vall-e/vall_e/models/ar_nar_v2.py
2025-03-12 23:17:27 -05:00

1056 lines
33 KiB
Python

"""
# an AR + NAR model that handles:
* inferencing the primary RVQ level in an autoregressive manner (AR)
* inferencing the remaining RVQ levels in parallel (NAR)
This model can fully handle being trained as a unified model (AR + NAR) or separate models (AR | NAR).
It's recommended to train as a unified model, then "distill" knowledge of each tasks separately, just in case.
"""
from .base_v2 import Base_V2, list_to_tensor, Categorical
from ..config import cfg
import torch
from torch.nn.utils.rnn import pad_sequence
import random
import math
import time
from einops import rearrange
from torch import Tensor
from tqdm import trange, tqdm
import logging
_logger = logging.getLogger(__name__)
from ..emb.qnt import trim, get_silence
from ..utils import get_devices, setup_logging, timer, clamp, convert_kwargs
from .lora import enable_lora
from ..samplers import cfg_logits
text_task = [ "stt", "phn", "un-phn" ]
class AR_NAR_V2(Base_V2):
# yikes
def forward_super(self, *args, **kwargs):
return super().forward(*args, **kwargs)
# parse inputs for training
# a lot of this could be delegated back to the dataloader, but it's just easier to keep the task of the dataloader to provide sufficient data, and the model to process the data for training
def forward_train(
self,
task_list: list[Tensor] | None = None,
phns_list: list[Tensor] | None = None,
proms_list: list[Tensor] | None = None,
resps_list: list[Tensor] | None = None,
lang_list: list[Tensor] | None = None,
tone_list: list[Tensor] | None = None,
len_list: list[Tensor] | None = None,
text_list: list[Tensor] | None = None,
):
# deduce batch_size
if phns_list:
device = phns_list[0].device
batch_size = len(phns_list)
elif text_list:
device = text_list[0].device
batch_size = len(text_list)
elif proms_list:
device = proms_list[0].device
batch_size = len(proms_list)
elif resps_list:
device = resps_list[0].device
batch_size = len(resps_list)
# specifies how to sample probabilities of which RVQ levels to train against
rvq_levels_p = self.config.experimental.rvq_levels_p if self.config is not None else "equal"
# determines which RVQ level to target per batch
quant_level_range = self.config.experimental.rvq_level_range if self.config is not None and self.config.experimental.rvq_level_range else [ 0 if self.causal else 1, self.n_resp_levels - 1 ]
# rate to perform token dropout errors
token_dropout_error = self.config.experimental.token_dropout_error
# RVQ levels to apply token dropout on
token_dropout_rvq_levels = self.config.experimental.token_dropout_rvq_levels
# RVQ levels to apply masking training on
masking_train_rvq_levels = [0,self.n_resp_levels] # self.config.experimental.masking_train_rvq_levels
# cringe
self.audio_frames_per_second = cfg.dataset.frames_per_second
# CFG
cfg_text_dropout_p = self.config.experimental.cfg_text_dropout_p if self.config is not None else 0.0
cfg_cond_dropout_p = self.config.experimental.cfg_cond_dropout_p if self.config is not None else 0.0
cfg_prom_dropout_p = self.config.experimental.cfg_prom_dropout_p if self.config is not None else 0.0
use_raw_text_p = self.config.experimental.use_raw_text_p if self.config is not None else 0.0
# rate to train RVQ level AR-ly or NAR-ly
masking_train_p = self.config.experimental.masking_train_p if self.config is not None else 0.5
masking_ratio = self.config.experimental.masking_ratio if self.config is not None else "random"
# force set mask training
if "len" not in self.capabilities:
masking_train_p = 0.0
elif "ar" not in self.capabilities:
masking_train_p = 1.0
# implicitly set it to all levels
if not token_dropout_rvq_levels:
token_dropout_rvq_levels = [0, self.resp_levels - 1]
if not token_dropout_rvq_levels:
token_dropout_rvq_levels = [0, 0]
# allow passing a specific distribution of RVQ levels
rvq_levels_p = rvq_levels_p if isinstance(rvq_levels_p, list) else []
