ymzhang319's picture
init
7f2690b
raw
history blame
No virus
21.4 kB
"""
taken from: https://github.com/karpathy/minGPT/
GPT model:
- the initial stem consists of a combination of token encoding and a positional encoding
- the meat of it is a uniform sequence of Transformer blocks
- each Transformer is a sequential combination of a 1-hidden-layer MLP block and a self-attention block
- all blocks feed into a central residual pathway similar to resnets
- the final decoder is a linear projection into a vanilla Softmax classifier
"""
import math
import logging
import torch
import torch.nn as nn
from torch.nn import functional as F
import sys
sys.path.insert(0, '.') # nopep8
from train import instantiate_from_config
logger = logging.getLogger(__name__)
class GPTConfig:
""" base GPT config, params common to all GPT versions """
embd_pdrop = 0.1
resid_pdrop = 0.1
attn_pdrop = 0.1
def __init__(self, vocab_size, block_size, **kwargs):
self.vocab_size = vocab_size
self.block_size = block_size
for k,v in kwargs.items():
setattr(self, k, v)
class GPT1Config(GPTConfig):
""" GPT-1 like network roughly 125M params """
n_layer = 12
n_head = 12
n_embd = 768
class GPT2Config(GPTConfig):
""" GPT-2 like network roughly 1.5B params """
# TODO
class CausalSelfAttention(nn.Module):
"""
A vanilla multi-head masked self-attention layer with a projection at the end.
It is possible to use torch.nn.MultiheadAttention here but I am including an
explicit implementation here to show that there is nothing too scary here.
"""
def __init__(self, config):
super().__init__()
assert config.n_embd % config.n_head == 0
# key, query, value projections for all heads
self.key = nn.Linear(config.n_embd, config.n_embd)
self.query = nn.Linear(config.n_embd, config.n_embd)
self.value = nn.Linear(config.n_embd, config.n_embd)
# regularization
self.attn_drop = nn.Dropout(config.attn_pdrop)
self.resid_drop = nn.Dropout(config.resid_pdrop)
# output projection
self.proj = nn.Linear(config.n_embd, config.n_embd)
# causal mask to ensure that attention is only applied to the left in the input sequence
mask = torch.tril(torch.ones(config.block_size,
config.block_size))
if hasattr(config, "n_unmasked"):
mask[:config.n_unmasked, :config.n_unmasked] = 1
self.register_buffer("mask", mask.view(1, 1, config.block_size, config.block_size))
self.n_head = config.n_head
def forward(self, x, layer_past=None):
B, T, C = x.size()
# calculate query, key, values for all heads in batch and move head forward to be the batch dim
k = self.key(x).view(B, T, self.n_head, C // self.n_head).transpose(1, 2) # (B, nh, T, hs)
q = self.query(x).view(B, T, self.n_head, C // self.n_head).transpose(1, 2) # (B, nh, T, hs)
v = self.value(x).view(B, T, self.n_head, C // self.n_head).transpose(1, 2) # (B, nh, T, hs)
# causal self-attention; Self-attend: (B, nh, T, hs) x (B, nh, hs, T) -> (B, nh, T, T)
att = (q @ k.transpose(-2, -1)) * (1.0 / math.sqrt(k.size(-1)))
att = att.masked_fill(self.mask[:,:,:T,:T] == 0, float('-inf'))
att = F.softmax(att, dim=-1)
y = self.attn_drop(att) @ v # (B, nh, T, T) x (B, nh, T, hs) -> (B, nh, T, hs)
y = y.transpose(1, 2).contiguous().view(B, T, C) # re-assemble all head outputs side by side
# output projection
y = self.resid_drop(self.proj(y))
return y, att
class Block(nn.Module):
""" an unassuming Transformer block """
def __init__(self, config):
super().__init__()
self.ln1 = nn.LayerNorm(config.n_embd)
self.ln2 = nn.LayerNorm(config.n_embd)
self.attn = CausalSelfAttention(config)
self.mlp = nn.Sequential(
nn.Linear(config.n_embd, 4 * config.n_embd),
nn.GELU(), # nice
nn.Linear(4 * config.n_embd, config.n_embd),
nn.Dropout(config.resid_pdrop),
)
def forward(self, x):
# x = x + self.attn(self.ln1(x))
# x is a tuple (x, attention)
x, _ = x
res = x
x = self.ln1(x)
x, att = self.attn(x)
x = res + x
x = x + self.mlp(self.ln2(x))
