|
|
|
|
|
import pyworld as pw |
|
|
import os |
|
|
import math, random |
|
|
import warnings |
|
|
import logging |
|
|
import gzip |
|
|
import base64 |
|
|
import torch |
|
|
import torchaudio |
|
|
import torch.nn.functional as F |
|
|
import torch.nn.init as init |
|
|
from torch import nn, Tensor |
|
|
import numpy as np |
|
|
from typing import Optional, Dict, Union, List, Tuple, Any |
|
|
from functools import partial |
|
|
from datetime import datetime |
|
|
from datasets import load_dataset, Audio, concatenate_datasets |
|
|
from transformers.trainer_seq2seq import Seq2SeqTrainer |
|
|
from transformers.training_args_seq2seq import Seq2SeqTrainingArguments |
|
|
import transformers |
|
|
import evaluate |
|
|
from dataclasses import dataclass |
|
|
import matplotlib.pyplot as plt |
|
|
|
|
|
device = torch.device(device="cuda:0") |
|
|
dtype = torch.float32 |
|
|
|
|
|
extractor = None |
|
|
tokenizer = None |
|
|
optimizer = None |
|
|
scheduler = None |
|
|
model = None |
|
|
Residual = None |
|
|
MultiheadA = None |
|
|
|
|
|
@dataclass |
|
|
class Dimensions: |
|
|
vocab: int |
|
|
text_ctx: int |
|
|
text_dims: int |
|
|
text_head: int |
|
|
text_idx: int |
|
|
mels: int |
|
|
aud_ctx: int |
|
|
aud_dims: int |
|
|
aud_head: int |
|
|
aud_idx: int |
|
|
act: str |
|
|
debug: List[str] |
|
|
cross_attn: bool |
|
|
features: List[str] |
|
|
f0_rotary: bool |
|
|
|
|
|
def plot_waveform(x=None, w=None, p=None, per=None, sample_idx=0, sr=16000, hop_length=160, |
|
|
title="", markers=None, marker_labels=None, |
|
|
show_voiced_regions=True, show_energy=False): |
|
|
num_plots = sum([x is not None, w is not None, p is not None, per is not None]) |
|
|
if num_plots == 0: |
|
|
raise ValueError("No data to plot. Please provide at least one input tensor.") |
|
|
time_spans = [] |
|
|
|
|
|
if w is not None: |
|
|
w_np = w[sample_idx].detach().cpu().numpy() |
|
|
if w_np.ndim > 1: |
|
|
w_np = w_np.squeeze() |
|
|
time_spans.append(len(w_np) / sr) |
|
|
if x is not None: |
|
|
x_np = x[sample_idx].detach().cpu().numpy() |
|
|
if x_np.shape[0] < x_np.shape[1]: |
|
|
x_np = x_np.T |
|
|
time_spans.append(x_np.shape[0] * hop_length / sr) |
|
|
if p is not None: |
|
|
p_np = p[sample_idx].detach().cpu().numpy() |
|
|
if p_np.ndim > 1: |
|
|
p_np = p_np.squeeze() |
|
|
time_spans.append(len(p_np) * hop_length / sr) |
|
|
if per is not None: |
|
|
per_np = per[sample_idx].detach().cpu().numpy() |
|
|
if per_np.ndim > 1: |
|
|
per_np = per_np.squeeze() |
|
|
time_spans.append(len(per_np) * hop_length / sr) |
|
|
max_time = max(time_spans) if time_spans else 0 |
|
|
fig, axs = plt.subplots(num_plots, 1, figsize=(14, 4*num_plots), sharex=True) |
|
|
if num_plots == 1: |
|
|
axs = [axs] |
|
|
if show_voiced_regions and per is not None: |
|
|
per_np = per[sample_idx].detach().cpu().numpy() |
|
|
if per_np.ndim > 1: |
|
|
per_np = per_np.squeeze() |
|
|
t_per = np.arange(len(per_np)) * hop_length / sr |
|
|
threshold = 0.5 |
|
|
for ax in axs: |
|
|
for i in range(len(per_np)-1): |
|
|
if per_np[i] > threshold: |
|
|
ax.axvspan(t_per[i], t_per[i+1], color='lightblue', alpha=0.2, zorder=0) |
|
|
current_ax = 0 |
|
|
if w is not None: |
|
|
w_np = w[sample_idx].detach().cpu().numpy() |
|
|
if w_np.ndim > 1: |
|
|
w_np = w_np.squeeze() |
|
|
t = np.arange(len(w_np)) / sr |
|
|
axs[current_ax].plot(t, w_np, color="tab:blue") |
|
|
|
|
|
if show_energy: |
|
|
frame_length = hop_length |
|
|
hop_length_energy = hop_length // 2 |
|
|
energy = [] |
|
|
for i in range(0, len(w_np)-frame_length, hop_length_energy): |
|
|
frame = w_np[i:i+frame_length] |
|
|
energy.append(np.sqrt(np.mean(frame**2))) |
|
|
energy = np.array(energy) |
|
|
energy = energy / np.max(energy) * 0.8 * max(abs(w_np.min()), abs(w_np.max())) |
|
|
t_energy = np.arange(len(energy)) * hop_length_energy / sr |
|
|
axs[current_ax].plot(t_energy, energy, color="red", alpha=0.7, label="Energy") |
|
|
axs[current_ax].legend(loc='upper right') |
|
|
axs[current_ax].set_title("Waveform") |
|
|
axs[current_ax].set_ylabel("Amplitude") |
|
|
axs[current_ax].set_xlim([0, max_time]) |
|
|
axs[current_ax].grid(True, axis='x', linestyle='--', alpha=0.3) |
|
|
current_ax += 1 |
|
|
|
|
|
if x is not None: |
|
|
x_np = x[sample_idx].detach().cpu().numpy() |
|
|
if x_np.shape[0] < x_np.shape[1]: |
|
|
x_np = x_np.T |
|
|
im = axs[current_ax].imshow(x_np.T, aspect="auto", origin="lower", cmap="magma", |
|
|
extent=[0, x_np.shape[0]*hop_length/sr, 0, x_np.shape[1]]) |
|
|
axs[current_ax].set_title("Spectrogram") |
|
|
axs[current_ax].set_ylabel("Mel Bin") |
|
|
axs[current_ax].set_xlim([0, max_time]) |
|
|
axs[current_ax].grid(True, axis='x', linestyle='--', alpha=0.3) |
|
|
current_ax += 1 |
|
|
|
|
|
if p is not None: |
|
|
p_np = p[sample_idx].detach().cpu().numpy() |
|
|
if p_np.ndim > 1: |
|
|
p_np = p_np.squeeze() |
|
|
t_p = np.arange(len(p_np)) * hop_length / sr |
|
|
axs[current_ax].plot(t_p, p_np, color="tab:green") |
|
|
axs[current_ax].set_title("Pitch") |
|
|
axs[current_ax].set_ylabel("Frequency (Hz)") |
|
|
axs[current_ax].set_xlim([0, max_time]) |
|
|
axs[current_ax].grid(True, axis='both', linestyle='--', alpha=0.3) |
|
|
