|
|
""" |
|
|
Main model for using CodecLM. This will combine all the required components |
|
|
and provide easy access to the generation API. |
|
|
""" |
|
|
|
|
|
import typing as tp |
|
|
import warnings |
|
|
import sys |
|
|
import time |
|
|
import torch |
|
|
import torch.nn as nn |
|
|
from torch.nn import functional as F |
|
|
import torchaudio |
|
|
import numpy as np |
|
|
import lightning as pl |
|
|
from torchmetrics.classification import MulticlassAccuracy |
|
|
import pdb |
|
|
from codeclm.models import builders |
|
|
import math |
|
|
from torch.optim import Optimizer |
|
|
from torch.optim.lr_scheduler import _LRScheduler |
|
|
from peft import LoraConfig, get_peft_model |
|
|
from datetime import datetime |
|
|
import os |
|
|
os.environ['TOKENIZERS_PARALLELISM'] = "false" |
|
|
|
|
|
|
|
|
class CodecLM_PL(pl.LightningModule): |
|
|
def __init__(self, cfg, ckpt_path): |
|
|
super().__init__() |
|
|
|
|
|
self.cfg = cfg |
|
|
|
|
|
|
|
|
self.audio_tokenizer = builders.get_audio_tokenizer_model(self.cfg.audio_tokenizer_checkpoint, self.cfg) |
|
|
if self.audio_tokenizer is not None: |
|
|
for param in self.audio_tokenizer.parameters(): |
|
|
param.requires_grad = False |
|
|
if "audio_tokenizer_checkpoint_sep" in self.cfg.keys(): |
|
|
self.seperate_tokenizer = builders.get_audio_tokenizer_model(self.cfg.audio_tokenizer_checkpoint_sep, self.cfg) |
|
|
for param in self.seperate_tokenizer.parameters(): |
|
|
param.requires_grad = False |
|
|
else: |
|
|
self.seperate_tokenizer = None |
|
|
|
|
|
|
|
|
self.audiolm = builders.get_lm_model(self.cfg) |
|
|
print(self.audiolm) |
|
|
|
|
|
checkpoint = torch.load(ckpt_path, map_location='cpu') |
|
|
missing, unexpected = self.load_state_dict(checkpoint, strict=False) |
|
|
print("successfully load pretrained model {}".format(ckpt_path)) |
|
|
|
|
|
self.val_steps = [] |
|
|
self.train_slide_acc = [] |
|
|
self.train_steps = [] |
|
|
self.top1_acc_metric = nn.ModuleList([MulticlassAccuracy( |
|
|
self.audiolm.code_size, |
|
|
top_k=1, |
|
|
average="micro", multidim_average="global", |
|
|
ignore_index=self.cfg.lm.code_size, |
|
|
) for _ in range(self.audiolm.code_depth)]) |
|
|
self.top10_acc_metric = nn.ModuleList([MulticlassAccuracy( |
|
|
self.audiolm.code_size, |
|
|
top_k=10, |
|
|
average="micro", multidim_average="global", |
|
|
ignore_index=self.cfg.lm.code_size, |
|
|
) for _ in range(self.audiolm.code_depth)]) |
|
|
|
|
|
self.epoch = 0 |
|
|
|
|
|
|
|
|
def generate_mask_and_end_token(self, x, sequence_lengths, end_id=16384): |
|
|
batch_size = sequence_lengths.size(0) |
|
|
max_length = x.size(2) |
|
|
|
|
|
|
|
|
if max_length == sequence_lengths.max(): |
|
|
x = F.pad(x, (0, 1), value=end_id) |
|
|
max_length = x.size(2) |
|
|
|
|
|
if max_length <= sequence_lengths.max() + 1: |
|
|
sequence_lengths = sequence_lengths - (sequence_lengths.max()+1 - max_length) |
|
|
|
|
|
|
|
|
x[torch.arange(batch_size), :, sequence_lengths] = end_id |
|
|
sequence_lengths += 1 |
|
|
|
|
|
mask = torch.arange(max_length).expand(batch_size, max_length) < sequence_lengths.unsqueeze(1) |
|
|
mask = mask.to(x.device) |
|
|
mask_3d = mask.unsqueeze(1).expand(batch_size, x.size(1), max_length) |
|
|
x = torch.where(mask_3d, x, end_id+1) |
|
|
return x, mask_3d |
|
|
|
|
|
def get_time(self): |
|
|
|
|
|
now = datetime.now() |
|
|
|
|
|
|
|
|
formatted_now = now.strftime("%Y-%m-%d %H:%M:%S.%f") |
|
|
return formatted_now |
|
|
|
|
|
class CosineLRScheduler(_LRScheduler): |
|
|
"""Cosine LR scheduler. |
|
|
|
|
|
Args: |
|
|
optimizer (Optimizer): Torch optimizer. |
|
|
warmup_steps (int): Number of warmup steps. |
|
|
total_steps (int): Total number of steps. |
|
|
lr_min_ratio (float): Minimum learning rate. |
|
|
cycle_length (float): Cycle length. |
|
|
""" |
|
|
def __init__(self, optimizer: Optimizer, total_steps: int, warmup_steps: int, |
|
|
lr_min_ratio: float = 0.0, cycle_length: float = 1.0): |
|
|
self.warmup_steps = warmup_steps |
|
|
assert self.warmup_steps >= 0 |
|
|
self.total_steps = total_steps |
|
|
assert self.total_steps >= 0 |
|
|
self.lr_min_ratio = lr_min_ratio |
|
|
self.cycle_length = cycle_length |
|
|
super().__init__(optimizer) |
|
|
|
|
|
def _get_sched_lr(self, lr: float, step: int): |
|
|
if step < self.warmup_steps: |
|
|
lr_ratio = step / self.warmup_steps |
|
|
lr = lr_ratio * lr |
|
|
elif step <= self.total_steps: |
|
|
s = (step - self.warmup_steps) / (self.total_steps - self.warmup_steps) |
|
|
lr_ratio = self.lr_min_ratio + 0.5 * (1 - self.lr_min_ratio) * \ |
|
|
(1. + math.cos(math.pi * s / self.cycle_length)) |
|
|
lr = lr_ratio * lr |
|
|
else: |
|
|
lr_ratio = self.lr_min_ratio |
|
|
lr = lr_ratio * lr |
|
|
return lr |
|
|
|
|
|
def get_lr(self): |
|
|
return [self._get_sched_lr(lr, self.last_epoch) for lr in self.base_lrs] |
|
|
|