In f0nzie/rTorch: R Bindings to 'PyTorch'

knitr::opts_chunk$set(
  collapse = FALSE,
  results = "markup",
  comment = ""
)

Sources:

• Linear Regression using PyTorch

• Linear Regression with PyTorch

• PyTorch basics - Linear Regression from scratch

library(rTorch)

invisible(torch$manual_seed(0))
device = torch$device('cpu')

# Input (temp, rainfall, humidity)
inputs = np$array(list(
                   list( 73,  67, 43),
                   list( 91,  88, 64),
                   list( 87, 134, 58),
                   list(102,  43, 37),
                   list( 69,  96, 70)), dtype='float32')

# Targets (apples, oranges)
targets = np$array(list(
                    list( 56,  70), 
                    list( 81, 101),
                    list(119, 133),
                    list( 22,  37), 
                    list(103, 119)), dtype='float32')

# Convert inputs and targets to tensors
inputs  = torch$from_numpy(inputs)
targets = torch$from_numpy(targets)
message("tensor shapes: input, targets")
print(inputs$shape); print(targets$shape)

# random numbers for weights and biases. Then convert to double()
torch$set_default_dtype(torch$double)

w = torch$randn(2L, 3L, requires_grad=TRUE)  #$double()
b = torch$randn(2L, requires_grad=TRUE)      #$double()
message("Weight tensor shapes: w, b")
print(w$shape); print(b$shape)

model <- function(x) {
  wt <- w$t()
  return(torch$add(torch$mm(x, wt), b))
}

# MSE loss function
mse = function(t1, t2) {
  diff <- torch$sub(t1, t2)
  mul <- torch$sum(torch$mul(diff, diff))
  return(torch$div(mul, diff$numel()))
}

# Running all together
# Adjust weights and reset gradients
num_epochs <- 100
message("training loop")
for (i in 1:num_epochs) {
  preds = model(inputs)
  loss = mse(preds, targets)
  loss$backward()
  with(torch$no_grad(), {
    w$data <- torch$sub(w$data, torch$mul(w$grad, torch$scalar_tensor(1e-5)))
    b$data <- torch$sub(b$data, torch$mul(b$grad, torch$scalar_tensor(1e-5)))
    w$grad$zero_()
    b$grad$zero_()
  })
}

# Calculate loss
preds = model(inputs)
loss = mse(preds, targets)
cat("loss as tensor = ", "\t"); print(loss)
cat("loss as scalar = ", "\t"); print(loss$item())

# predictions
message("predictions"); print(preds)
# Targets
message("targets"); print(targets)

library(rTorch)

invisible(torch$manual_seed(0))
device = torch$device('cpu')

# Input (temp, rainfall, humidity)
inputs = np$array(list(
                   list( 73,  67, 43),
                   list( 91,  88, 64),
                   list( 87, 134, 58),
                   list(102,  43, 37),
                   list( 69,  96, 70)), dtype='float32')

# Targets (apples, oranges)
targets = np$array(list(
                    list( 56,  70), 
                    list( 81, 101),
                    list(119, 133),
                    list( 22,  37), 
                    list(103, 119)), dtype='float32')

# Convert inputs and targets to tensors
inputs  = torch$from_numpy(inputs)
targets = torch$from_numpy(targets)
message("tensor shapes: input, targets")
print(inputs$shape); print(targets$shape)

# random numbers for weights and biases. Then convert to double()
torch$set_default_dtype(torch$double)

w = torch$randn(2L, 3L, requires_grad=TRUE)  #$double()
b = torch$randn(2L, requires_grad=TRUE)      #$double()
message("Weight tensor shapes: w, b")
print(w$shape); print(b$shape)

model <- function(x) {
  wt <- w$t()
  return(torch$add(torch$mm(x, wt), b))
}

# MSE loss function
mse = function(t1, t2) {
  diff <- torch$sub(t1, t2)
  mul <- torch$sum(torch$mul(diff, diff))
  return(torch$div(mul, diff$numel()))
}

num_epochs <- 100
message("training loop")
for (i in 1:num_epochs) {
  preds = model(inputs)
  loss = mse(preds, targets)
  loss$backward()
  with(torch$no_grad(), {
    w$data <- torch$sub(w$data, torch$mul(w$grad, torch$scalar_tensor(1e-5)))
    b$data <- torch$sub(b$data, torch$mul(b$grad, torch$scalar_tensor(1e-5)))

    w$grad$zero_()
    b$grad$zero_()
  })
}

# Calculate loss
preds = model(inputs)
loss = mse(preds, targets)
cat("loss as tensor = ", "\t"); print(loss)
cat("loss as scalar = ", "\t"); print(loss$item())

# predictions
message("predictions"); print(preds)
# Targets
message("targets"); print(targets)

f0nzie/rTorch documentation built on Oct. 28, 2021, 5:40 a.m.