axs[current_ax].set_ylim([0, min(1000, p_np.max() * 1.2)]) |
|
|
current_ax += 1 |
|
|
|
|
|
if per is not None: |
|
|
per_np = per[sample_idx].detach().cpu().numpy() |
|
|
if per_np.ndim > 1: |
|
|
per_np = per_np.squeeze() |
|
|
t_per = np.arange(len(per_np)) * hop_length / sr |
|
|
axs[current_ax].plot(t_per, per_np, color="tab:red") |
|
|
axs[current_ax].set_title("Period (Voice Activity)") |
|
|
axs[current_ax].set_ylabel("periodocity") |
|
|
axs[current_ax].set_xlim([0, max_time]) |
|
|
axs[current_ax].grid(True, axis='both', linestyle='--', alpha=0.3) |
|
|
axs[current_ax].set_ylim([-0.05, 1.05]) |
|
|
axs[current_ax].axhline(y=0.5, color='k', linestyle='--', alpha=0.3) |
|
|
|
|
|
if markers is not None: |
|
|
for i, t in enumerate(markers): |
|
|
label = marker_labels[i] if marker_labels and i < len(marker_labels) else None |
|
|
for ax in axs: |
|
|
ax.axvline(x=t, color='k', linestyle='-', alpha=0.7, label=label if i == 0 else None) |
|
|
if marker_labels: |
|
|
axs[0].legend(loc='upper right', fontsize='small') |
|
|
axs[-1].set_xlabel("Time (s)") |
|
|
fig.suptitle(title, fontsize=16) |
|
|
plt.tight_layout(rect=[0, 0, 1, 0.97]) |
|
|
plt.show() |
|
|
return fig |
|
|
|
|
|
def exists(v): |
|
|
return v is not None |
|
|
|
|
|
def default(v, b): |
|
|
return v if exists(v) else b |
|
|
|
|
|
class Conv1d(nn.Conv1d): |
|
|
def _conv_forward( |
|
|
self, x: Tensor, weight: Tensor, bias) -> Tensor: |
|
|
return super()._conv_forward(x, weight.to(x.device, x.dtype), None if bias is None else bias.to(x.device, x.dtype)) |
|
|
|
|
|
class Conv2d(nn.Conv2d): |
|
|
def _conv_forward( |
|
|
self, x: Tensor, weight: Tensor, bias) -> Tensor: |
|
|
return super()._conv_forward(x, weight.to(x.device, x.dtype), None if bias is None else bias.to(x.device, x.dtype)) |
|
|
|
|
|
class Linear(nn.Module): |
|
|
def __init__(self, in_features: int, out_features: int, bias: bool = True) -> None: |
|
|
super(Linear, self).__init__() |
|
|
self.linear = nn.Linear(in_features, out_features, bias=bias) |
|
|
init.xavier_uniform_(self.linear.weight) |
|
|
if bias: |
|
|
init.zeros_(self.linear.bias) |
|
|
def forward(self, x: Tensor) -> Tensor: |
|
|
return self.linear(x) |
|
|
|
|
|
class RMSNorm(nn.Module): |
|
|
def __init__(self, dims: Union[int, Tensor, List, Tuple], |
|
|
eps = 1e-8, elementwise_affine = True): |
|
|
super(RMSNorm, self).__init__() |
|
|
if isinstance(dims, int): |
|
|
self.normalized_shape = (dims,) |
|
|
else: |
|
|
self.normalized_shape = tuple(dims) |
|
|
self.eps = eps |
|
|
self.elementwise_affine = elementwise_affine |
|
|
if self.elementwise_affine: |
|
|
self.weight = nn.Parameter(torch.empty(self.normalized_shape)) |
|
|
init.ones_(self.weight) |
|
|
else: |
|
|
self.register_parameter("weight", None) |
|
|
def forward(self, x): |
|
|
return F.rms_norm(x, self.normalized_shape, self.weight, self.eps) |
|
|
|
|
|
def LayerNorm(x: Tensor, normalized_shape: Union[int, Tensor, List, Tuple], |
|
|
weight: Optional[Tensor] = None, bias: Optional[Tensor] = None, |
|
|
eps: float = 1e-5) -> Tensor: |
|
|
return F.layer_norm(x, normalized_shape, weight, bias, eps) |
|
|
|
|
|
def get_device(): |
|
|
return torch.device("cuda:0" if torch.cuda.is_available() else "cpu") |
|
|
|
|
|
def get_dtype(): |
|
|
return torch.float32 if torch.cuda.is_available() else torch.float64 |
|
|
|
|
|
def get_tox(): |
|
|
return {"device": get_device(), "dtype": get_dtype()} |
|
|
|
|
|
def sinusoids(length, channels, max_timescale=10000): |
|
|
"""Returns sinusoids for positional embedding""" |
|
|
assert channels % 2 == 0 |
|
|
log_timescale_increment = np.log(max_timescale) / (channels // 2 - 1) |
|
|
inv_timescales = torch.exp(-log_timescale_increment * torch.arange(channels // 2)) |
|
|
scaled_time = torch.arange(length)[:, np.newaxis] * inv_timescales[np.newaxis, :] |
|
|
return torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], dim=1) |
|
|
|
|
|
class ParameterCycler: |
|
|
def __init__(self, parameters): |
|
|
self.parameters = parameters |
|
|
self.current_idx = 0 |
|
|
def toggle_requires_grad(self): |
|
|
x = random.randint(0, len(self.parameters) - 1) |
|
|
for x, param in enumerate(self.parameters): |
|
|
param.requires_grad = (x == self.current_idx) |
|
|
print(f"Parameter {x}: requires_grad={param.requires_grad}") |
|
|
self.current_idx = (self.current_idx + 1) % len(self.parameters) |
|
|
|
|
|
def extract_f0(waveform, sampling_rate=16000, hop_length=128, device="cuda:0"): |
|
|
"""Extract F0 from waveform - handle various input types""" |
|
|
if waveform is None: |
|
|
return None |
|
|
|
|
|
if isinstance(waveform, list): |
|
|
if len(waveform) == 0: |
|
|
return None |
|
|
waveform = waveform[0] |
|
|
print(f"DEBUG: Converted list to tensor, new type: {type(waveform)}") |
|
|
|
|
|
if not isinstance(waveform, torch.Tensor): |
|
|
waveform = torch.tensor(waveform) |
|
|
|
|
|
if isinstance(waveform, torch.Tensor): |
|
|
if waveform.dim() == 3: |
|
|
waveform = waveform.squeeze(1) |
|
|
if waveform.dim() == 2: |
|
|
waveform = waveform[0] |
|
|
|
|
|
wav_np = waveform.detach().cpu().numpy().astype(np.float64) |
|
|
else: |
|
|
wav_np = np.array(waveform).astype(np.float64) |
|
|
|
|
|
