb = self.unconditioned_embedding.repeat(x.shape[0], x.shape[-1], 1) else: if precomputed_code_embeddings is not None: code_emb = precomputed_code_embeddings else: code_emb = self.timestep_independent(codes, x.shape[-1]) unused_params.append(self.unconditioned_embedding) with torch.autocast(x.device.type, enabled=self.enable_fp16): blk_emb = self.time_embed(timestep_embedding(timesteps, self.time_embed_dim)) x = self.inp_block(x).permute(0,2,1) rotary_pos_emb = self.rotary_embeddings(x.shape[1], x.device) x = self.intg(torch.cat([x, code_emb], dim=-1)) for layer in self.layers: x = checkpoint(layer, x, blk_emb, rotary_pos_emb) x = x.float().permute(0,2,1) out = self.out(x) # Involve probabilistic or possibly unused parameters in loss so we don't get DDP errors. extraneous_addition = 0 for p in unused_params: extraneous_addition = extraneous_addition + p.mean() out = out + extraneous_addition * 0 return out class TransformerDiffusionWithQuantizer(nn.Module): def __init__(self, quantizer_dims=[1024], quantizer_codebook_size=256, quantizer_codebook_groups=2, freeze_quantizer_until=20000, **kwargs): super().__init__() self.internal_step = 0 self.freeze_quantizer_until = freeze_quantizer_until self.diff = TransformerDiffusion(**kwargs) self.quantizer = MusicQuantizer2(inp_channels=kwargs['in_channels'], inner_dim=quantizer_dims, codevector_dim=quantizer_dims[0], codebook_size=quantizer_codebook_size, codebook_groups=quantizer_codebook_groups, max_gumbel_temperature=4, min_gumbel_temperature=.5) self.quantizer.quantizer.temperature = self.quantizer.min_gumbel_temperature del self.quantizer.up def update_for_step(self, step, *args): self.internal_step = step qstep = max(0, self.internal_step - self.freeze_quantizer_until) self.quantizer.quantizer.temperature = max( self.quantizer.max_gumbel_temperature * self.quantizer.gumbel_temperature_decay ** qstep, self.quantizer.min_gumbel_temperature, ) def forward(self, x, timesteps, truth_mel, conditioning_input=None, disable_diversity=False, conditioning_free=False): quant_grad_enabled = self.internal_step > self.freeze_quantizer_until with torch.set_grad_enabled(quant_grad_enabled): proj, diversity_loss = self.quantizer(truth_mel, return_decoder_latent=True) proj = proj.permute(0,2,1) # Make sure this does not cause issues in DDP by explicitly using the parameters for nothing. if not quant_grad_enabled: unused = 0 for p in self.quantizer.parameters(): unused = unused + p.mean() * 0 proj = proj + unused diversity_loss = diversity_loss * 0 diff = self.diff(x, timesteps, codes=proj, conditioning_input=conditioning_input, conditioning_free=conditioning_free) if disable_diversity: return diff return diff, diversity_loss def get_debug_values(self, step, __): if self.quantizer.total_codes > 0: return {'histogram_quant_codes': self.quantizer.codes[:self.quantizer.total_codes], 'gumbel_temperature': self.quantizer.quantizer.temperature} else: return {} def get_grad_norm_parameter_groups(self): attn1 = list(itertools.chain.from_iterable([lyr.block1.attn.parameters() for lyr in self.diff.layers])) attn2 = list(itertools.chain.from_iterable([lyr.block2.attn.parameters() for lyr in self.diff.layers])) ff1 = list(itertools.chain.from_iterable([lyr.block1.ff.parameters() for lyr in self.diff.layers])) ff2 = list(itertools.chain.from_iterable([lyr.block2.ff.parameters() for lyr in self.diff.layers])) blkout_layers = list(itertools.chain.from_iterable([lyr.out.parameters() for lyr in self.diff.layers])) groups = { 'prenorms': list(itertools.chain.from_iterable([lyr.prenorm.parameters() for lyr in self.diff.layers])), 'blk1_attention_layers': attn1, 'blk2_attention_layers': attn2, 'attention_layers': attn1 + attn2, 'blk1_ff_layers': ff1, 'blk2_ff_layers': ff2, 'ff_layers': ff1 + ff2, 'block_out_layers': blkout_layers, 'quantizer_encoder': list(self.quantizer.encoder.parameters()), 'quant_codebook': [self.quantizer.quantizer.codevectors], 'rotary_embeddings': list(self.diff.rotary_embeddings.parameters()), 'out': list(self.diff.out.parameters()), 'x_proj': list(self.diff.inp_block.parameters()), 'layers': list(self.diff.layers.parameters()), 'code_converters': list(self.diff.input_converter.parameters()) + list(self.diff.code_converter.parameters()), 