return x, att
class GPT(nn.Module):
""" the full GPT language model, with a context size of block_size """
def __init__(self, vocab_size, block_size, n_layer=12, n_head=8, n_embd=256,
embd_pdrop=0., resid_pdrop=0., attn_pdrop=0., n_unmasked=0):
super().__init__()
config = GPTConfig(vocab_size=vocab_size, block_size=block_size,
embd_pdrop=embd_pdrop, resid_pdrop=resid_pdrop, attn_pdrop=attn_pdrop,
n_layer=n_layer, n_head=n_head, n_embd=n_embd,
n_unmasked=n_unmasked)
# input embedding stem
self.tok_emb = nn.Embedding(config.vocab_size, config.n_embd)
self.pos_emb = nn.Parameter(torch.zeros(1, config.block_size, config.n_embd))
self.drop = nn.Dropout(config.embd_pdrop)
# transformer
self.blocks = nn.Sequential(*[Block(config) for _ in range(config.n_layer)])
# decoder head
self.ln_f = nn.LayerNorm(config.n_embd)
self.head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
self.block_size = config.block_size
self.apply(self._init_weights)
self.config = config
logger.info("number of parameters: %e", sum(p.numel() for p in self.parameters()))
def get_block_size(self):
return self.block_size
def _init_weights(self, module):
if isinstance(module, (nn.Linear, nn.Embedding)):
module.weight.data.normal_(mean=0.0, std=0.02)
if isinstance(module, nn.Linear) and module.bias is not None:
module.bias.data.zero_()
elif isinstance(module, nn.LayerNorm):
module.bias.data.zero_()
module.weight.data.fill_(1.0)
def forward(self, idx, embeddings=None, targets=None):
# forward the GPT model
token_embeddings = self.tok_emb(idx) # each index maps to a (learnable) vector
if embeddings is not None: # prepend explicit embeddings
token_embeddings = torch.cat((embeddings, token_embeddings), dim=1)
t = token_embeddings.shape[1]
assert t <= self.block_size, "Cannot forward, model block size is exhausted."
position_embeddings = self.pos_emb[:, :t, :] # each position maps to a (learnable) vector
x = self.drop(token_embeddings + position_embeddings)
# returns only last layer attention
# giving tuple (x, None) just because Sequential takes a single input but outputs two (x, atttention).
# att is (B, H, T, T)
x, att = self.blocks((x, None))
x = self.ln_f(x)
logits = self.head(x)
# if we are given some desired targets also calculate the loss
loss = None
if targets is not None:
loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1))
return logits, loss, att
class DummyGPT(nn.Module):
# for debugging
def __init__(self, add_value=1):
super().__init__()
self.add_value = add_value
def forward(self, idx):
raise NotImplementedError('Model should output attention')
return idx + self.add_value, None
class CodeGPT(nn.Module):
"""Takes in semi-embeddings"""
def __init__(self, vocab_size, block_size, in_channels, n_layer=12, n_head=8, n_embd=256,
embd_pdrop=0., resid_pdrop=0., attn_pdrop=0., n_unmasked=0):
super().__init__()
config = GPTConfig(vocab_size=vocab_size, block_size=block_size,
embd_pdrop=embd_pdrop, resid_pdrop=resid_pdrop, attn_pdrop=attn_pdrop,
n_layer=n_layer, n_head=n_head, n_embd=n_embd,
n_unmasked=n_unmasked)
# input embedding stem
self.tok_emb = nn.Linear(in_channels, config.n_embd)
self.pos_emb = nn.Parameter(torch.zeros(1, config.block_size, config.n_embd))
self.drop = nn.Dropout(config.embd_pdrop)
# transformer
self.blocks = nn.Sequential(*[Block(config) for _ in range(config.n_layer)])
# decoder head
self.ln_f = nn.LayerNorm(config.n_embd)
self.head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
self.block_size = config.block_size
self.apply(self._init_weights)
self.config = config
logger.info("number of parameters: %e", sum(p.numel() for p in self.parameters()))
def get_block_size(self):
return self.block_size
def _init_weights(self, module):
if isinstance(module, (nn.Linear, nn.Embedding)):
module.weight.data.normal_(mean=0.0, std=0.02)
if isinstance(module, nn.Linear) and module.bias is not None:
module.bias.data.zero_()
elif isinstance(module, nn.LayerNorm):
module.bias.data.zero_()
module.weight.data.fill_(1.0)
elif isinstance(module, (nn.Conv1d, nn.Conv2d)):
torch.nn.init.xavier_uniform(module.weight)
if module.bias is not None:
module.bias.data.fill_(0.01)
def forward(self, idx, embeddings=None, targets=None):
raise NotImplementedError('Model should output attention')
# forward the GPT model
token_embeddings = self.tok_emb(idx) # each index maps to a (learnable) vector
if embeddings is not None: # prepend explicit embeddings
token_embeddings = torch.cat((embeddings, token_embeddings), dim=1)
t = token_embeddings.shape[1]
assert t <= self.block_size, "Cannot forward, model block size is exhausted."
position_embeddings = self.pos_emb[:, :t, :] # each position maps to a (learnable) vector
x = self.drop(token_embeddings + position_embeddings)
x = self.blocks(x)
x = self.ln_f(x)
logits = self.head(x)
# if we are given some desired targets also calculate the loss
loss = None
if targets is not None:
loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1))
return logits, loss
class GPTFeats(GPT):
def __init__(self, feat_embedding_config, GPT_config):
super().__init__(**GPT_config)
# patching the config by removing the default parameters for Conv1d
if feat_embedding_config.target.split('.')[-1] in ['LSTM', 'GRU']:
for p in ['in_channels', 'out_channels', 'padding', 'kernel_size']:
if p in feat_embedding_config.params:
feat_embedding_config.params.pop(p)
self.embedder = instantiate_from_config(config=feat_embedding_config)
if isinstance(self.embedder, nn.Linear):
print('Checkout cond_transformer.configure_optimizers. Make sure not to use decay with Linear')
def forward(self, idx, feats):
if isinstance(self.embedder, nn.Linear):
feats = feats.permute(0, 2, 1)
feats = self.embedder(feats)
elif isinstance(self.embedder, (nn.LSTM, nn.GRU)):
feats = feats.permute(0, 2, 1)
feats, _ = self.embedder(feats)
elif isinstance(self.embedder, (nn.Conv1d, nn.Identity)):
# (B, D', T) <- (B, D, T)
feats = self.embedder(feats)
# (B, T, D') <- (B, T, D)
feats = feats.permute(0, 2, 1)
else:
raise NotImplementedError
# calling forward from super
return super().forward(idx, embeddings=feats)
class GPTFeatsPosEnc(GPT):
def __init__(self, feat_embedding_config, GPT_config, PosEnc_config):
super().__init__(**GPT_config)
# patching the config by removing the default parameters for Conv1d
if feat_embedding_config.target.split('.')[-1] in ['LSTM', 'GRU']:
for p in ['in_channels', 'out_channels', 'padding', 'kernel_size']:
if p in feat_embedding_config.params:
feat_embedding_config.params.pop(p)
self.embedder = instantiate_from_config(config=feat_embedding_config)
self.pos_emb_vis = nn.Parameter(torch.zeros(1, PosEnc_config['block_size_v'], PosEnc_config['n_embd']))
self.pos_emb_aud = nn.Parameter(torch.zeros(1, PosEnc_config['block_size_a'], PosEnc_config['n_embd']))
if isinstance(self.embedder, nn.Linear):
print('Checkout cond_transformer.configure_optimizers. Make sure not to use decay with Linear')
def foward(self, idx, feats):
if isinstance(self.embedder, nn.Linear):
feats = feats.permute(0, 2, 1)
feats = self.embedder(feats)
elif isinstance(self.embedder, (nn.LSTM, nn.GRU)):
feats = feats.permute(0, 2, 1)
feats, _ = self.embedder(feats)
elif isinstance(self.embedder, (nn.Conv1d, nn.Identity)):
# (B, D', T) <- (B, D, T)
feats = self.embedder(feats)
# (B, T, D') <- (B, T, D)
feats = feats.permute(0, 2, 1)
else:
raise NotImplementedError
# calling forward from super
# forward the GPT model
token_embeddings = self.tok_emb(idx) # each index maps to a (learnable) vector
if feats is not None: # prepend explicit feats
token_embeddings = torch.cat((feats, token_embeddings), dim=1)
t = token_embeddings.shape[1]
assert t <= self.block_size, "Cannot forward, model block size is exhausted."