if not rvq_levels_p:
lo, hi = quant_level_range[0], quant_level_range[1] + 1
# randomly select a target RVQ-bin level (0 being AR, 1+ being NAR)
if rvq_levels_p == "equal":
rvq_levels_p = [ i for i in range( lo, hi ) ]
elif rvq_levels_p == "normal":
# yuck
rvq_levels_p = [
0,
1, 1,
2, 2, 2, 2,
3, 3, 3, 3, 3, 3, 3, 3,
4, 4, 4, 4, 4, 4, 4, 4,
5, 5, 5, 5,
6, 6,
7,
]
else:
rvq_levels_p = sum([[i for _ in range(hi - i)] for i in range( lo, hi ) ], [])
# input RVQ levels
quant_levels = [ random.choice( rvq_levels_p ) for i in range(batch_size) ]
# timestep levels (for TTS NAR)
timesteps = [ None for _ in range(batch_size) ]
for i, task in enumerate( task_list ):
lo, hi = masking_train_rvq_levels[0], masking_train_rvq_levels[1]
if task in text_task:
quant_levels[i] = 0 # self.n_resp_levels - 1
elif lo <= quant_levels[i] and quant_levels[i] <= hi and random.random() < masking_train_p:
# to-do: prioritize lower timesteps over later timesteps
# ...except that the masking rate is still tied to the cosine scheduling, which does this already
#r = random.random()
#p = math.acos(r) / (math.pi * 0.5)
#timesteps[i] = 1.0 - clamp(p, 0.0, 1.0)
timesteps[i] = random.random()
# instead make it between [0.2, 0.8]
if masking_ratio == "rand":
timesteps[i] = (timesteps[i] * 0.6) + 0.2
# tensor to cat for RVQ level 0
text_stop_sequence = torch.tensor([2], device=device, dtype=torch.int16)
text_start_stop_sequence = torch.tensor([1, 2], device=device, dtype=torch.int16)
audio_stop_sequence = torch.tensor([[self.stop_token]], device=device, dtype=torch.int16)
# final validations and stuff
for i, quant_level, resps, proms, task in zip(range(batch_size), quant_levels, resps_list, proms_list, task_list):
# cap quant_level if it exceeds its corresponding resp/prom
# this was needed for when my DAC-encoded audio was erroneously trimmed to 8 RVQ levels instead of 9
if quant_level >= resps.shape[-1]:
quant_levels[i] = resps.shape[-1] - 1
# proms could be a Tensor, list[Tensor], or None
if isinstance( proms, torch.Tensor ):
if quant_level >= proms.shape[-1]:
quant_levels[i] = proms.shape[-1] - 1
elif isinstance( proms, list ):
for j, prom in enumerate( proms ):
if not isinstance( prom, torch.Tensor ):
continue
if quant_level >= prom.shape[-1]:
quant_levels[i] = prom.shape[-1] - 1
# apply token dropout error compensation
"""
if token_dropout_error > 0 and (token_dropout_rvq_levels[0] <= quant_level and quant_level <= token_dropout_rvq_levels[1]):
steps = resps.shape[0]
for l in range( quant_level ):
for t in range( steps ):
token = resps[t, l].item()
if random.random() < token_dropout_error:
offset = 1 * ( 1 if random.random() < 0.5 else -1 )
resps_list[i][t, l] = clamp(token + offset, 1, 1022) # +- 1
"""
# only apply stop token for RVQ level 0
if timesteps[i] is None or (self.predict_causally):
# append stop tokens for AR
if task not in text_task:
resps_list[i] = torch.cat([ resps, audio_stop_sequence.repeat((1, resps.shape[-1])) ])
if task == "len":
quant_levels[i] = 0
# apply CFG (should probably only apply to NAR quant level 0)
if task not in text_task + ["len"]:
drop_text = False
drop_audio = False
swap_text = False
if random.random() < cfg_prom_dropout_p:
drop_audio = True
if random.random() < cfg_cond_dropout_p:
drop_audio = True
drop_text = True
if random.random() < use_raw_text_p and text_list[i] is not None:
swap_text = True
if drop_text:
phns_list[i] = text_start_stop_sequence
if drop_audio:
proms_list[i] = None
if swap_text and not drop_text:
phns_list[i] = None
inputs = self.inputs(
phns_list=phns_list,
proms_list=proms_list,
resps_list=resps_list,
lang_list=lang_list,
tone_list=tone_list,
task_list=task_list,
text_list=text_list,
time_list=timesteps,
quant_levels=quant_levels,
)
return super().forward(
inputs=inputs,
quant_levels=quant_levels,
)
# handles doing demasking inferencing in parallel to inference all tokens
# it works if the underlying model is trained properly (which is a pain)
def forward_nar_masked(
