radam

config.json

@@ -71,7 +71,7 @@
     "dataset": "ljspeech",      // DATASET-RELATED: one of TTS.dataset.preprocessors depending on your target dataset. Use "tts_cache" for pre-computed dataset by extract_features.py
     "min_seq_len": 6,       // DATASET-RELATED: minimum text length to use in training
     "max_seq_len": 150,     // DATASET-RELATED: maximum text length
-    "output_path": "/media/erogol/data_ssd/Models/libri_tts/",      // DATASET-RELATED: output path for all training outputs.
+    "output_path": "../keep/",      // DATASET-RELATED: output path for all training outputs.
     "num_loader_workers": 4,        // number of training data loader processes. Don't set it too big. 4-8 are good values.
     "num_val_loader_workers": 4,    // number of evaluation data loader processes.
     "phoneme_cache_path": "mozilla_us_phonemes",  // phoneme computation is slow, therefore, it caches results in the given folder.

train.py

@@ -27,6 +27,7 @@ from utils.speakers import load_speaker_mapping, save_speaker_mapping, \
 from utils.synthesis import synthesis
 from utils.text.symbols import phonemes, symbols
 from utils.visual import plot_alignment, plot_spectrogram
+from utils.radam import RAdam
 from datasets.preprocess import get_preprocessor_by_name
 
 torch.backends.cudnn.enabled = True
@@ -476,9 +477,9 @@ def main(args): #pylint: disable=redefined-outer-name
 
     print(" | > Num output units : {}".format(ap.num_freq), flush=True)
 
-    optimizer = optim.Adam(model.parameters(), lr=c.lr, weight_decay=0)
+    optimizer = RAdam(model.parameters(), lr=c.lr, weight_decay=0)
     if c.stopnet and c.separate_stopnet:
-        optimizer_st = optim.Adam(
+        optimizer_st = RAdam(
             model.decoder.stopnet.parameters(), lr=c.lr, weight_decay=0)
     else:
         optimizer_st = None