f0, t = pw.dio(wav_np, sampling_rate, |
|
|
frame_period=hop_length/sampling_rate*1000) |
|
|
f0 = pw.stonemask(wav_np, f0, t, sampling_rate) |
|
|
|
|
|
f0_tensor = torch.from_numpy(f0).float().to(device) |
|
|
return f0_tensor.unsqueeze(0).unsqueeze(0) |
|
|
|
|
|
class rotary(nn.Module): |
|
|
_seen = set() |
|
|
def __init__(self, dims, max_ctx=1500, theta=10000, learned_freq=False, radii=False, |
|
|
learned_radius=False, learned_theta=False, learned_pitch=False, debug: List[str] = [], use_pbias = False): |
|
|
super().__init__() |
|
|
|
|
|
self.use_pbias = use_pbias |
|
|
self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') |
|
|
self.dtype = torch.float32 |
|
|
self.debug = debug |
|
|
self._counter = 0 |
|
|
self.dims = dims |
|
|
self.max_ctx = max_ctx |
|
|
self.radii = radii |
|
|
f0_factor = 0.5 |
|
|
self.learned_adaptation: bool = False |
|
|
pitch_scale = 1.0 |
|
|
radius = 1 |
|
|
|
|
|
if self.learned_adaptation: |
|
|
self.f0_scale = nn.Parameter(torch.tensor(f0_factor, device=self.device, dtype=self.dtype), requires_grad=True) |
|
|
else: |
|
|
self.register_buffer('f0_scale', torch.tensor(f0_factor)) |
|
|
|
|
|
self.theta = nn.Parameter(torch.tensor(theta, device=self.device, dtype=self.dtype), requires_grad=True) |
|
|
self.pitch_scale = nn.Parameter(torch.tensor(pitch_scale, device=self.device, dtype=self.dtype), requires_grad=True) |
|
|
freqs = 1. / (theta ** (torch.arange(0, dims, 2, device=self.device, dtype=self.dtype)[:(dims // 2)].float() / dims)) |
|
|
self.freqs = nn.Parameter(torch.tensor(freqs, device=self.device, dtype=self.dtype), requires_grad=True) |
|
|
self.radius = nn.Parameter(torch.ones(radius, device=self.device, dtype=self.dtype), requires_grad=True) |
|
|
|
|
|
def forward(self, x=None, layer=None, enc=None) -> Tensor: |
|
|
|
|
|
f0 = enc.get("f0") if enc else None |
|
|
if isinstance(x, int): |
|
|
ctx = x |
|
|
else: |
|
|
batch, ctx, dims = x.shape |
|
|
t = torch.arange(ctx, device=self.device).float() |
|
|
|
|
|
if f0 is not None: |
|
|
f0_mean=f0.mean()+1e-8 |
|
|
theta=f0_mean*self.pitch_scale |
|
|
freqs = 1. / (theta ** (torch.arange(0, self.dims, 2, device=self.device, dtype=self.dtype)[:(self.dims // 2)].float() /self.dims)) |
|
|
else: |
|
|
freqs = self.freqs |
|
|
|
|
|
freqs = torch.einsum('i,j->ij', t, freqs) |
|
|
freqs = freqs.float() |
|
|
|
|
|
if self.radii: |
|
|
|
|
|
radius = enc.get("f0d") if enc else self.radius |
|
|
radius = radius.float() |
|
|
|
|
|
else: |
|
|
radius = self.radius |
|
|
|
|
|
freqs = torch.polar(radius.unsqueeze(-1), freqs) |
|
|
|
|
|
if "rotary" in self.debug: |
|
|
if f0 is not None: |
|
|
key = f"{self._counter}_{theta:.2f}" |
|
|
if key not in rotary._seen: |
|
|
if not hasattr(self, '_prev_f0_theta'): |
|
|
self._prev_f0_theta = theta |
|
|
|
|
|
elif abs(self._prev_f0_theta - theta) > 100.0: |
|
|
|
|
|
print(f"{layer} : {f0_mean} : Theta: {theta:.2f} : {theta:.2f} : {ctx} ") |
|
|
if self.radii: |
|
|
print(f"radius: {radius} Hz, enc: {layer} Hz, ctx: {ctx}") |
|
|
self._prev_f0_theta = theta |
|
|
rotary._seen.add(key) |
|
|
self._counter += 1 |
|
|
return freqs |
|
|
|
|
|
@staticmethod |
|
|
def apply_rotary(x, freqs): |
|
|
multihead_format = len(freqs.shape) == 4 |
|
|
if multihead_format: |
|
|
x1 = x[..., :freqs.shape[-1]*2] |
|
|
x2 = x[..., freqs.shape[-1]*2:] |
|
|
x1 = x1.float().reshape(*x1.shape[:-1], -1, 2).contiguous() |
|
|
x1 = torch.view_as_complex(x1) |
|
|
x1 = x1 * freqs |
|
|
x1 = torch.view_as_real(x1).flatten(-2) |
|
|
return torch.cat([x1.type_as(x), x2], dim=-1) |
|
|
else: |
|
|
x1 = x[..., :freqs.shape[-1]*2] |
|
|
x2 = x[..., freqs.shape[-1]*2:] |
|
|
|
|
|
if x.ndim == 2: |
|
|
x1 = x1.unsqueeze(0) |
|
|
x1 = x1.float().reshape(*x1.shape[:-1], -1, 2).contiguous() |
|
|
x1 = torch.view_as_complex(x1) |
|
|
x1 = x1 * freqs |
|
|
x1 = torch.view_as_real(x1).flatten(-2) |
|
|
x1 = x1.squeeze(0) |
|
|
return torch.cat([x1.type_as(x), x2], dim=-1) |
|
|
else: |
|
|
x1 = x1.float().reshape(*x1.shape[:-1], -1, 2).contiguous() |
|
|
x1 = torch.view_as_complex(x1) |
|
|
x1 = x1 * freqs |
|
|
x1 = torch.view_as_real(x1).flatten(-2) |
|
|
return torch.cat([x1.type_as(x), x2], dim=-1) |
|
|
|
|
|
class MultiheadA(nn.Module): |
|
|
_seen = set() |
|
|
rbf = False |
|
|
def __init__(self, dims: int, head: int, rotary_emb: bool = True, |
|
|
zero_val: float = 0.0001, minz: float = 0.0, maxz: float = 0.001, debug: List[str] = [], optim_attn=False): |
|
|
|
|
|
super(MultiheadA, self).__init__() |
|
|
|
|
|
self.dims = dims |
|
|
self.head = head |
|
|
self.head_dim = dims // head |
|
|
|
|
|
self.q = Linear(dims, dims) |
|
|
self.k = Linear(dims, dims, bias=False) |
|
|
self.v = Linear(dims, dims) |
|
|
self.o = Linear(dims, dims) |
|
|
|
|
|
self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') |
|
|
self.dtype = torch.float32 |
|
|
self.debug = debug |
|
|
self._counter = 0 |
|
|
|
|
|
self.pad_token = 0 |
|
|
self.rotary_emb = rotary_emb |
|
|
self.minz = minz |
|
|
self.maxz = maxz |
|
|
self.zero_val = zero_val |
|
|
self.optim_attn = optim_attn |
|
|
self.fzero = nn.Parameter(torch.tensor(zero_val, dtype=torch.float32), requires_grad=False) |
|
|
|
|
|
if rotary_emb: |
|
|