self,
task_list: list[Tensor] | None = None,
phns_list: list[Tensor] | None = None,
proms_list: list[Tensor] | None = None,
resps_list: list[Tensor] | None = None,
lang_list: list[Tensor] | None = None,
tone_list: list[Tensor] | None = None,
len_list: list[Tensor] | None = None,
text_list: list[Tensor] | None = None,
disable_tqdm=False,
use_lora=None,
**sampling_kwargs,
):
device = phns_list[0].device
batch_size = len(phns_list)
level = 0
if cfg.lora is not None:
enable_lora( self, cfg.lora.active_level( level ) if use_lora is None else use_lora )
# convert (N)AR specific args
sampling_kwargs = convert_kwargs( sampling_kwargs, "ar_" )
min_length = sampling_kwargs.pop("min_duration", 1)
max_length = sampling_kwargs.pop("max_duration", 500)
max_steps = sampling_kwargs.get("max_steps", 25)
refine_on_stop = sampling_kwargs.get("refine_on_stop", False)
entropix_sampling = sampling_kwargs.get("entropix_sampling", False)
annealed_sampling = sampling_kwargs.get("annealed_sampling", True)
# greedy sampling is very, very much preferred, but using greedy logit scores later helps enough
temperature = sampling_kwargs.pop("temperature", 0.0)
minimum_cfg_strength = sampling_kwargs.get("minimum_cfg_strength", 2.5)
# this really helps keep audio coherent so far
cfg_strength = sampling_kwargs.get("cfg_strength", minimum_cfg_strength)
cfg_rescale = sampling_kwargs.pop("cfg_rescale", 0.75)
start_noise = sampling_kwargs.get("denoise_start", 0.0)
end_noise = sampling_kwargs.get("denoise_end", 1.0)
remasking = sampling_kwargs.get("remasking", True)
max_steps = math.floor(max_steps * (end_noise - start_noise))
# to specify the initial mask used
vc_list = sampling_kwargs.pop("vc_list", None)
vc_threshold = sampling_kwargs.pop("vc_threshold", 0.25)
vc_mask_p = sampling_kwargs.pop("vc_mask_p", 0.25)
len_list = [ clamp(l, min_length, max_length) for l in len_list ]
# force set CFG because too low / no CFG causes issues
original_cfg_strength = cfg_strength
cfg_strength = max( cfg_strength, minimum_cfg_strength )
prefix_context = sampling_kwargs.get("prefix_context", None)
# fill with masked tokens (even though they get masked anyways)
resps_list = [ torch.ones((seq_len, self.n_resp_levels), dtype=torch.int16, device=device) * self.mask_token for seq_len in len_list ]
# fill scores
scores = [ torch.ones((seq_len), dtype=torch.float32, device=device) for seq_len in len_list ]
quant_levels = [ level for _ in range(batch_size) ]
null_text = [ torch.tensor([1, 2], device=device, dtype=torch.int16) for _ in range(batch_size) ]
null_prom = [ None for _ in range(batch_size) ]
iterator = tqdm(torch.linspace(start_noise, end_noise, max_steps), desc="NAR Masked", disable=disable_tqdm)
for timestep in iterator:
# update previous list of tokens
prev_list = resps_list
# ramp down over time
annealing = 1.0 - timestep
# get noise level, per cosine scheduling
noise_p = math.cos( timestep * math.pi * 0.5 )
# proportion of tokens to remask
remask_p = 1.0 / (max_steps * 2) if remasking else 0
# pick the worst scoring tokens to mask off
masked_indices = [ score.topk( clamp( int( noise_p * seq_len + remask_p * seq_len ), 1, seq_len), dim=-1 ).indices for score, seq_len in zip(scores, len_list) ]
# normal masking
# mask off inputs
resps_list = [ torch.stack([resp[:, l].scatter(0, indices, self.mask_token) for l in range(self.n_resp_levels)], dim=-1) for resp, indices in zip( resps_list, masked_indices ) ]
# boolean mask
is_masked = [ resps == self.mask_token for resps in resps_list ]
# timestep inputs
time_list = [ timestep for _ in range(batch_size) ]
sampling_temperature = temperature * annealing if annealed_sampling else temperature
sampling_cfg = cfg_strength * timestep if annealed_sampling else cfg_strength
input_resps_list = resps_list
# setup inputs
inputs = super().inputs(
phns_list=phns_list,
proms_list=proms_list,
