diff --git a/tools/interpolation_to_kmm.py b/tools/interpolation_to_kmm.py
new file mode 100644
index 0000000..2e05c36
--- /dev/null
+++ b/tools/interpolation_to_kmm.py
@@ -0,0 +1,211 @@
+#!/usr/bin/python3
+
+import os
+import sys
+from argparse import ArgumentParser
+
+items = []
+arraycount = {}
+unigram = {}
+
+phrases = {}
+seen_in_unigram = set()
+
+def writeRestUnigram(outfile):
+    # Write out
+    for token, count in arraycount.items():
+        if token in seen_in_unigram:
+            pass
+        else:
+            string = phrases[token]
+            count = unigram[token] if token in unigram else 0
+            inline = "\item {0} {1} count {2}".format(token, string, count)
+            outline = writeUnigramItem(inline)
+            outfile.writelines([outline, os.linesep])
+    for token, freq in unigram.items():
+        if token in seen_in_unigram:
+            pass
+        else:
+            string = phrases[token]
+            count = unigram[token]
+            inline = "\item {0} {1} count {2}".format(token, string, count)
+            outline = writeUnigramItem(inline)
+            outfile.writelines([outline, os.linesep])
+
+def parseUnigramItem(oneline):
+    assert oneline.startswith("\\item")
+
+
+def parseBigramItem(oneline):
+    assert oneline.startswith("\\item")
+    (linetype, firsttoken, firststring,
+     secondtoken, secondstring, tagcount, count) = oneline.split(" ")
+    assert linetype == "\\item"
+    assert tagcount == "count"
+    firsttoken = int(firsttoken)
+    secondtoken = int(secondtoken)
+    count = int(count)
+    items.append((firsttoken, secondtoken, count))
+
+    if firsttoken in arraycount:
+        arraycount[firsttoken] += count
+    else:
+        arraycount[firsttoken] = count
+        
+    if secondtoken in unigram:
+        unigram[secondtoken] += count
+    else:
+        unigram[secondtoken] = count
+
+    # save in phrases
+    if firsttoken in phrases:
+        assert firststring == phrases[firsttoken]
+    else:
+        phrases[firsttoken] = firststring
+
+    if secondtoken in phrases:
+        assert secondstring == phrases[secondtoken]
+    else:
+        phrases[secondtoken] = secondstring
+
+
+def parseHeader(oneline):
+    assert oneline == "\\data model interpolation"
+
+def parseEnd(oneline):
+    assert oneline == "\\end"
+
+def parseBody(infile):
+    state = None
+    for oneline in infile.readlines():
+        oneline = oneline.rstrip(os.linesep)
+        if oneline.startswith("\\data"):
+            parseHeader(oneline)
+            continue
+        if oneline.startswith("\\1-gram"):
+            state = "1-gram"
+            continue
+        if oneline.startswith("\\2-gram"):
+            state = "2-gram"
+            continue
+        if oneline.startswith("\\end"):
+            state = None
+            parseEnd(oneline)
+            continue
+        if oneline.startswith("\\item"):
+            if "1-gram" == state:
+                parseUnigramItem(oneline)
+                continue
+            if "2-gram" == state:
+                parseBigramItem(oneline)
+                continue
+
+def writeUnigramItem(inline):
+    assert inline.startswith("\\item")
+    (linetype, token, string, tagcount, count) = inline.split(" ")
+    assert linetype == "\\item"
+    assert tagcount == "count"
+    token = int(token)
+    oldfreq = int(count)
+
+    count = 0
+    if token in arraycount:
+        count = arraycount[token]
+    freq = 0
+    if token in unigram:
+        freq = unigram[token]
+
+    assert oldfreq == freq
+    outline = "\\item {0} {1} count {2} freq {3}".format(
+        token, string, count, freq)
+
+    # write out already
+    seen_in_unigram.add(token)
+
+    return outline
+
+
+def writeBigramItem(inline):
+    assert inline.startswith("\\item")
+    (linetype, firsttoken, firststring, secondtoken, secondstring, tagcount, count) = inline.split(" ")
+    assert linetype == "\\item"
+    assert tagcount == "count"
+    firsttoken = int(firsttoken)
+    secondtoken = int(secondtoken)
+    count = int(count)
+    T = count
+    N_n_0 = 1
+    n_1 = 1 if count == 1 else 0
+    Mr = count
+    outline = "\\item {0} {1} {2} {3} count {4} T {5} N_n_0 {6} n_1 {7} Mr {8}".format(firsttoken, firststring, secondtoken, secondstring, count, T, N_n_0, n_1, Mr)
+    return outline
+
+def writeHeader(inline):
+    assert inline == "\\data model interpolation"
+    count = sum([x[2] for x in items])
+    N = 1
+    total_freq = count
+    outline = "\data model \"k mixture model\" count {0} N {1} total_freq {2}".format(count, N, total_freq)
+    return outline
+
+def writeEnd(inline):
+    assert inline == "\\end"
+    outline = "\\end"
+    return outline
+
+def writeBody(infile, outfile):
+    state = None
+    for inline in infile.readlines():
+        inline = inline.rstrip(os.linesep)
+        if inline.startswith("\\data"):
+            outline = writeHeader(inline)
+            outfile.writelines([outline, os.linesep])
+            continue
+        if inline.startswith("\\1-gram"):
+            state = "1-gram"
+            outline = "\\1-gram"
+            outfile.writelines([outline, os.linesep])
+            # write the <start> tag value
+            outline = writeUnigramItem("\item 1 <start> count 0")
+            outfile.writelines([outline, os.linesep])
+            continue
+        if inline.startswith("\\2-gram"):
+            # assume \2-gram is after \1-gram
+            writeRestUnigram(outfile)
+            state = "2-gram"
+            outline = "\\2-gram"
+            outfile.writelines([outline, os.linesep])
+            continue
+        if inline.startswith("\\end"):
+            state = None
+            outline = writeEnd(inline)
+            outfile.writelines([outline, os.linesep])
+            continue
+        if inline.startswith("\\item"):
+            if "1-gram" == state:
+                outline = writeUnigramItem(inline)
+                outfile.writelines([outline, os.linesep])
+                continue
+            if "2-gram" == state:
+                outline = writeBigramItem(inline)
+                outfile.writelines([outline, os.linesep])
+                continue
+
+
+if __name__ == '__main__':
+    parser = ArgumentParser(description='Simple converter.')
+    parser.add_argument('infile', help='input file')
+    parser.add_argument('outfile', help='output file')
+
+    args = parser.parse_args()
+    print(args)
+
+    infile = open(args.infile, "r")
+    parseBody(infile)
+    infile.close()
+
+    infile = open(args.infile, "r")
+    outfile = open(args.outfile, "w")
+    writeBody(infile, outfile)
+    outfile.close()
+    infile.close()