summaryrefslogtreecommitdiff
path: root/dtrain/parallelize.rb
diff options
context:
space:
mode:
authorMichael Denkowski <michael.j.denkowski@gmail.com>2012-12-22 16:01:23 -0500
committerMichael Denkowski <michael.j.denkowski@gmail.com>2012-12-22 16:01:23 -0500
commit597d89c11db53e91bc011eab70fd613bbe6453e8 (patch)
tree83c87c07d1ff6d3ee4e3b1626f7eddd49c61095b /dtrain/parallelize.rb
parent65e958ff2678a41c22be7171456a63f002ef370b (diff)
parent201af2acd394415a05072fbd53d42584875aa4b4 (diff)
Merge branch 'master' of git://github.com/redpony/cdec
Diffstat (limited to 'dtrain/parallelize.rb')
-rwxr-xr-xdtrain/parallelize.rb79
1 files changed, 0 insertions, 79 deletions
diff --git a/dtrain/parallelize.rb b/dtrain/parallelize.rb
deleted file mode 100755
index 1d277ff6..00000000
--- a/dtrain/parallelize.rb
+++ /dev/null
@@ -1,79 +0,0 @@
-#!/usr/bin/env ruby
-
-
-if ARGV.size != 5
- STDERR.write "Usage: "
- STDERR.write "ruby parallelize.rb <#shards> <input> <refs> <epochs> <dtrain.ini>\n"
- exit
-end
-
-dtrain_bin = '/home/pks/bin/dtrain_local'
-ruby = '/usr/bin/ruby'
-lplp_rb = '/home/pks/mt/cdec-dtrain/dtrain/hstreaming/lplp.rb'
-lplp_args = 'l2 select_k 100000'
-gzip = '/bin/gzip'
-
-num_shards = ARGV[0].to_i
-input = ARGV[1]
-refs = ARGV[2]
-epochs = ARGV[3].to_i
-ini = ARGV[4]
-
-
-`mkdir work`
-
-def make_shards(input, refs, num_shards)
- lc = `wc -l #{input}`.split.first.to_i
- shard_sz = lc / num_shards
- leftover = lc % num_shards
- in_f = File.new input, 'r'
- refs_f = File.new refs, 'r'
- shard_in_files = []
- shard_refs_files = []
- 0.upto(num_shards-1) { |shard|
- shard_in = File.new "work/shard.#{shard}.in", 'w+'
- shard_refs = File.new "work/shard.#{shard}.refs", 'w+'
- 0.upto(shard_sz-1) { |i|
- shard_in.write in_f.gets
- shard_refs.write refs_f.gets
- }
- shard_in_files << shard_in
- shard_refs_files << shard_refs
- }
- while leftover > 0
- shard_in_files[-1].write in_f.gets
- shard_refs_files[-1].write refs_f.gets
- leftover -= 1
- end
- (shard_in_files + shard_refs_files).each do |f| f.close end
- in_f.close
- refs_f.close
-end
-
-make_shards input, refs, num_shards
-
-0.upto(epochs-1) { |epoch|
- pids = []
- input_weights = ''
- if epoch > 0 then input_weights = "--input_weights work/weights.#{epoch-1}" end
- weights_files = []
- 0.upto(num_shards-1) { |shard|
- pids << Kernel.fork {
- `#{dtrain_bin} -c #{ini}\
- --input work/shard.#{shard}.in\
- --refs work/shard.#{shard}.refs #{input_weights}\
- --output work/weights.#{shard}.#{epoch}\
- &> work/out.#{shard}.#{epoch}`
- }
- weights_files << "work/weights.#{shard}.#{epoch}"
- }
- pids.each { |pid| Process.wait(pid) }
- cat = File.new('work/weights_cat', 'w+')
- weights_files.each { |f| cat.write File.new(f, 'r').read }
- cat.close
- `#{ruby} #{lplp_rb} #{lplp_args} #{num_shards} < work/weights_cat &> work/weights.#{epoch}`
-}
-
-`rm work/weights_cat`
-`#{gzip} work/*`
-