self.rope = rotary( |
|
|
dims=self.head_dim, |
|
|
debug = debug, |
|
|
radii=False, |
|
|
learned_pitch=False, |
|
|
learned_freq=False, |
|
|
learned_theta=False, |
|
|
learned_radius=False, |
|
|
) |
|
|
else: |
|
|
self.rope = None |
|
|
|
|
|
def enhanced_attention_scores(self, q, k, rbf_sigma=1.0, rbf_ratio=0.0): |
|
|
scale = (self.dims // self.head) ** -0.25 |
|
|
dot_scores = torch.matmul(q, k.transpose(-1, -2)) * scale |
|
|
if rbf_ratio <= 0.0: |
|
|
return dot_scores |
|
|
q_norm = q.pow(2).sum(dim=-1, keepdim=True) |
|
|
k_norm = k.pow(2).sum(dim=-1, keepdim=True) |
|
|
qk = torch.matmul(q, k.transpose(-1, -2)) |
|
|
dist_sq = q_norm + k_norm.transpose(-1, -2) - 2 * qk |
|
|
rbf_scores = torch.exp(-dist_sq / (2 * rbf_sigma**2)) |
|
|
return (1 - rbf_ratio) * dot_scores + rbf_ratio * rbf_scores |
|
|
|
|
|
def forward(self, x: Tensor, xa: Tensor = None, mask: Tensor = None, feat=None, layer = None) -> tuple: |
|
|
|
|
|
scale = (self.dims // self.head) ** -0.25 |
|
|
|
|
|
z = xa if xa is not None else x |
|
|
q = self.q(x).to(x.dtype) |
|
|
k = self.k(z).to(x.dtype) |
|
|
v = self.v(z).to(x.dtype) |
|
|
batch, ctx, dims = q.shape |
|
|
|
|
|
if self.rotary_emb: |
|
|
qf = self.rope(q.size(1), layer=layer, feat=feat) |
|
|
kf = self.rope(k.size(1), layer=layer, feat=feat) |
|
|
|
|
|
q = q.view(*q.shape[:2], self.head, -1).permute(0, 2, 1, 3) |
|
|
k = k.view(*k.shape[:2], self.head, -1).permute(0, 2, 1, 3) |
|
|
v = v.view(*v.shape[:2], self.head, -1).permute(0, 2, 1, 3) |
|
|
|
|
|
q = self.rope.apply_rotary(q, qf) |
|
|
k = self.rope.apply_rotary(k, kf) |
|
|
|
|
|
else: |
|
|
q = q.view(*q.shape[:2], self.head, -1).permute(0, 2, 1, 3) |
|
|
k = k.view(*k.shape[:2], self.head, -1).permute(0, 2, 1, 3) |
|
|
v = v.view(*v.shape[:2], self.head, -1).permute(0, 2, 1, 3) |
|
|
batch, head, ctx, head_dim = q.shape |
|
|
|
|
|
if self.rbf: |
|
|
qk = self.enhanced_attention_scores(q * scale, k * scale, rbf_sigma=1.0, rbf_ratio=0.3) |
|
|
|
|
|
qk = (q * scale) @ (k * scale).transpose(-1, -2) |
|
|
if self.rope.use_pbias: |
|
|
pbias = self.rope.pbias(feat.get("f0")) |
|
|
if pbias is not None: |
|
|
qk = qk + pbias[:,:,:q.shape[2],:q.shape[2]] |
|
|
token_ids = k[:, :, :, 0] |
|
|
zscale = torch.ones_like(token_ids) |
|
|
fzero = torch.clamp(F.softplus(self.fzero), self.minz, self.maxz) |
|
|
zscale[token_ids.float() == self.pad_token] = fzero.to(q.device, q.dtype) |
|
|
|
|
|
if mask is not None: |
|
|
mask = mask[:q.shape[2], :q.shape[2]] |
|
|
qk = qk + mask.unsqueeze(0).unsqueeze(0) * zscale.unsqueeze(-2).expand(qk.shape) |
|
|
qk = qk * zscale.unsqueeze(-2) |
|
|
w = F.softmax(qk, dim=-1).to(q.dtype) |
|
|
wv = (w @ v).permute(0, 2, 1, 3).flatten(start_dim=2) |
|
|
|
|
|
if "multihead" in self.debug and self._counter % 100 == 0: |
|
|
print(f"Step {self._counter}: Using rotary embeddings: {self.rotary_emb}") |
|
|
print(f"MHA: q={q.shape}, k={k.shape}, v={v.shape}") |
|
|
print(f"Attention shape: {qk.shape}, wv shape: {wv.shape}") |
|
|
self._counter += 1 |
|
|
return self.o(wv), qk.detach() |
|
|
|
|
|
class t_gate(nn.Module): |
|
|
def __init__(self, dims, num_types=4): |
|
|
super().__init__() |
|
|
self.gate_projections = nn.ModuleList([ |
|
|
nn.Sequential(Linear(dims, 1), nn.Sigmoid()) |
|
|
for _ in range(num_types)]) |
|
|
self.type_classifier = nn.Sequential( |
|
|
Linear(dims, num_types), |
|
|
nn.Softmax(dim=-1)) |
|
|
def forward(self, x): |
|
|
type_probs = self.type_classifier(x) |
|
|
gates = torch.stack([gate(x) for gate in self.gate_projections], dim=-1) |
|
|
comb_gate = torch.sum(gates * type_probs.unsqueeze(2), dim=-1) |
|
|
return comb_gate |
|
|
|
|
|
class m_gate(nn.Module): |
|
|
def __init__(self, dims, mem_size=64): |
|
|
super().__init__() |
|
|
self.m_key = nn.Parameter(torch.randn(mem_size, dims)) |
|
|
self.m_val = nn.Parameter(torch.randn(mem_size, 1)) |
|
|
self.gate_proj = nn.Sequential(Linear(dims, dims//2), nn.SiLU(), Linear(dims//2, 1)) |
|
|
|
|
|
def forward(self, x): |
|
|
d_gate = torch.sigmoid(self.gate_proj(x)) |
|
|
attention = torch.matmul(x, self.m_key.transpose(0, 1)) |
|
|
attention = F.softmax(attention / math.sqrt(x.shape[-1]), dim=-1) |
|
|
m_gate = torch.matmul(attention, self.m_val) |
|
|
m_gate = torch.sigmoid(m_gate) |
|
|
return 0.5 * (d_gate + m_gate) |
|
|
|
|
|
class c_gate(nn.Module): |
|
|
def __init__(self, dims): |
|
|
super().__init__() |
|
|
self.s_gate = nn.Sequential(Linear(dims, 1), nn.Sigmoid()) |
|
|
self.w_gate = nn.Sequential(Linear(dims, 1), nn.Sigmoid()) |
|
|
self.p_gate = nn.Sequential(Linear(dims, 1), nn.Sigmoid()) |
|
|
self.integ = Linear(dims*3, dims) |
|
|
|
|
|
def forward(self, x, features): |
|
|
s_feat = features.get("spectrogram", x) |
|
|
w_feat = features.get("waveform", x) |
|
|
p_feat = features.get("pitch", x) |
|
|
s = self.s_gate(x) * s_feat |
|
|
w = self.w_gate(x) * w_feat |
|
|
p = self.p_gate(x) * p_feat |
|
|
|
|
|
comb = torch.cat([s, w, p], dim=-1) |
|
|
return self.integ(comb) |
|
|
|
|
|
class Residual(nn.Module): |
|
|
_seen = set() |
|
|
def __init__(self, ctx, dims, head, act, cross_attn=True, debug: List[str] = [], |
|
|
tgate=True, mgate=False, cgate=False, mem_size=512, features=None): |