vis_t = self.pos_emb_vis.shape[1]
position_embeddings = torch.cat([self.pos_emb_vis, self.pos_emb_aud[:, :t-vis_t, :]])
x = self.drop(token_embeddings + position_embeddings)
# returns only last layer attention
# giving tuple (x, None) just because Sequential takes a single input but outputs two (x, atttention).
# att is (B, H, T, T)
x, att = self.blocks((x, None))
x = self.ln_f(x)
logits = self.head(x)
# if we are given some desired targets also calculate the loss
loss = None
return logits, loss, att
class GPTClass(GPT):
def __init__(self, token_embedding_config, GPT_config):
super().__init__(**GPT_config)
self.embedder = instantiate_from_config(config=token_embedding_config)
def forward(self, idx, token):
token = self.embedder(token)
# calling forward from super
return super().forward(idx, embeddings=token)
class GPTFeatsClass(GPT):
def __init__(self, feat_embedding_config, token_embedding_config, GPT_config):
super().__init__(**GPT_config)
# patching the config by removing the default parameters for Conv1d
if feat_embedding_config.target.split('.')[-1] in ['LSTM', 'GRU']:
for p in ['in_channels', 'out_channels', 'padding', 'kernel_size']:
if p in feat_embedding_config.params:
feat_embedding_config.params.pop(p)
self.feat_embedder = instantiate_from_config(config=feat_embedding_config)
self.cls_embedder = instantiate_from_config(config=token_embedding_config)
if isinstance(self.feat_embedder, nn.Linear):
print('Checkout cond_transformer.configure_optimizers. Make sure not to use decay with Linear')
def forward(self, idx, feats_token_dict: dict):
feats = feats_token_dict['feature']
token = feats_token_dict['target']
# Features. Output size: (B, T, D')
if isinstance(self.feat_embedder, nn.Linear):
feats = feats.permute(0, 2, 1)
feats = self.feat_embedder(feats)
elif isinstance(self.feat_embedder, (nn.LSTM, nn.GRU)):
feats = feats.permute(0, 2, 1)
feats, _ = self.feat_embedder(feats)
elif isinstance(self.feat_embedder, (nn.Conv1d, nn.Identity)):
# (B, D', T) <- (B, D, T)
feats = self.feat_embedder(feats)
# (B, T, D') <- (B, T, D)
feats = feats.permute(0, 2, 1)
else:
raise NotImplementedError
# Class. Output size: (B, 1, D')
token = self.cls_embedder(token)
# Concat
condition_emb = torch.cat([feats, token], dim=1)
# calling forward from super
return super().forward(idx, embeddings=condition_emb)
#### sampling utils
def top_k_logits(logits, k):
v, ix = torch.topk(logits, k)
out = logits.clone()
out[out < v[:, [-1]]] = -float('Inf')
return out
@torch.no_grad()
def sample(model, x, steps, temperature=1.0, sample=False, top_k=None):
"""
take a conditioning sequence of indices in x (of shape (b,t)) and predict the next token in
the sequence, feeding the predictions back into the model each time. Clearly the sampling
has quadratic complexity unlike an RNN that is only linear, and has a finite context window
of block_size, unlike an RNN that has an infinite context window.
