train.lua

-- Copyright 2016 Anurag Ranjan and the Max Planck Gesellschaft.
-- All rights reserved.
-- This software is provided for research purposes only.  
-- By using this software you agree to the terms of the license file 
-- in the root folder.
-- For commercial use, please contact ps-license@tue.mpg.de.
--
--  Copyright (c) 2014, Facebook, Inc.
--  All rights reserved.
--
--  This source code is licensed under the BSD-style license found in the
--  LICENSE file in the root directory of this source tree. An additional grant
--  of patent rights can be found in the PATENTS file in the same directory.
--
require 'optim'

--[[
   1. Setup SGD optimization state and learning rate schedule
   2. Create loggers.
   3. train - this function handles the high-level training loop,
              i.e. load data, train model, save model and state to disk
   4. trainBatch - Used by train() to train a single batch after the data is loaded.
]]--

-- Setup a reused optimization state (for sgd). If needed, reload it from disk
local optimState = {
    learningRate = opt.LR,
    learningRateDecay = 0.0,
    momentum = opt.momentum,
    dampening = 0.0,
    weightDecay = opt.weightDecay
}

if opt.optimState ~= 'none' then
    assert(paths.filep(opt.optimState), 'File not found: ' .. opt.optimState)
    print('Loading optimState from file: ' .. opt.optimState)
    optimState = torch.load(opt.optimState)
end

-- Learning rate annealing schedule. We will build a new optimizer for
-- each epoch.
--
-- By default we follow a known recipe for a 55-epoch training. If
-- the learningRate command-line parameter has been specified, though,
-- we trust the user is doing something manual, and will use her
-- exact settings for all optimization.
--
-- Return values:
--    diff to apply to optimState,
--    true IFF this is the first epoch of a new regime
local function paramsForEpoch(epoch)
    if opt.LR ~= 0.0 then -- if manually specified
        return { }
    end
    local regimes
    if opt.level % 2 == 0 then
	 regimes= {
          -- start, end,       LR,   WD,
          { 1,    35,   5e-4,   0 },
          { 36,   50,   1e-4,   0 },
          { 51,  1e8,   1e-5,   0 },
       }
    else
	 regimes = {
	-- start, end,       LR,   WD,
          { 1,    30,   5e-4,   0 },
          { 31,   40,   1e-4,   0 },
          { 41,  1e8,   1e-5,   0 },   
	 }
    end

    for _, row in ipairs(regimes) do
        if epoch >= row[1] and epoch <= row[2] then
            return { learningRate=row[3], weightDecay=row[4] }, epoch == row[1]
        end
    end
end

-- 2. Create loggers.
trainLogger = optim.Logger(paths.concat(opt.save, 'train.log'))
local batchNumber
local top1_epoch, loss_epoch

-- 3. train - this function handles the high-level training loop,
--            i.e. load data, train model, save model and state to disk
function train()
   print('==> doing epoch on training data:')
   print("==> online epoch # " .. epoch)

   local params, newRegime = paramsForEpoch(epoch)
   if newRegime then
      optimState = {
         learningRate = params.learningRate,
         learningRateDecay = 0.0,
         momentum = opt.momentum,
         dampening = 0.0,
         weightDecay = params.weightDecay
      }
   end
   batchNumber = 0
   cutorch.synchronize()

   -- set the dropouts to training mode
   model:training()

   local tm = torch.Timer()
   top1_epoch = 0
   loss_epoch = 0
   for i=1,opt.epochSize do
      -- queue jobs to data-workers
      donkeys:addjob(
         -- the job callback (runs in data-worker thread)
         function()
            local inputs, labels = trainLoader:sample(opt.batchSize)
            return inputs, labels
         end,
         -- the end callback (runs in the main thread)
         trainBatch
      )
   end

   donkeys:synchronize()
   cutorch.synchronize()

   top1_epoch = top1_epoch * 100 / (opt.batchSize * opt.epochSize)
   loss_epoch = loss_epoch / opt.epochSize

   trainLogger:add{
      ['% top1 accuracy (train set)'] = top1_epoch,
      ['avg loss (train set)'] = loss_epoch
   }
   print(string.format('Epoch: [%d][TRAINING SUMMARY] Total Time(s): %.2f\t'
                          .. 'average loss (per batch): %.2f \t '
                          .. 'accuracy(%%):\t top-1 %.2f\t',
                       epoch, tm:time().real, loss_epoch, top1_epoch))
   print('\n')

   -- save model
   collectgarbage()

   -- clear the intermediate states in the model before saving to disk
   -- this saves lots of disk space
   model:clearState()
   saveDataParallel(paths.concat(opt.save, 'model_' .. epoch .. '.t7'), model) -- defined in util.lua
   torch.save(paths.concat(opt.save, 'optimState_' .. epoch .. '.t7'), optimState)
end -- of train()
-------------------------------------------------------------------------------------------
-- GPU inputs (preallocate)
local inputs = torch.CudaTensor()
local labels = torch.CudaTensor()

local timer = torch.Timer()
local dataTimer = torch.Timer()

local parameters, gradParameters = model:getParameters()
assert(opt.batchSize % opt.iterSize==0,'can not divide big batch into small batches')
local batchsize = opt.batchSize/opt.iterSize
-- 4. trainBatch - Used by train() to train a single batch after the data is loaded.
function trainBatch(inputsCPU, labelsCPU)
   cutorch.synchronize()
   collectgarbage()
   local dataLoadingTime = dataTimer:time().real
   timer:reset()
   local err, outputs
   feval = function(x)
      model:zeroGradParameters()
	  for i = 1, opt.iterSize, 1 do
	     	inputs:resize(batchsize, inputsCPU:size(2), inputsCPU:size(3),inputsCPU:size(4)):copy(inputsCPU[{{(i-1)*batchsize+1, i*batchsize},{},{},{}}])
		labels:resize(batchsize, labelsCPU:size(2), labelsCPU:size(3), labelsCPU:size(4)):copy(labelsCPU[{{(i-1)*batchsize+1, i*batchsize},{},{},{}}])
         	outputs = model:forward(inputs)
         	err = criterion:forward(outputs, labels)
         	local gradOutputs = criterion:backward(outputs, labels)
         	model:backward(inputs, gradOutputs)
	  end
	  print(gradParameters:size())
      return err, gradParameters/opt.iterSize
   end

   if opt.optimizer == 'adam' then
      optim.adam(feval, parameters, optimState)
   elseif opt.optimizer == 'sgd' then
      optim.sgd(feval, parameters, optimState)
   else
      error("Specify Optimizer")
   end

   -- DataParallelTable's syncParameters
   if model.needsSync then
      model:syncParameters()
   end
   
   cutorch.synchronize()
   batchNumber = batchNumber + 1
   loss_epoch = loss_epoch + err

   -- Calculate top-1 error, and print information
   print(('Epoch: [%d][%d/%d]\tTime %.3f Err %.4f LR %.0e DataLoadingTime %.3f'):format(
          epoch, batchNumber, opt.epochSize, timer:time().real, err,
          optimState.learningRate, dataLoadingTime))

   dataTimer:reset()
end