|
|
super().__init__() |
|
|
|
|
|
self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') |
|
|
self.dtype = torch.float32 |
|
|
self.dims = dims |
|
|
self.head = head |
|
|
self.ctx = ctx |
|
|
self.head_dim = dims // head |
|
|
self.cross_attn = cross_attn |
|
|
self.features = features |
|
|
self.debug = debug |
|
|
self._counter = 0 |
|
|
self.dropout = 0.01 |
|
|
|
|
|
self.t_gate = tgate |
|
|
self.m_gate = mgate |
|
|
self.c_gate = cgate |
|
|
|
|
|
self.blend = nn.Parameter(torch.tensor(0.5)) |
|
|
|
|
|
act_map = {"gelu": nn.GELU(), "relu": nn.ReLU(), "sigmoid": nn.Sigmoid(), |
|
|
"tanh": nn.Tanh(), "swish": nn.SiLU(), "tanhshrink": nn.Tanhshrink(), |
|
|
"softplus": nn.Softplus(), "softshrink": nn.Softshrink(), |
|
|
"leaky_relu": nn.LeakyReLU(), "elu": nn.ELU()} |
|
|
act_fn = act_map.get(act, nn.GELU()) |
|
|
|
|
|
self.attna = MultiheadA(dims, head, rotary_emb=True, debug=debug) |
|
|
self.attnb = (MultiheadA(dims, head, rotary_emb=True, debug=debug) if cross_attn else None) |
|
|
|
|
|
mlp = dims * 4 |
|
|
self.mlp = nn.Sequential(Linear(dims, mlp), act_fn, Linear(mlp, dims)) |
|
|
|
|
|
self.t_gate = t_gate(dims=dims, num_types=4) if t_gate else None |
|
|
self.m_gate = m_gate(dims=dims, mem_size=mem_size) if m_gate else None |
|
|
self.c_gate = c_gate(dims=dims) if c_gate else None |
|
|
|
|
|
self.lna = RMSNorm(dims) |
|
|
self.lnb = RMSNorm(dims) if cross_attn else None |
|
|
self.lnc = RMSNorm(dims) |
|
|
|
|
|
if not any([t_gate, m_gate, c_gate]): |
|
|
self.mlp_gate = nn.Sequential(Linear(dims, 1), nn.Sigmoid()) |
|
|
|
|
|
def forward(self, x: Tensor, xa: Tensor = None, mask: Tensor = None, feat=None, layer = None): |
|
|
bln = self.blend |
|
|
x = x + self.attna(self.lna(x), xa=None, mask=mask, layer=layer, feat=feat)[0] |
|
|
|
|
|
if self.attnb and xa is not None: |
|
|
c = self.attnb(self.lnb(x), xa, mask=None, layer=layer, feat=feat)[0] |
|
|
b = torch.sigmoid(bln) |
|
|
x = b * x + (1 - b) * c |
|
|
|
|
|
normx = self.lnc(x) |
|
|
mlp_out = self.mlp(normx) |
|
|
|
|
|
if self.t_gate: |
|
|
gate = self.t_gate(normx) |
|
|
x = x + gate * mlp_out |
|
|
|
|
|
elif self.m_gate: |
|
|
gate = self.m_gate(normx) |
|
|
x = x + gate * mlp_out |
|
|
|
|
|
elif self.c_gate is not None: |
|
|
gate_output = self.c_gate(normx, self.features) |
|
|
x = x + gate_output |
|
|
|
|
|
else: |
|
|
if hasattr(self, 'mlp_gate'): |
|
|
mlp_gate = self.mlp_gate(normx) |
|
|
x = x + mlp_gate * mlp_out |
|
|
else: |
|
|
x = x + mlp_out |
|
|
|
|
|
if "residual" in self.debug and self._counter % 100 == 0: |
|
|
print(f"Step {self._counter}: Residual block output shape: {x.shape}, xa shape: {xa.shape if xa is not None else None}") |
|
|
if self.t_gate: |
|
|
print(f"Step {self._counter}: Using t_gate: {self.t_gate}") |
|
|
elif self.m_gate: |
|
|
print(f"Step {self._counter}: Using m_gate: {self.m_gate}") |
|
|
elif self.c_gate: |
|
|
print(f"Step {self._counter}: Using c_gate: {self.c_gate}") |
|
|
else: |
|
|
print(f"Step {self._counter}: Using MLP gate: {self.mlp_gate if hasattr(self, 'mlp_gate') else None}") |
|
|
self._counter += 1 |
|
|
|
|
|
return x |
|
|
|
|
|
class PEncoder(nn.Module): |
|
|
def __init__(self, input_dims, dims, head, layer, kernel_size, act): |
|
|
super().__init__() |
|
|
|
|
|
self.head_dim = dims // head |
|
|
self.dropout = 0.01 |
|
|
|
|
|
act_map = {"gelu": nn.GELU(), "relu": nn.ReLU(), "sigmoid": nn.Sigmoid(), "tanh": nn.Tanh(), "swish": nn.SiLU(), "tanhshrink": nn.Tanhshrink(), "softplus": nn.Softplus(), "softshrink": nn.Softshrink(), "leaky_relu": nn.LeakyReLU(), "elu": nn.ELU()} |
|
|
act_fn = act_map.get(act, nn.GELU()) |
|
|
|
|
|
self.encoder = nn.Sequential( |
|
|
Conv1d(input_dims, dims//4, kernel_size=7, stride=8, padding=3), act_fn, |
|
|
Conv1d(dims//4, dims//2, kernel_size=5, stride=4, padding=2), act_fn, |
|
|
Conv1d(dims//2, dims, kernel_size=5, stride=5, padding=2),act_fn) |
|
|
|
|
|
def forward(self, x, feat=None, layer=None): |
|
|
x = self.encoder(x).permute(0, 2, 1) |
|
|
x = x + self.positional(x.shape[1]).to(x.device, x.dtype) |
|
|
x = nn.functional.dropout(x, p=self.dropout, training=self.training) |
|
|
x = self.norm(x) |
|
|
return x |
|
|
|
|
|
class WEncoder(nn.Module): |
|
|
def __init__(self, input_dims, dims, head, layer, kernel_size, act): |
|
|
super().__init__() |
|
|
|
|
|
self.head_dim = dims // head |
|
|
self.dropout = 0.01 |
|
|
|
|
|
act_map = {"gelu": nn.GELU(), "relu": nn.ReLU(), "sigmoid": nn.Sigmoid(), "tanh": nn.Tanh(), "swish": nn.SiLU(), "tanhshrink": nn.Tanhshrink(), "softplus": nn.Softplus(), "softshrink": nn.Softshrink(), "leaky_relu": nn.LeakyReLU(), "elu": nn.ELU()} |
|
|
act_fn = act_map.get(act, nn.GELU()) |
|
|
|
|
|
self.downsample = nn.Sequential( |
|
|
Conv1d(input_dims, dims//8, kernel_size=15, stride=8, padding=7), act_fn, |
|
|
Conv1d(dims//8, dims//4, kernel_size=7, stride=4, padding=3), act_fn, |
|
|
Conv1d(dims//4, dims, kernel_size=9, stride=5, padding=4), act_fn) |
|
|
|
|
|
self.encoder = nn.Sequential( |
|
|
Conv1d(dims, dims, kernel_size=3, padding=1, groups=dims//8), act_fn, |
|
|
Conv1d(dims, dims, kernel_size=1), act_fn) |
|
|
|
|
|
self.positional = lambda length: sinusoids(length, dims) |
