Still experimenting

reasoning_gym/data/gsm_data/direct_template_symbolic/0002.json

@@ -0,0 +1,28 @@
+{
+    "question": "Jerry is rolling a six-sided die. How much more likely is it (expressed as a percentage) that he rolls a number greater than 3 than that he rolls two even numbers in a row?",
+    "answer": "There are 3 numbers greater than 3 on the dice, so the chances of rolling one of them are 3 / 6 = 50%.
+The chance of rolling one even number is 50%, so the chance of rolling two in a row is 50% * 50% = 25%.
+The difference between these two probabilities is 50% - 25% = 25%.
+#### 25",
+    "id_orig": 184,
+    "id_shuffled": 2,
+    "question_annotated": "{name} is rolling a {sides} die. How much more likely is it (expressed as a percentage) that he rolls a number greater than {target} than that he rolls two {z} numbers in a row?
+
+#init:
+- name=sample(names_male)
+- $sides = sample([(\"four-sided\", 4), (\"six-sided\", 6), (\"ten-sided\", 10), (\"twelve-sided\", 12), (\"twenty-sided\", 20), (\"d4\", 4),  (\"d6\", 6), (\"d8\", 8), (\"d10\", 10), (\"d12\", 12), (\"d20\", 20), (\"4-sided\", 4),  (\"6-sided\", 6), (\"8-sided\", 8), (\"10-sided\", 10), (\"12-sided\", 12), (\"20-sided\", 20) ] * 3)
+- $target = range(1, 20)
+- z = sample([\"even\", \"odd\"])
+
+#conditions: 
+- target < sides
+- divides(sides-target, target)
+- is_int((sides-target)/sides * 100)
+-  int((sides-target)/sides * 100) > 25
+
+#answer: int((sides-target)/sides * 100) - 25",
+    "answer_annotated": "There are {sides-target} numbers greater than {target} on the dice, so the chances of rolling one of them are {sides-target} / {sides} = {int((sides-target)/sides * 100)}%.
+The chance of rolling one {z} number is 50%, so the chance of rolling two in a row is 50% * 50% = 25%.
+The difference between these two probabilities is {int((sides-target)/sides * 100)}% - 25% = { int((sides-target)/sides * 100) - 25}%.
+#### { int((sides-target)/sides * 100) - 25}"
+}