"""
block_size = model.get_block_size()
model.eval()
for k in range(steps):
x_cond = x if x.size(1) <= block_size else x[:, -block_size:] # crop context if needed
raise NotImplementedError('v-iashin: the model outputs (logits, loss, attention)')
logits, _ = model(x_cond)
# pluck the logits at the final step and scale by temperature
logits = logits[:, -1, :] / temperature
# optionally crop probabilities to only the top k options
if top_k is not None:
logits = top_k_logits(logits, top_k)
# apply softmax to convert to probabilities
probs = F.softmax(logits, dim=-1)
# sample from the distribution or take the most likely
if sample:
ix = torch.multinomial(probs, num_samples=1)
else:
_, ix = torch.topk(probs, k=1, dim=-1)
# append to the sequence and continue
x = torch.cat((x, ix), dim=1)
return x
#### clustering utils
class KMeans(nn.Module):
def __init__(self, ncluster=512, nc=3, niter=10):
super().__init__()
self.ncluster = ncluster
self.nc = nc
self.niter = niter
self.shape = (3,32,32)
self.register_buffer("C", torch.zeros(self.ncluster,nc))
self.register_buffer('initialized', torch.tensor(0, dtype=torch.uint8))
def is_initialized(self):
return self.initialized.item() == 1
@torch.no_grad()
def initialize(self, x):
N, D = x.shape
assert D == self.nc, D
c = x[torch.randperm(N)[:self.ncluster]] # init clusters at random
for i in range(self.niter):
# assign all pixels to the closest codebook element
a = ((x[:, None, :] - c[None, :, :])**2).sum(-1).argmin(1)
# move each codebook element to be the mean of the pixels that assigned to it
c = torch.stack([x[a==k].mean(0) for k in range(self.ncluster)])
# re-assign any poorly positioned codebook elements
nanix = torch.any(torch.isnan(c), dim=1)
ndead = nanix.sum().item()
print('done step %d/%d, re-initialized %d dead clusters' % (i+1, self.niter, ndead))
c[nanix] = x[torch.randperm(N)[:ndead]] # re-init dead clusters
self.C.copy_(c)
self.initialized.fill_(1)
def forward(self, x, reverse=False, shape=None):
if not reverse:
# flatten
bs,c,h,w = x.shape
assert c == self.nc
x = x.reshape(bs,c,h*w,1)
C = self.C.permute(1,0)
C = C.reshape(1,c,1,self.ncluster)
a = ((x-C)**2).sum(1).argmin(-1) # bs, h*w indices
return a
else:
# flatten
bs, HW = x.shape
"""
c = self.C.reshape( 1, self.nc, 1, self.ncluster)
c = c[bs*[0],:,:,:]
c = c[:,:,HW*[0],:]
x = x.reshape(bs, 1, HW, 1)
x = x[:,3*[0],:,:]
x = torch.gather(c, dim=3, index=x)
"""
x = self.C[x]
x = x.permute(0,2,1)
shape = shape if shape is not None else self.shape
x = x.reshape(bs, *shape)
return x
if __name__ == '__main__':
import torch
from omegaconf import OmegaConf
import numpy as np
from tqdm import tqdm
device = torch.device('cuda:2')
torch.cuda.set_device(device)
cfg = OmegaConf.load('./configs/vggsound_transformer.yaml')
model = instantiate_from_config(cfg.model.params.transformer_config)
model = model.to(device)
mel_num = cfg.data.params.mel_num
spec_crop_len = cfg.data.params.spec_crop_len
feat_depth = cfg.data.params.feat_depth
feat_crop_len = cfg.data.params.feat_crop_len
gcd = np.gcd(mel_num, spec_crop_len)
z_idx_size = (2, int(mel_num / gcd) * int(spec_crop_len / gcd))
for i in tqdm(range(300)):
z_indices = torch.randint(0, cfg.model.params.transformer_config.params.GPT_config.vocab_size, z_idx_size).to(device)
c = torch.rand(2, feat_depth, feat_crop_len).to(device)
logits, loss, att = model(z_indices[:, :-1], feats=c)