|
|
self.norm = RMSNorm(dims) |
|
|
|
|
|
def forward(self, x, feat=None, layer=None): |
|
|
x = self.downsample(x) |
|
|
x = self.encoder(x) |
|
|
x = x.permute(0, 2, 1) |
|
|
x = x + self.positional(x.shape[1]).to(x.device, x.dtype) |
|
|
x = nn.functional.dropout(x, p=self.dropout, training=self.training) |
|
|
return self.norm(x) |
|
|
|
|
|
class FEncoder(nn.Module): |
|
|
def __init__(self, input_dims, dims, head, layer, kernel_size, act, stride=1): |
|
|
super().__init__() |
|
|
|
|
|
self.head_dim = dims // head |
|
|
self.dropout = 0.01 |
|
|
|
|
|
act_map = {"gelu": nn.GELU(), "relu": nn.ReLU(), "sigmoid": nn.Sigmoid(), "tanh": nn.Tanh(), "swish": nn.SiLU(), "tanhshrink": nn.Tanhshrink(), "softplus": nn.Softplus(), "softshrink": nn.Softshrink(), "leaky_relu": nn.LeakyReLU(), "elu": nn.ELU()} |
|
|
act_fn = act_map.get(act, nn.GELU()) |
|
|
|
|
|
self.encoder = nn.Sequential( |
|
|
Conv1d(input_dims, dims, kernel_size=kernel_size, stride=stride, padding=kernel_size//2), act_fn, |
|
|
Conv1d(dims, dims, kernel_size=5, padding=2), act_fn, |
|
|
Conv1d(dims, dims, kernel_size=3, padding=1, groups=dims), act_fn) |
|
|
|
|
|
self.positional = lambda length: sinusoids(length, dims) |
|
|
self.norm = RMSNorm(dims) |
|
|
self._norm = RMSNorm(dims) |
|
|
|
|
|
def forward(self, x, feat=None, layer=None): |
|
|
x = self.encoder(x).permute(0, 2, 1) |
|
|
x = x + self.positional(x.shape[1]).to(x.device, x.dtype) |
|
|
x = nn.functional.dropout(x, p=self.dropout, training=self.training) |
|
|
x = self._norm(x) |
|
|
return x |
|
|
|
|
|
class F0Encoder(nn.Module): |
|
|
def __init__(self, input_dims, dims, head, layer, kernel_size, act, stride=1): |
|
|
super().__init__() |
|
|
|
|
|
self.head_dim = dims // head |
|
|
self.dropout = 0.01 |
|
|
|
|
|
act_map = {"gelu": nn.GELU(), "relu": nn.ReLU(), "sigmoid": nn.Sigmoid(), |
|
|
"tanh": nn.Tanh(), "swish": nn.SiLU(), "tanhshrink": nn.Tanhshrink(), |
|
|
"softplus": nn.Softplus(), "softshrink": nn.Softshrink(), |
|
|
"leaky_relu": nn.LeakyReLU(), "elu": nn.ELU()} |
|
|
act_fn = act_map.get(act, nn.GELU()) |
|
|
|
|
|
self.encoder = nn.Sequential( |
|
|
Conv1d(input_dims, dims, kernel_size=kernel_size, stride=stride, padding=kernel_size//2), act_fn, |
|
|
Conv1d(dims, dims, kernel_size=5, padding=2), act_fn, |
|
|
Conv1d(dims, dims, kernel_size=3, padding=1, groups=dims), act_fn) |
|
|
|
|
|
self.positional = lambda length: sinusoids(length, dims) |
|
|
self.norm = RMSNorm(dims) |
|
|
self._norm = RMSNorm(dims) |
|
|
|
|
|
def forward(self, x, feat=None, layer=None): |
|
|
if x.dim() == 3 and x.shape[0] == 1 and x.shape[1] == 1: |
|
|
pass |
|
|
elif x.dim() == 2: |
|
|
x = x.unsqueeze(1) |
|
|
elif x.dim() == 1: |
|
|
x = x.unsqueeze(0).unsqueeze(0) |
|
|
x = self.encoder(x) |
|
|
x = x.permute(0, 2, 1) |
|
|
x = x + self.positional(x.shape[1]).to(x.device, x.dtype) |
|
|
x = nn.functional.dropout(x, p=self.dropout, training=self.training) |
|
|
x = self._norm(x) |
|
|
return x |
|
|
|
|
|
class AudioEncoder(nn.Module): |
|
|
_seen = set() |
|
|
def __init__(self, mels: int, ctx: int, dims: int, head: int, layer: int, debug: List[str], features: List[str], |
|
|
f0_rotary: bool = False, act: str = "gelu"): |
|
|
super(AudioEncoder, self).__init__() |
|
|
|
|
|
self.dims = dims |
|
|
self.head = head |
|
|
self.ctx = ctx |
|
|
self.head_dim = dims // head |
|
|
|
|
|
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') |
|
|
dtype = torch.float32 |
|
|
self.device = device |
|
|
self.dtype = dtype |
|
|
self.debug = debug |
|
|
self._counter = 0 |
|
|
|
|
|
self.features = features |
|
|
self.dropout = 0.01 |
|
|
self.f0_rotary = f0_rotary |
|
|
|
|
|
self.rope = rotary( |
|
|
dims=self.head_dim) |
|
|
|
|
|
act_map = {"gelu": nn.GELU(), "relu": nn.ReLU(), "sigmoid": nn.Sigmoid(), "tanh": nn.Tanh(), "swish": nn.SiLU(), |
|
|
"tanhshrink": nn.Tanhshrink(), "softplus": nn.Softplus(), "softshrink": nn.Softshrink(), "leaky_relu": nn.LeakyReLU(), "elu": nn.ELU()} |
|
|
act_fn = act_map.get(act, nn.GELU()) |
|
|
|
|
|
if features == ["spectrogram", "waveform", "pitch"]: |
|
|
cgate=True |
|
|
else: |
|
|
cgate = False |
|
|
|
|
|
self.blocks = nn.ModuleDict({ |
|
|
"spectrogram": nn.ModuleList( |
|
|
[FEncoder(input_dims=mels, dims=dims, head=head, layer=layer, kernel_size=3, act=act_fn)] + |
|
|
[Residual(ctx=ctx, dims=dims, head=head, act=act, debug=debug, features=features, cgate=cgate) for _ in range(layer)] if "spectrogram" in features else None |
|
|
), |
|
|
"waveform": nn.ModuleList( |
|
|
[WEncoder(input_dims=1, dims=dims, head=head, layer=layer, kernel_size=11, act=act_fn)] + |
|
|
[Residual(ctx=ctx, dims=dims, head=head, act=act, debug=debug, features=features, cgate=cgate) for _ in range(layer)] if "waveform" in features else None |
|
|
), |
|
|
"pitch": nn.ModuleList( |
|
|
[FEncoder(input_dims=1, dims=dims, head=head, layer=layer, kernel_size=9, act=act, stride=2)] + |
|
|
[Residual(ctx=ctx, dims=dims, head=head, act=act, debug=debug, features=features, cgate=cgate) for _ in range(layer)] if "pitch" in features else None |
|
|
), |
|
|
"spec_envelope": nn.ModuleList( |
|
|
[FEncoder(input_dims=mels, dims=dims, head=head, layer=layer, kernel_size=3, act=act_fn)] + |
