gchhablani

sudo apt install g++
pip install git+https://github.com/pytorch/fairseq.git@698e3b91ffa832c286c48035bdff78238b0de8ae
git clone https://github.com/wasiahmad/PLBART



# PreTrained
## Download pre-trained checkpoint
cd PLBART/pretrain
bash download.sh

gchhablani

#!/usr/bin/bash

intermediate() {
    local seed=$1
    local ckpt=$2
    local step=$((ckpt/1000))

    local multiberts_ckpt_dir="multiberts-seed-${seed}-${step}k"
    mkdir $multiberts_ckpt_dir
    huggingface-cli repo create --yes ${multiberts_ckpt_dir} --organization google

gchhablani

#!/bin/bash 
hda-verb /dev/snd/hwC0D0 0x20 SET_COEF_INDEX 0x67
hda-verb /dev/snd/hwC0D0 0x20 SET_PROC_COEF 0x3000

gchhablani

#!/bin/bash

cd ~/Downloads

# Google Chrome
wget https://dl.google.com/linux/direct/google-chrome-stable_current_amd64.deb
sudo dpkg -i google-chrome-stable_current_amd64.deb

# Anaconda
wget https://repo.anaconda.com/archive/Anaconda3-2021.05-Linux-x86_64.sh # Change version if needed

gchhablani

inp = torch.from_numpy(np.random.normal(0,1,size=(100,784))).to(device).float()
temp = vae(inp)
temp = temp.data.reshape(100,1,28,28)
grid = torchvision.utils.make_grid(temp,nrow=10)
print(grid.shape)
plt.imshow(grid.to('cpu').permute(1,2,0))
plt.gcf().set_size_inches(20,10)
plt.show()

gchhablani

vae = VAE()
vae.to(device)
criterion = nn.MSELoss()
optimizer = optim.Adamax(vae.parameters(),lr = 1e-4)

l = None
for epoch in range(100):
  for i, data in enumerate(loader,0):
    inputs,classes = data
    inputs,classes = Variable(inputs.resize_(batch_size,784)).to(device),Variable(classes).to(device)

gchhablani

class VAE(nn.Module):
  def __init__(self):
    super(VAE,self).__init__()
    self.encoder = nn.Sequential(nn.Linear(784,128),nn.ReLU(),nn.Linear(128,64),nn.ReLU())
    self.decoder = nn.Sequential(nn.Linear(64,128),nn.ReLU(),nn.Linear(128,784))
    self._mu = nn.Linear(64,64)
    self._log_sigma = nn.Linear(64,64)

  def sampler(self,encoding):
    mu = self._mu(encoding)

gchhablani

inp = torch.from_numpy(np.random.normal(0,1,size=(100,64))).to(device).float()
temp = ae.decoder(inp)
temp = temp.data.reshape(100,1,28,28)
grid = torchvision.utils.make_grid(temp,nrow=10)
print(grid.shape)
plt.imshow(grid.to('cpu').permute(1,2,0))
plt.gcf().set_size_inches(20,10)
plt.show()

gchhablani

ae = AE()
ae.to(device)
criterion = nn.MSELoss()
optimizer = optim.Adamax(ae.parameters(),lr = 1e-4)

l = None
for epoch in range(100):
  for i, data in enumerate(loader,0):
    inputs,classes = data
    inputs,classes = Variable(inputs.resize_(batch_size,784)).to(device),Variable(classes).to(device)

gchhablani

class AE(nn.Module):
  def __init__(self):
    super(AE,self).__init__()
    self.encoder = nn.Sequential(nn.Linear(784,50),nn.ReLU(),nn.Linear(50,50),nn.ReLU())
    self.decoder = nn.Sequential(nn.Linear(14,50),nn.ReLU(),nn.Linear(50,50),nn.ReLU(),nn.Linear(50,784),nn.ReLU())

  def forward(self,inp):
    return self.decoder(self.encoder(inp))