utils/radam.py

@@ -0,0 +1,207 @@
+import math
+import torch
+from torch.optim.optimizer import Optimizer, required
+
+class RAdam(Optimizer):
+
+    def __init__(self, params, lr=1e-3, betas=(0.9, 0.999), eps=1e-8, weight_decay=0):
+        defaults = dict(lr=lr, betas=betas, eps=eps, weight_decay=weight_decay)
+        self.buffer = [[None, None, None] for ind in range(10)]
+        super(RAdam, self).__init__(params, defaults)
+
+    def __setstate__(self, state):
+        super(RAdam, self).__setstate__(state)
+
+    def step(self, closure=None):
+
+        loss = None
+        if closure is not None:
+            loss = closure()
+
+        for group in self.param_groups:
+
+            for p in group['params']:
+                if p.grad is None:
+                    continue
+                grad = p.grad.data.float()
+                if grad.is_sparse:
+                    raise RuntimeError('RAdam does not support sparse gradients')
+
+                p_data_fp32 = p.data.float()
+
+                state = self.state[p]
+
+                if len(state) == 0:
+                    state['step'] = 0
+                    state['exp_avg'] = torch.zeros_like(p_data_fp32)
+                    state['exp_avg_sq'] = torch.zeros_like(p_data_fp32)
+                else:
+                    state['exp_avg'] = state['exp_avg'].type_as(p_data_fp32)
+                    state['exp_avg_sq'] = state['exp_avg_sq'].type_as(p_data_fp32)
+
+                exp_avg, exp_avg_sq = state['exp_avg'], state['exp_avg_sq']
+                beta1, beta2 = group['betas']
+
+                exp_avg_sq.mul_(beta2).addcmul_(1 - beta2, grad, grad)
+                exp_avg.mul_(beta1).add_(1 - beta1, grad)
+
+                state['step'] += 1
+                buffered = self.buffer[int(state['step'] % 10)]
+                if state['step'] == buffered[0]:
+                    N_sma, step_size = buffered[1], buffered[2]
+                else:
+                    buffered[0] = state['step']
+                    beta2_t = beta2 ** state['step']
+                    N_sma_max = 2 / (1 - beta2) - 1
+                    N_sma = N_sma_max - 2 * state['step'] * beta2_t / (1 - beta2_t)
+                    buffered[1] = N_sma
+
+                    # more conservative since it's an approximated value
+                    if N_sma >= 5:
+                        step_size = group['lr'] * math.sqrt((1 - beta2_t) * (N_sma - 4) / (N_sma_max - 4) * (N_sma - 2) / N_sma * N_sma_max / (N_sma_max - 2)) / (1 - beta1 ** state['step'])
+                    else:
+                        step_size = group['lr'] / (1 - beta1 ** state['step'])
+                    buffered[2] = step_size
+
+                if group['weight_decay'] != 0:
+                    p_data_fp32.add_(-group['weight_decay'] * group['lr'], p_data_fp32)
+
+                # more conservative since it's an approximated value
+                if N_sma >= 5:            
+                    denom = exp_avg_sq.sqrt().add_(group['eps'])
+                    p_data_fp32.addcdiv_(-step_size, exp_avg, denom)
+                else:
+                    p_data_fp32.add_(-step_size, exp_avg)
+
+                p.data.copy_(p_data_fp32)
+
+        return loss
+
+class PlainRAdam(Optimizer):
+
+    def __init__(self, params, lr=1e-3, betas=(0.9, 0.999), eps=1e-8, weight_decay=0):
+        defaults = dict(lr=lr, betas=betas, eps=eps, weight_decay=weight_decay)
+
+        super(PlainRAdam, self).__init__(params, defaults)
+
+    def __setstate__(self, state):
+        super(PlainRAdam, self).__setstate__(state)
+
+    def step(self, closure=None):
+
+        loss = None
+        if closure is not None:
+            loss = closure()
+
+        for group in self.param_groups:
+
+            for p in group['params']:
+                if p.grad is None:
+                    continue
+                grad = p.grad.data.float()
+                if grad.is_sparse:
+                    raise RuntimeError('RAdam does not support sparse gradients')
+
+                p_data_fp32 = p.data.float()
+
+                state = self.state[p]
+
+                if len(state) == 0:
+                    state['step'] = 0
+                    state['exp_avg'] = torch.zeros_like(p_data_fp32)
+                    state['exp_avg_sq'] = torch.zeros_like(p_data_fp32)
+                else:
+                    state['exp_avg'] = state['exp_avg'].type_as(p_data_fp32)
+                    state['exp_avg_sq'] = state['exp_avg_sq'].type_as(p_data_fp32)
+
+                exp_avg, exp_avg_sq = state['exp_avg'], state['exp_avg_sq']
+                beta1, beta2 = group['betas']
+
+                exp_avg_sq.mul_(beta2).addcmul_(1 - beta2, grad, grad)
+                exp_avg.mul_(beta1).add_(1 - beta1, grad)
+
+                state['step'] += 1
+                beta2_t = beta2 ** state['step']
+                N_sma_max = 2 / (1 - beta2) - 1
+                N_sma = N_sma_max - 2 * state['step'] * beta2_t / (1 - beta2_t)
+
+                if group['weight_decay'] != 0:
+                    p_data_fp32.add_(-group['weight_decay'] * group['lr'], p_data_fp32)
+
+                # more conservative since it's an approximated value
+                if N_sma >= 5:                    
+                    step_size = group['lr'] * math.sqrt((1 - beta2_t) * (N_sma - 4) / (N_sma_max - 4) * (N_sma - 2) / N_sma * N_sma_max / (N_sma_max - 2)) / (1 - beta1 ** state['step'])
+                    denom = exp_avg_sq.sqrt().add_(group['eps'])
+                    p_data_fp32.addcdiv_(-step_size, exp_avg, denom)
+                else:
+                    step_size = group['lr'] / (1 - beta1 ** state['step'])
+                    p_data_fp32.add_(-step_size, exp_avg)
+
+                p.data.copy_(p_data_fp32)
+
+        return loss
+
+
+class AdamW(Optimizer):
+
+    def __init__(self, params, lr=1e-3, betas=(0.9, 0.999), eps=1e-8, weight_decay=0, warmup = 0):
+        defaults = dict(lr=lr, betas=betas, eps=eps,
+                        weight_decay=weight_decay, warmup = warmup)
+        super(AdamW, self).__init__(params, defaults)
+
+    def __setstate__(self, state):
+        super(AdamW, self).__setstate__(state)
+
+    def step(self, closure=None):
+        loss = None
+        if closure is not None:
+            loss = closure()
+
+        for group in self.param_groups:
+
+            for p in group['params']:
+                if p.grad is None:
+                    continue
+                grad = p.grad.data.float()
+                if grad.is_sparse:
+                    raise RuntimeError('Adam does not support sparse gradients, please consider SparseAdam instead')
+
+                p_data_fp32 = p.data.float()
+
+                state = self.state[p]
+
+                if len(state) == 0:
+                    state['step'] = 0
+                    state['exp_avg'] = torch.zeros_like(p_data_fp32)
+                    state['exp_avg_sq'] = torch.zeros_like(p_data_fp32)
+                else:
+                    state['exp_avg'] = state['exp_avg'].type_as(p_data_fp32)
+                    state['exp_avg_sq'] = state['exp_avg_sq'].type_as(p_data_fp32)
+
+                exp_avg, exp_avg_sq = state['exp_avg'], state['exp_avg_sq']
+                beta1, beta2 = group['betas']
+
+                state['step'] += 1
+
+                exp_avg_sq.mul_(beta2).addcmul_(1 - beta2, grad, grad)
+                exp_avg.mul_(beta1).add_(1 - beta1, grad)
+
+                denom = exp_avg_sq.sqrt().add_(group['eps'])
+                bias_correction1 = 1 - beta1 ** state['step']
+                bias_correction2 = 1 - beta2 ** state['step']
+                
+                if group['warmup'] > state['step']:
+                    scheduled_lr = 1e-8 + state['step'] * group['lr'] / group['warmup']
+                else:
+                    scheduled_lr = group['lr']
+
+                step_size = group['lr'] * math.sqrt(bias_correction2) / bias_correction1
+                
+                if group['weight_decay'] != 0:
+                    p_data_fp32.add_(-group['weight_decay'] * scheduled_lr, p_data_fp32)
+
+                p_data_fp32.addcdiv_(-step_size, exp_avg, denom)
+
+                p.data.copy_(p_data_fp32)
+
+        return loss