|
|
[Residual(ctx=ctx, dims=dims, head=head, act=act, debug=debug) for _ in range(layer)] if "spec_envelope" in features else None |
|
|
), |
|
|
"spec_phase": nn.ModuleList( |
|
|
[FEncoder(input_dims=mels, dims=dims, head=head, layer=layer, kernel_size=3, act=act_fn)] + |
|
|
[Residual(ctx=ctx, dims=dims, head=head, act=act, debug=debug) for _ in range(layer)] if "spec_phase" in features else None), |
|
|
}) |
|
|
|
|
|
self.f0 = nn.ModuleList([ |
|
|
FEncoder(input_dims=1, dims=dims, head=head, layer=layer, kernel_size=9, act=act, stride=2) |
|
|
for _ in range(layer)]) |
|
|
|
|
|
def forward(self, feat, layer="encoder"): |
|
|
|
|
|
if self._counter < 1: |
|
|
s = feat.get("spectrogram") |
|
|
w = feat.get("waveform") |
|
|
p = default(feat.get("f0"), feat.get("pitch")) |
|
|
plot_waveform(x=s, w=w, p=p, hop_length=128) |
|
|
|
|
|
enc = {} |
|
|
enc.update(feat) |
|
|
|
|
|
for f in self.features: |
|
|
if f in feat and f in self.blocks: |
|
|
x = feat[f] |
|
|
for block in self.blocks[f]: |
|
|
x = block(x, feat=feat, layer=layer) |
|
|
enc[f] = x |
|
|
|
|
|
if "encoder" in self.debug and self._counter % 100 == 0: |
|
|
names = list(feat.keys()) |
|
|
shapes = {k: v.shape for k, v in feat.items()} |
|
|
print(f"Step {self._counter}: mode: {names}") |
|
|
print(f"shapes: {shapes}") |
|
|
for name, param in self.named_parameters(): |
|
|
if param.requires_grad: |
|
|
print(f"ENCODER LAYER {name}: grad_norm={param.median():.4f}") |
|
|
self._counter += 1 |
|
|
return enc |
|
|
|
|
|
class TextDecoder(nn.Module): |
|
|
def __init__(self, vocab: int, ctx: int, dims: int, head: int, layer: int, cross_attn: bool, |
|
|
debug: List[str], features: List[str], f0_rotary: bool = False, sequential=False): |
|
|
super(TextDecoder, self).__init__() |
|
|
|
|
|
self.dims = dims |
|
|
self.head = head |
|
|
self.ctx = ctx |
|
|
self.head_dim = dims // head |
|
|
|
|
|
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') |
|
|
dtype = torch.float32 |
|
|
self.device = device |
|
|
self.dtype = dtype |
|
|
self.debug = debug |
|
|
self._counter = 0 |
|
|
|
|
|
self.dropout = 0.01 |
|
|
self.sequential = sequential |
|
|
self.features = features |
|
|
self.f0_rotary = f0_rotary |
|
|
|
|
|
self.token = nn.Embedding(num_embeddings=vocab, embedding_dim=dims) |
|
|
with torch.no_grad(): |
|
|
self.token.weight[0].zero_() |
|
|
self.positional = nn.Parameter(data=torch.empty(ctx, dims), requires_grad=True) |
|
|
|
|
|
self.block = nn.ModuleList([ |
|
|
Residual(ctx=ctx, dims=dims, head=head, act="gelu", cross_attn=cross_attn, debug=debug, features=features) |
|
|
for _ in range(layer)]) |
|
|
|
|
|
self.blocks = nn.ModuleDict({ |
|
|
f: nn.ModuleList([Residual(ctx=ctx, dims=dims, head=head, act="gelu", cross_attn=cross_attn, debug=debug, features=features) |
|
|
for _ in range(layer)]) for f in features}) |
|
|
|
|
|
self.blend = nn.ParameterDict({f: nn.Parameter(torch.tensor(0.5)) for f in features}) |
|
|
self.ln_dec = RMSNorm(dims) |
|
|
|
|
|
mask = torch.tril(torch.ones(ctx, ctx), diagonal=0) |
|
|
self.register_buffer("mask", mask, persistent=False) |
|
|
|
|
|
rotary_emb = False |
|
|
if rotary_emb: |
|
|
self.rope = rotary( |
|
|
dims=self.head_dim, |
|
|
debug = debug, |
|
|
radii=False, |
|
|
learned_pitch=False, |
|
|
learned_freq=False, |
|
|
learned_theta=False, |
|
|
learned_radius=False, |
|
|
) |
|
|
else: |
|
|
self.rope = None |
|
|
|
|
|
def forward(self, x, feat, order=None, layer='decoder') -> Tensor: |
|
|
|
|
|
bln = self.blend |
|
|
x = x.to(device) |
|
|
if order is None: |
|
|
order = self.features |
|
|
mask = self.mask[:x.shape[1], :x.shape[1]] |
|
|
x = self.token(x) + self.positional[:x.shape[1]] |
|
|
x = F.dropout(x, p=self.dropout, training=self.training) |
|
|
|
|
|
for block in self.block: |
|
|
x = block(x, xa=None, mask=mask, feat=feat, layer=layer) |
|
|
|
|
|
for f in order: |
|
|
if f in feat: |
|
|
xa = feat[f] |
|
|
for block in self.blocks[f]: |
|
|
out = block(x=x, xa=xa, mask=None, feat=feat, layer=layer) |
|
|
a = torch.sigmoid(bln[f]) |
|
|
x = a * out + (1 - a) * x |
|
|
x = self.ln_dec(x) |
|
|
|
|
|
if "decoder" in self.debug and self._counter % 100 == 0: |
|
|
for name, param in self.named_parameters(): |
|
|
if param.requires_grad: |
|
|
print(f"DECODER LAYER {name}: grad_norm={param.median():.4f}") |
|
|
self._counter += 1 |
|
|
return x @ torch.transpose(self.token.weight.to(dtype), 0, 1).float() |
|
|
|
|
|
class Echo(nn.Module): |
|
|
def __init__(self, param: Dimensions): |
|
|
super().__init__() |
|
|
self.param = param |
|
|
self.count = 0 |
|
|
|
|
|
self.encoder = AudioEncoder( |
|
|
mels=param.mels, |
|
|
ctx=param.aud_ctx, |
|
|
dims=param.aud_dims, |
|
|
head=param.aud_head, |
|
|
layer=param.aud_idx, |
|
|
act=param.act, |
|
|
debug=param.debug, |
|
|
features=param.features, |
|
|
f0_rotary=param.f0_rotary, |
|
|
) |
|
|
|
|
|
self.decoder = TextDecoder( |
|
|
vocab=param.vocab, |
|
|
ctx=param.text_ctx, |
|
|
dims=param.text_dims, |
|
|
head=param.text_head, |
|
|
layer=param.text_idx, |
|
|
cross_attn=param.cross_attn, |
|
|
debug=param.debug, |
|
|
features=param.features, |
|
|
f0_rotary=param.f0_rotary, |
|
|
) |
|