resps_list=input_resps_list,
lang_list=lang_list,
tone_list=tone_list,
time_list=time_list,
quant_levels=quant_levels,
)
output = super().forward(
inputs=inputs,
quant_levels=quant_levels,
)
logits = output.logits
if cfg_strength > 0:
null_inputs = super().inputs(
phns_list=null_text,
proms_list=null_prom,
resps_list=input_resps_list,
lang_list=lang_list,
tone_list=tone_list,
time_list=time_list,
quant_levels=quant_levels,
)
null_output = super().forward(
inputs=null_inputs,
quant_levels=quant_levels,
)
logits = cfg_logits( logits=output.logits, null=null_output.logits, strength=cfg_strength, rescale=cfg_rescale, lens=[ l for l in len_list ] )
l_scores = []
l_resps_list = []
# cringe hack because we're able to sample multiple levels at once
for l in range(self.n_resp_levels):
# sample with sampler settings
filtered_sampled = super().sample(
logits=[ logit[l] for logit in logits ],
prev_list=[ resp[..., l] for resp in prev_list ],
quant_levels=quant_levels,
temperature=sampling_temperature,
**sampling_kwargs,
)
# retrieves unfiltered logits
unfiltered_sampled = super().sample(
logits=[ logit[l] for logit in logits ],
prev_list=[ resp[..., l] for resp in prev_list ],
quant_levels=quant_levels,
temperature=0.0,
**sampling_kwargs,
)
# get sampled tokens
sampled_ids = filtered_sampled.ids
# keep unmasked tokens
l_resps_list.append([ torch.where( masked[..., l], input_ids, resps[..., l] ).to(torch.int16) for masked, input_ids, resps in zip( is_masked, sampled_ids, resps_list ) ])
# get probability scores
l_scores.append([
# conjugate to have worse scoring tokens picked for topk
1.0 -
# only keep scores of tokens we are predicting (and ignore the tokens previously finalized)
torch.where( masked[..., l], torch.tensor([score for index, score in enumerate(scores)], device=device), torch.ones(masked[..., l].shape, device=device) )
# use unmodified logit scores for this, as it offers better stability
for scores, masked in zip( unfiltered_sampled.scores, is_masked )
])
resps_list = []
scores = []
for batch_index in range(batch_size):
score = sum([ l_scores[level][batch_index] for level in range(self.n_resp_levels) ]) / self.n_resp_levels
resp = torch.stack([ l_resps_list[level][batch_index] for level in range(self.n_resp_levels) ], dim=-1)
scores.append( score )
resps_list.append( resp )
return resps_list
def forward_len(
self,
task_list: list[Tensor],
phns_list: list[Tensor] | None = None,
text_list: list[Tensor] | None = None,
proms_list: list[Tensor] | None = None,
resps_list: list[Tensor] | None = None,
lang_list: list[Tensor] | None = None,
tone_list: list[Tensor] | None = None,
len_list: list[Tensor] | None = None,
disable_tqdm=False,
use_lora=None,
**sampling_kwargs,
):
# deduce batch_size
if phns_list:
device = phns_list[0].device
batch_size = len(phns_list)
elif text_list:
device = text_list[0].device
batch_size = len(text_list)
elif proms_list:
device = proms_list[0].device
batch_size = len(proms_list)
if cfg.lora is not None:
enable_lora( self, cfg.lora.active_level( 0 ) if use_lora is None else use_lora )
task_list = [ "len" for _ in range( batch_size ) ]
quant_levels = [ 0 for _ in range( batch_size ) ]
inputs = self.inputs(
task_list=task_list,
phns_list=phns_list,
proms_list=proms_list,
resps_list=None,
lang_list=lang_list,
tone_list=tone_list,
len_list=None,
text_list=text_list,
quant_levels=quant_levels,
)
output = super().forward(
inputs=inputs,
quant_levels=quant_levels,
)
logits = output.logits
return [ int(logit * cfg.dataset.frames_per_second) for logit in logits ]
def forward_ar(
self,
task_list: list[Tensor],
phns_list: list[Tensor] | None = None,
text_list: list[Tensor] | None = None,
proms_list: list[Tensor] | None = None,
resps_list: list[Tensor] | None = None,
lang_list: list[Tensor] | None = None,
tone_list: list[Tensor] | None = None,
len_list: list[Tensor] | None = None,
disable_tqdm=False,
use_lora=None,
**sampling_kwargs,
):