'time_embed': list(self.diff.time_embed.parameters()), } return groups def before_step(self, step): scaled_grad_parameters = list(itertools.chain.from_iterable([lyr.out.parameters() for lyr in self.diff.layers])) + \ list(itertools.chain.from_iterable([lyr.prenorm.parameters() for lyr in self.diff.layers])) # Scale back the gradients of the blkout and prenorm layers by a constant factor. These get two orders of magnitudes # higher gradients. Ideally we would use parameter groups, but ZeroRedundancyOptimizer makes this trickier than # directly fiddling with the gradients. for p in scaled_grad_parameters: p.grad *= .2 class TransformerDiffusionWithARPrior(nn.Module): def __init__(self, freeze_diff=False, **kwargs): super().__init__() self.internal_step = 0 from models.audio.music.gpt_music import GptMusicLower self.ar = GptMusicLower(dim=512, layers=12) for p in self.ar.parameters(): p.DO_NOT_TRAIN = True p.requires_grad = False self.diff = TransformerDiffusion(ar_prior=True, **kwargs) if freeze_diff: for p in self.diff.parameters(): p.DO_NOT_TRAIN = True p.requires_grad = False for p in list(self.diff.ar_prior_intg.parameters()) + list(self.diff.ar_input.parameters()): del p.DO_NOT_TRAIN p.requires_grad = True def get_grad_norm_parameter_groups(self): groups = { 'attention_layers': list(itertools.chain.from_iterable([lyr.attn.parameters() for lyr in self.diff.layers])), 'ff_layers': list(itertools.chain.from_iterable([lyr.ff.parameters() for lyr in self.diff.layers])), 'rotary_embeddings': list(self.diff.rotary_embeddings.parameters()), 'out': list(self.diff.out.parameters()), 'x_proj': list(self.diff.inp_block.parameters()), 'layers': list(self.diff.layers.parameters()), 'ar_prior_intg': list(self.diff.ar_prior_intg.parameters()), 'time_embed': list(self.diff.time_embed.parameters()), } return groups def forward(self, x, timesteps, truth_mel, disable_diversity=False, conditioning_input=None, conditioning_free=False): with torch.no_grad(): prior = self.ar(truth_mel, conditioning_input, return_latent=True) diff = self.diff(x, timesteps, prior, conditioning_free=conditioning_free) return diff @register_model def register_transformer_diffusion12(opt_net, opt): return TransformerDiffusion(**opt_net['kwargs']) @register_model def register_transformer_diffusion12_with_quantizer(opt_net, opt): return TransformerDiffusionWithQuantizer(**opt_net['kwargs']) @register_model def register_transformer_diffusion12_with_ar_prior(opt_net, opt): return TransformerDiffusionWithARPrior(**opt_net['kwargs']) def test_quant_model(): clip = torch.randn(2, 256, 400) ts = torch.LongTensor([600, 600]) # For music: model = TransformerDiffusionWithQuantizer(in_channels=256, model_channels=1536, contraction_dim=768, prenet_channels=1024, num_heads=10, input_vec_dim=1024, num_layers=24, prenet_layers=4, dropout=.1) quant_weights = torch.load('D:\\dlas\\experiments\\train_music_quant_r4\\models\\5000_generator.pth') model.quantizer.load_state_dict(quant_weights, strict=False) torch.save(model.state_dict(), 'sample.pth') print_network(model) o = model(clip, ts, clip) pg = model.get_grad_norm_parameter_groups() t = 0 for k, vs in pg.items(): s = 0 for v in vs: m = 1 for d in v.shape: m *= d s += m t += s print(k, s/1000000) print(t) def test_ar_model(): clip = torch.randn(2, 256, 400) cond = torch.randn(2, 256, 400) ts = torch.LongTensor([600, 600]) model = TransformerDiffusionWithARPrior(model_channels=2048, prenet_channels=1536, input_vec_dim=512, num_layers=16, prenet_layers=6, freeze_diff=True, unconditioned_percentage=.4) model.get_grad_norm_parameter_groups() ar_weights = torch.load('D:\\dlas\\experiments\\train_music_gpt\\models\\44500_generator_ema.pth') model.ar.load_state_dict(ar_weights, strict=True) diff_weights = torch.load('X:\\dlas\\experiments\\train_music_diffusion_tfd8\\models\\47500_generator_ema.pth') pruned_diff_weights = {} for k,v in diff_weights.items(): if k.startswith('diff.'): pruned_diff_weights[k.replace('diff.', '')] = v model.diff.load_state_dict(pruned_diff_weights, strict=False) torch.save(model.state_dict(), 'sample.pth') model(clip, ts, cond, conditioning_input=cond) if __name__ == '__main__': test_quant_model()