|
|
|
|
all_head = torch.zeros(self.param.text_idx, self.param.text_head, dtype=torch.bool) |
|
|
all_head[self.param.text_idx // 2 :] = True |
|
|
self.register_buffer("alignment_head", all_head.to_sparse(), persistent=False) |
|
|
|
|
|
def set_alignment_head(self, dump: bytes): |
|
|
array = np.frombuffer( |
|
|
gzip.decompress(base64.b85decode(dump)), dtype=bool).copy() |
|
|
mask = torch.from_numpy(array).reshape( |
|
|
self.param.text_idx, self.param.text_head) |
|
|
self.register_buffer("alignment_head", mask.to_sparse(), persistent=False) |
|
|
|
|
|
def embed_audio(self, spectrogram: torch.Tensor): |
|
|
return self.encoder(spectrogram) |
|
|
|
|
|
def logits(self,input_ids: torch.Tensor, encoder_output: torch.Tensor): |
|
|
return self.decoder(input_ids, encoder_output) |
|
|
|
|
|
def forward(self, |
|
|
decoder_input_ids=None, |
|
|
labels=None, |
|
|
waveform: Optional[torch.Tensor]=None, |
|
|
input_ids=None, |
|
|
spectrogram: torch.Tensor=None, |
|
|
pitch: Optional[torch.Tensor]=None, |
|
|
f0: Optional[torch.Tensor]=None, |
|
|
f0d: Optional[torch.Tensor]=None, |
|
|
envelope: Optional[torch.Tensor]=None, |
|
|
phase: Optional[torch.Tensor]=None, |
|
|
) -> Dict[str, torch.Tensor]: |
|
|
|
|
|
decoder_input_ids = input_ids |
|
|
encoder_inputs = {} |
|
|
if spectrogram is not None: |
|
|
encoder_inputs["spectrogram"] = spectrogram |
|
|
if waveform is not None: |
|
|
encoder_inputs["waveform"] = waveform |
|
|
if pitch is not None: |
|
|
encoder_inputs["pitch"] = pitch |
|
|
if envelope is not None: |
|
|
encoder_inputs["envelope"] = envelope |
|
|
if phase is not None: |
|
|
encoder_inputs["phase"] = phase |
|
|
if f0 is not None: |
|
|
encoder_inputs["f0"] = f0 |
|
|
if f0d is not None: |
|
|
encoder_inputs["f0d"] = f0d |
|
|
|
|
|
encoder_outputs = self.encoder(encoder_inputs) |
|
|
logits = self.decoder(input_ids, encoder_outputs) |
|
|
|
|
|
loss = None |
|
|
if labels is not None: |
|
|
loss = F.cross_entropy( |
|
|
logits.view(-1, logits.shape[-1]), labels.view(-1), ignore_index=0) |
|
|
|
|
|
self.count += 1 |
|
|
return { |
|
|
"logits": logits, |
|
|
"loss": loss, |
|
|
"labels": labels, |
|
|
"input_ids": input_ids, |
|
|
"decoder_input_ids": decoder_input_ids, |
|
|
"encoder_output": encoder_outputs, |
|
|
} |
|
|
|
|
|
def device(self): |
|
|
return next(self.parameters()).device |
|
|
@property |
|
|
def dtype(self): |
|
|
return next(self.parameters()).dtype |
|
|
|
|
|
def _init_weights(self, module): |
|
|
std = 0.02 |
|
|
self.init_counts = { |
|
|
"Linear": 0, "Conv1d": 0, "LayerNorm": 0, "RMSNorm": 0, |
|
|
"Conv2d": 0, "SEBlock": 0, "TextDecoder": 0, "AudioEncoder": 0, |
|
|
"Residual": 0, "MultiheadA": 0, "MultiheadB - Cross Attention": 0, |
|
|
"MultiheadC": 0, "MultiheadD": 0, "FEncoder": 0, |
|
|
"WEncoder": 0, "PEncoder": 0} |
|
|
|
|
|
for name, module in self.named_modules(): |
|
|
if isinstance(module, RMSNorm): |
|
|
nn.init.ones_(module.weight) |
|
|
self.init_counts["RMSNorm"] += 1 |
|
|
elif isinstance(module, nn.Linear): |
|
|
if module.weight is not None: |
|
|
nn.init.xavier_uniform_(module.weight) |
|
|
if module.bias is not None: |
|
|
nn.init.zeros_(module.bias) |
|
|
self.init_counts["Linear"] += 1 |
|
|
elif isinstance(module, Conv1d): |
|
|
nn.init.normal_(module.weight, mean=0.0, std=std) |
|
|
if module.bias is not None: |
|
|
nn.init.zeros_(module.bias) |
|
|
self.init_counts["Conv1d"] += 1 |
|
|
elif isinstance(module, Conv2d): |
|
|
nn.init.normal_(module.weight, mean=0.0, std=std) |
|
|
if module.bias is not None: |
|
|
nn.init.zeros_(module.bias) |
|
|
self.init_counts["Conv2d"] += 1 |
|
|
elif isinstance(module, MultiheadA): |
|
|
|
|
|
self.init_counts["MultiheadA"] += 1 |
|
|
elif isinstance(module, TextDecoder): |
|
|
self.init_counts["TextDecoder"] += 1 |
|
|
elif isinstance(module, AudioEncoder): |
|
|
self.init_counts["AudioEncoder"] += 1 |
|
|
elif isinstance(module, Residual): |
|
|
self.init_counts["Residual"] += 1 |
|
|
|
|
|
def init_weights(self): |
|
|
print("Initializing model weights...") |
|
|
self.apply(self._init_weights) |
|
|
print("Initialization summary:") |
|
|
for module_type, count in self.init_counts.items(): |
|
|
if count > 0: |
|
|
print(f"{module_type}: {count}") |
|
|
|
|
|
def register_gradient_hooks(self): |
|
|
|
|
|
for name, param in self.named_parameters(): |
|
|
if param.requires_grad: |
|
|
if "encoder" in name: |
|
|
param.register_hook(lambda grad, n=name: self._print_encoder_grad(n, grad)) |
|
|
elif "decoder" in name: |
|
|
param.register_hook(lambda grad, n=name: self._print_decoder_grad(n, grad)) |
|
|
|
|
|
print("Gradient debugging hooks registered") |
|
|
return self |
|
|
|
|
|
def _print_encoder_grad(self, name, grad): |
|
|
if grad is not None and self.count == 10: |
|
|
norm = grad.median().item() |
|
|
print(f"ENCODER GRAD: {name} = {norm:.6f}") |
|
|
|
|
|
return None |
|
|
|
|
|
def _print_decoder_grad(self, name, grad): |
|
|
if grad is not None and self.count == 10: |
|
|
norm = grad.median().item() |
|
|
print(f"DECODER GRAD: {name} = {norm:.6f}") |
|
|
return None |
|
|
|
|
|
def reset_counter(self): |
|
|
self._counter = 0 |
|
|
print("Counter reset to 0.") |
|
|
|
|
|
|