# deduce batch_size
if phns_list:
device = phns_list[0].device
batch_size = len(phns_list)
elif text_list:
device = text_list[0].device
batch_size = len(text_list)
elif proms_list:
device = proms_list[0].device
batch_size = len(proms_list)
elif resps_list:
device = resps_list[0].device
batch_size = len(resps_list)
if cfg.lora is not None:
enable_lora( self, cfg.lora.active_level( 0 ) if use_lora is None else use_lora )
# convert AR specific args
sampling_kwargs = convert_kwargs( sampling_kwargs, "ar_" )
temperature = sampling_kwargs.get("temperature", 1.0)
cfg_strength = sampling_kwargs.get("cfg_strength", 0.0)
cfg_rescale = sampling_kwargs.pop("cfg_rescale", 0.7)
min_temperature = sampling_kwargs.get("min_temperature", -1.0)
max_duration = sampling_kwargs.get("max_duration", 500)
beam_width = sampling_kwargs.get("beam_width", 0)
entropix_sampling = sampling_kwargs.get("entropix_sampling", False)
refine_on_stop = sampling_kwargs.get("refine_on_stop", False)
input_prompt_prefix = sampling_kwargs.get("input_prompt_prefix", False)
layer_skip = sampling_kwargs.get("layer_skip", False)
prefix_silence = sampling_kwargs.get("prefix_silence", 0.0)
mirostat_tau = sampling_kwargs.get("mirostat_tau", 0.0)
mirostat_eta = sampling_kwargs.get("mirostat_eta", 0.0)
start_slice = [ 0 for _ in range(batch_size) ]
sequence_list = [ torch.zeros((0, 8), device=device).to(torch.int16) for _ in range(batch_size) ]
stopped = torch.zeros(batch_size, device=device).bool()
audio_stop_token = self.stop_token
text_stop_token = 2
state = None
mirostat = [
{"n": 1024, "tau": mirostat_tau, "eta": mirostat_eta, "max_surprise": mirostat_eta * 2, "error_surprise": 0, "running_total_surprise": 0}
] * batch_size if mirostat_tau > 0.0 else None
scores = [ 1.0 ] * beam_width
metrics = []
null_text = [ torch.tensor([1, 2], device=device, dtype=torch.int16) for _ in range(batch_size) ]
null_prom = [ None for _ in range(batch_size) ]
# get next in sequence
iterator = trange(max_duration // max(1, self.causal_size), desc="AR", disable=disable_tqdm)
for n in iterator:
if text_list is not None:
text_list = [ sequence_list[i] if task in text_task else text_list[i] for i, task in enumerate(task_list) ]
else:
phns_list = [ sequence_list[i] if task in text_task else phns_list[i] for i, task in enumerate(task_list) ]
resps_list = [ sequence_list[i] if task not in text_task else resps_list[i] for i, task in enumerate(task_list) ]
quant_levels = [ 0 for _ in range( max( batch_size, beam_width ) ) ]
inputs = self.inputs(
task_list=task_list,
phns_list=phns_list,
proms_list=proms_list,
resps_list=resps_list,
lang_list=lang_list,
tone_list=tone_list,
len_list=len_list,
text_list=text_list,
quant_levels=quant_levels,
)
# to-do: find an elegant way to write this
output = super().forward(
inputs=inputs,
state=state,
#layer_skip_variables=sampling_layer_skip_variables,
output_attentions=entropix_sampling,
)
if cfg_strength > 0:
null_inputs = super().inputs(
phns_list=null_text,
proms_list=null_prom,
resps_list=resps_list,
lang_list=lang_list,
tone_list=tone_list,
quant_levels=quant_levels,
)
null_output = super().forward(
inputs=null_inputs,
quant_levels=quant_levels,
#layer_skip_variables=sampling_layer_skip_variables,
)
logits = cfg_logits( logits=output.logits, null=null_output.logits, strength=cfg_strength, rescale=cfg_rescale, lens=[ resp.shape[0] + 1 for resp in resps_list ] )
logits, state = output.logits, output.state
l_resps_list = [ [] for _ in range(batch_size) ]
for l in range(self.n_resp_levels):
sampled = super().sample(
logits=[ logit[l] for logit in logits ],
#prev_list=[ resp[..., l] for resp in resps_list ],
**(sampling_kwargs | {"attentions": output.attentions if entropix_sampling else None}),
)
ids = sampled.ids
# append tokens
for i, token in enumerate(ids):
if audio_stop_token in token:
stopped[i] = True
l_resps_list[i].append(token.to(device))
for i, l in enumerate(l_resps_list):
