From eecc4cbf5a6e28f910130cd4a58444e9d4701ea9 Mon Sep 17 00:00:00 2001
From: Kenneth Heafield <github@kheafield.com>
Date: Mon, 27 Jan 2014 17:42:19 -0800
Subject: KenLM 5cc905bc2d214efa7de2db56a9a672b749a95591

---
 klm/lm/builder/pipeline.cc | 1 +
 1 file changed, 1 insertion(+)

(limited to 'klm/lm/builder/pipeline.cc')

diff --git a/klm/lm/builder/pipeline.cc b/klm/lm/builder/pipeline.cc
index b89ea6ba..44a2313c 100644
--- a/klm/lm/builder/pipeline.cc
+++ b/klm/lm/builder/pipeline.cc
@@ -226,6 +226,7 @@ void CountText(int text_file /* input */, int vocab_file /* output */, Master &m
 
   util::stream::Sort<SuffixOrder, AddCombiner> sorter(chain, config.sort, SuffixOrder(config.order), AddCombiner());
   chain.Wait(true);
+  std::cerr << "Unigram tokens " << token_count << " types " << type_count << std::endl;
   std::cerr << "=== 2/5 Calculating and sorting adjusted counts ===" << std::endl;
   master.InitForAdjust(sorter, type_count);
 }
-- 
cgit v1.2.3