sequence_list[i] = torch.cat([sequence_list[i], torch.stack(l, dim=-1)])
# stop token found
# stopped |= r == stop_token
if stopped.all().item():
iterator.close()
break
for i, l in enumerate( sequence_list ):
index = (l == audio_stop_token).nonzero()
# kludge for when it doesnt actually hit a stop token but i cant be bothered to properly address it right now since it only came up in test training at the moment
try:
index = index[:, 0].min()
sequence_list[i] = sequence_list[i][:index]
except Exception as e:
pass
return sequence_list
def forward(
self,
task_list: list[Tensor] | None = None,
phns_list: list[Tensor] | None = None,
proms_list: list[Tensor] | None = None,
resps_list: list[Tensor] | None = None,
lang_list: list[Tensor] | None = None,
tone_list: list[Tensor] | None = None,
len_list: list[Tensor] | None = None,
text_list: list[Tensor] | None = None,
training: bool | None = None,
disable_tqdm=False,
use_lora=None,
**sampling_kwargs,
):
# deduce batch_size
if phns_list:
device = phns_list[0].device
batch_size = len(phns_list)
elif text_list:
device = text_list[0].device
batch_size = len(text_list)
elif proms_list:
device = proms_list[0].device
batch_size = len(proms_list)
elif resps_list:
device = resps_list[0].device
batch_size = len(resps_list)
# implicitly set for training
if training is None and phns_list is not None and resps_list is not None:
n_levels_set = {r.shape[-1] for r in resps_list}
n_levels = next(iter(n_levels_set))
training = n_levels == self.n_resp_levels
# is training
if training:
return self.forward_train(
task_list=task_list,
phns_list=phns_list,
proms_list=proms_list,
resps_list=resps_list,
lang_list=lang_list,
tone_list=tone_list,
len_list=len_list,
text_list=text_list,
)
# is NAR
if (len_list is not None or resps_list is not None) and phns_list is not None:
return self.forward_nar_masked(
task_list=task_list,
phns_list=phns_list,
proms_list=proms_list,
resps_list=resps_list,
lang_list=lang_list,
tone_list=tone_list,
len_list=len_list,
text_list=text_list,
disable_tqdm=disable_tqdm,
use_lora=use_lora,
**sampling_kwargs,
)
# NAR demasking for all levels
"""
resps_lists = [ None for _ in range(batch_size) ]
for level in range(self.n_resp_levels):
resp_list = self.forward_nar_masked(
task_list=task_list,
phns_list=phns_list,
proms_list=proms_list,
resps_list=resps_list,
lang_list=lang_list,
tone_list=tone_list,
len_list=len_list,
text_list=text_list,
disable_tqdm=disable_tqdm,
use_lora=use_lora,
quant_levels=[ level for _ in range(batch_size) ],
**sampling_kwargs,
)
for batch_index, resp in enumerate(resp_list):
if resps_lists[batch_index] is None:
resps_lists[batch_index] = []
resps_lists[batch_index].append( resp )
for batch_index, resps in enumerate(resps_lists):
resps_lists[batch_index] = torch.stack( resps, dim=-1 )
return resps_lists
"""
if task_list is not None and task_list[0] == "len":
return self.forward_len(
task_list=task_list,
phns_list=phns_list,
proms_list=proms_list,
resps_list=resps_list,
lang_list=lang_list,
tone_list=tone_list,
len_list=len_list,
text_list=text_list,
disable_tqdm=disable_tqdm,
use_lora=use_lora,
**sampling_kwargs,
)
# is AR
return self.forward_ar(
task_list=task_list,
phns_list=phns_list,
proms_list=proms_list,
resps_list=resps_list,
lang_list=lang_list,
tone_list=tone_list,
len_list=len_list,
text_list=text_list,
disable_tqdm=disable_tqdm,
use_lora=use_lora,
**sampling_kwargs,
)
def example_usage():
#cfg.device = "cuda"
#cfg.trainer.backend = "local"
from functools import partial
from einops import repeat
from tqdm import tqdm
from ..emb.qnt import decode_to_file, unload_model, trim_random, repeat_extend_audio, concat_audio, merge_audio
from ..data import _load_artifact
from ..engines import Engine, Engines
from ..utils import ml
from ..utils import setup_logging
import numpy as np
import re
# cfg.model.experimental.masking_train_p = 0.5
cfg.hyperparameters.batch_size = 1
cfg.hyperparameters.gradient_accumulation_steps = 1
cfg.model.experimental.use_raw_text_p = 0
setup_logging()
def load_artifact( path ):
audio, metadata = _load_artifact(path, return_metadata=True)
audio = audio.to(cfg.device)
text = torch.tensor( cfg.tokenizer.encode( metadata["phonemes"] ) ).to(dtype=torch.uint8, device=cfg.device)
return text, audio
text, audio = load_artifact(f"./data/qnt.{cfg.audio_backend_extension}")
batch_size = cfg.hyperparameters.batch_size
phns_list = [ text ] * batch_size
proms_list = [ audio[:int(cfg.dataset.frames_per_second), :] ] * batch_size
resps_list = [ audio[:int(cfg.dataset.frames_per_second * 4), :] ] * batch_size
kwargs = {
'n_audio_tokens': cfg.model.audio_tokens,
'd_model': cfg.model.dim,
'd_ffn': cfg.model.ffn,
'n_heads': cfg.model.heads,
'n_layers': cfg.model.layers,
'n_experts': cfg.model.experts,
'p_dropout': 0.1,
'config': cfg.model
}
bos_id, space_id, eos_id = cfg.tokenizer.encode( " " )
available_tasks = [] + (["tts-ar"] if "ar" in cfg.model.capabilities else []) + (["tts-nar"] if "len" in cfg.model.capabilities else [])
if cfg.model.experimental.masking_train_p == 0:
available_tasks = ["tts-ar"]
elif cfg.model.experimental.masking_train_p == 1:
available_tasks = ["tts-nar"]
model = AR_NAR_V2(**kwargs).to(cfg.device)
steps = 250 # // batch_size
optimizer = cfg.hyperparameters.optimizer.lower() if cfg.yaml_path is not None else "prodigy"
scheduler = cfg.hyperparameters.scheduler.lower() if cfg.yaml_path is not None else ""
learning_rate = cfg.hyperparameters.learning_rate if cfg.yaml_path is not None else None
params = {
"params": model.parameters()
}
if cfg.optimizations.dadaptation:
# do not combine the two
if scheduler == "schedulefree":
scheduler = ""
learning_rate = 1.0
if optimizer == "prodigy":
if learning_rate is None:
learning_rate = 1.0
optimizer = ml.Prodigy
elif optimizer == "adagrad":
if learning_rate is None:
learning_rate = 1.0e-2
optimizer = ml.Adagrad
elif optimizer == "adamw":
if learning_rate is None:
learning_rate = 1.0e-4
optimizer = ml.AdamW
elif optimizer == "sdg":
if learning_rate is None:
learning_rate = 1.0e-4
optimizer = ml.SGD
elif optimizer == "apollo":
if learning_rate is None:
learning_rate = 0.01
optimizer = ml.Apollo
params["params"] = [
{'params': params, 'rank': 1, 'proj': 'random', 'scale_type': 'tensor', 'scale': 128,'update_proj_gap': 200, 'proj_type': 'std'}
]
elif optimizer == "muon":
optimizer = ml.Muon
muon_params = [ param for name, param in model.model.named_parameters() if param.ndim >= 2 ]
adamw_params = [ param for name, param in model.model.named_parameters() if param.ndim < 2 ]
adamw_params += [ param for name, param in model.named_parameters() if not name.startswith('model.') ]
params["params"] = [
{ "params": muon_params, "muon": True },
{ "params": adamw_params, "muon": False, "betas": (0.95, 0.95), "eps": 1e-8 },
]
elif optimizer == "cosmos":
optimizer = ml.COSMOS
else:
raise ValueError(f"Unrecognized optimizer: {optimizer}")
_logger.info(f"Optimizer: {optimizer}\tLearning rate: {learning_rate}")
params["lr"] = learning_rate
optimizer = optimizer(**params)
if scheduler == "schedulefree":
if isinstance(optimizer, ml.AdamW):
scheduler = ml.schedulefree.AdamWScheduleFree
elif isinstance(optimizer, ml.SGD):
scheduler = ml.schedulefree.SGDScheduleFree
else:
scheduler = None
if scheduler is not None:
_logger.info(f"Scheduler: {scheduler}")
optimizer = scheduler( model.parameters(), lr = learning_rate )
elif cfg.hyperparameters.scheduler:
scheduler_kwargs = {}
if scheduler == "onecycle":
scheduler_class = ml.OneCycleLR
scheduler_kwargs["max_lr"] = params['lr']
elif scheduler == "cosineannealing":
scheduler_class = ml.CosineAnnealingLR
elif scheduler == "noam":
scheduler_class = ml.NoamLR
scheduler_kwargs["d_model"] = model.d_model
scheduler_kwargs["warmup_steps"] = cfg.hyperparameters.warmup_steps
elif scheduler == "warmup":
scheduler_class = ml.WarmupLR
scheduler_kwargs["warmup_steps"] = cfg.hyperparameters.warmup_steps
else:
raise ValueError(f'Scheduler specified not implemented: {cfg.hyperparameters.scheduler}')
scheduler_kwargs.update(cfg.hyperparameters.scheduler_params)
scheduler = scheduler_class(
optimizer,
**scheduler_kwargs,
)
if isinstance(scheduler, str):
scheduler = None
if cfg.optimizations.replace and cfg.optimizations.linear:
model = ml.replace_linear( model )
if cfg.optimizations.replace and cfg.optimizations.embedding:
model = ml.replace_embedding( model )
"""
cfg.optimizations.model_offloading = {
"devices": ["cuda:0", "cpu"],
# "limits": [ 0.9, -1 ],
"assign": [[ f'layers.{i}.' for i in range(0,10) ], [ f'layers.{i}.' for i in range(11,12) ] + [ "model.norm" ]],
# "limits": [ 256 * (1024 ** 2), -1 ]
}
"""
engine = Engine(model=model, optimizer=optimizer, lr_scheduler=scheduler)
engines = Engines({"ar+nar": engine})
engines.setup()
"""
if cfg.optimizations.model_offloading:
model = ml.offload_model( model, policy=cfg.optimizations.model_offloading )
"""
"""
torch.save( {
'module': model.state_dict()
}, f"./data/{cfg.model.arch_type}.pth" )
"""
_logger.info(f"AR+NAR ({cfg.model.arch_type}, {cfg.audio_backend}) parameter count: {sum(p.numel() for p in model.parameters() if p.requires_grad)}")
@torch.no_grad()
def sample_data(t=None):
if isinstance(t, list):
tasks = t
texts = [ phns_list[0].to(cfg.device) if task not in text_task else None for i, task in enumerate( tasks ) ]
proms = [ proms_list[0].to(cfg.device) if task not in text_task else [ "stt" ] for i, task in enumerate( tasks ) ]
resps = [ None if task not in text_task else resps_list[0].to(cfg.device) for i, task in enumerate( tasks ) ]
return texts, proms, resps, tasks
texts = []
proms = []
resps = []
tasks = []
for i in range(batch_size):
task = random.choice(available_tasks) if t is None else t
text = phns_list[i].to(cfg.device)
prom = proms_list[i].to(cfg.device)
resp = resps_list[i].to(cfg.device)
# do nothing
if task == "stt":
prom = [ task ]
else:
task = "tts" # if random.random() > 0.1 or "len" not in cfg.model.capabilities else "len"
texts.append( text )
proms.append( prom )
resps.append( resp )
tasks.append( task )
return texts, proms, resps, tasks
@torch.inference_mode()
def sample( name, steps=500, task=None ):
engine.eval()
phns_list, proms_list, resp_list, task_list = sample_data( task )
if task == "tts-nar":
# len_list = engine( phns_list=phns_list, proms_list=proms_list, task_list=["len"], max_steps=5, temperature=0.0 )
len_list = [ r.shape[0] for r in resp_list ]
resps_list = engine( phns_list=phns_list, proms_list=proms_list, len_list=len_list )
else:
resps_list = engine( phns_list=phns_list, proms_list=proms_list, task_list=["tts"], max_duration=steps, temperature=1.0 )
if resps_list[0].dim() == 1 or resps_list[0].shape[-1] == 1:
resps_list = engine( phns_list=phns_list, proms_list=proms_list, resps_list=resps_list, temperature=0.0 )
for i, o in enumerate(resps_list):
print( o.shape, o )
_ = decode_to_file(o.to(dtype=torch.int32), f"data/{cfg.model.arch_type}.{cfg.audio_backend}.{i}.{name}.{task}.wav", device=cfg.device)
unload_model()
def train():
engine.train()
t = trange(steps)
for i in t:
texts, proms, resps, tasks = sample_data()
stats = {"step": i, "lr": engine.get_lr()[0]}
with torch.autograd.set_detect_anomaly(cfg.trainer.detect_grad_anomaly):
stats |= engine.traverse(phns_list=texts, proms_list=proms, resps_list=resps, task_list=tasks, training=True)
stats |= {"grad_norm": engine.get_global_grad_norm()}
tqdm.write(f"{stats}")
"""
torch.save( {
'module': model.state_dict()
}, f"./data/{cfg.model.arch_type}.pth" )
"""
task = available_tasks[0]
#sample("init", task=task)
train()
"""
if cfg.optimizations.compile:
model = ml.compile_model(model, backend=cfg.optimizations.compile)
"""
for task in available_tasks:
sample("final", task=task)
engines.quit()
if __name__ == "__main__":
example_usage()