diff --git a/galoisenne b/galoisenne
index 138b8751..7f1bec28 160000
--- a/galoisenne
+++ b/galoisenne
@@ -1 +1 @@
-Subproject commit 138b87519f05203e80e5b90da523864543f1a1e9
+Subproject commit 7f1bec28d0351ed260a3efd251a9f6119693c3b3
diff --git a/src/main/kotlin/edu/mcgill/cstk/experiments/repair/KotlinStatementRepair.kt b/src/main/kotlin/edu/mcgill/cstk/experiments/repair/KotlinStatementRepair.kt
index ed0b511e..67081582 100644
--- a/src/main/kotlin/edu/mcgill/cstk/experiments/repair/KotlinStatementRepair.kt
+++ b/src/main/kotlin/edu/mcgill/cstk/experiments/repair/KotlinStatementRepair.kt
@@ -73,7 +73,13 @@ fun evaluateSyntheticRepairBenchmarkOn(dataset: String, postprocess: List<Repair
    .forEach { (groundTruth, prompt) ->
      println("Original:  $groundTruth\nCorrupted: ${prettyDiffNoFrills(groundTruth, prompt)}")
      val startTime = System.currentTimeMillis()
-     parallelRepair(prompt, deck, edits + 1, { joinToString("").isSyntacticallyValidKotlin() }, scoreEdit).postprocess()
+     parallelRepair(
+       prompt = prompt,
+       fillers = deck,
+       maxEdits = edits + 1,
+       admissibilityFilter = { joinToString("").isSyntacticallyValidKotlin() },
+       scoreEdit = scoreEdit
+     ).postprocess()
        .also {
          //    repairKotlinStatement(prompt).also {
          val gtSeq = groundTruth.tokenizeByWhitespace().joinToString(" ")
diff --git a/src/main/kotlin/edu/mcgill/cstk/experiments/repair/PythonSnippetRepair.kt b/src/main/kotlin/edu/mcgill/cstk/experiments/repair/PythonSnippetRepair.kt
index 1cf882ff..4d192eb1 100644
--- a/src/main/kotlin/edu/mcgill/cstk/experiments/repair/PythonSnippetRepair.kt
+++ b/src/main/kotlin/edu/mcgill/cstk/experiments/repair/PythonSnippetRepair.kt
@@ -160,7 +160,7 @@ fun evaluateSeq2ParseOnStackOverflowDataset() {
 
 class RankStats(val name: String = "Total") {
   val upperBound = TIMEOUT_MS / 1000
-  val time = (300..3000 step 300) //(1..10).toSet()//setOf(2, 5, 10) + (20..upperBound step 20).toSet()
+  val time = (100..2500 step 200).toSet() //(1..10).toSet()//setOf(2, 5, 10) + (20..upperBound step 20).toSet()
   // Mean Reciprocal Rank
   val timedMRR = time.associateWith { 0.0 }.toMutableMap()
   // Precision at K, first int is K, second is the time cutoff
@@ -189,10 +189,12 @@ class RankStats(val name: String = "Total") {
         }
     }
 
+    fun Int.roundToTenths() = (toDouble() / 1000).round(1)
+
     var summary = "$name ranking statistics across $samplesEvaluated samples...\n"
     val latestMRRs = timedMRR.entries.sortedByDescending { it.key }
       .joinToString(", ") { (k, v) ->
-        "${k}s: ${(v / samplesEvaluated).round(3)}"
+        "${(k.roundToTenths()).round(1)}s: ${(v / samplesEvaluated).round(3)}"
       }
     summary += "\nMRR=  $latestMRRs"
 
@@ -200,7 +202,7 @@ class RankStats(val name: String = "Total") {
       .mapValues { (_, v) ->
         v.sortedByDescending { it.key.second }
           .joinToString(", ") { (p, v) ->
-            "${p.second}ms: ${(v / samplesEvaluated).round(3)}"
+            "${p.second.roundToTenths()}s: ${(v / samplesEvaluated).round(3)}"
           }
       }.entries.joinToString("\n") { (k, v) ->
         "P@${if (k == Int.MAX_VALUE) "All" else k}=".padEnd(6) + v
@@ -253,6 +255,7 @@ fun evaluateTidyparseOnStackoverflow() {
       parallelRepair(
         prompt = coarseBrokeStr,
         fillers = topTokens,
+//        hints = pythonErrorLocations(humanError.lexToIntTypesAsPython()),
         maxEdits = 4,
         admissibilityFilter = { map { pythonVocabBindex.getUnsafe(it) ?: it.toInt() }.isValidPython() },
         // TODO: incorporate parseable segmentations into scoring mechanism to prioritize chokepoint repairs
@@ -280,6 +283,10 @@ fun evaluateTidyparseOnStackoverflow() {
     }
   }
 
+fun pythonErrorLocations(coarseBrokeTks: List<Int>): List<Int> =
+  listOf(coarseBrokeTks.getIndexOfFirstPythonError())
+
+// Returns a triple of: (1) the broken source, (2) the human fix, and (3) the minimized fix
 private fun preprocessStackOverflow(
   brokeSnippets: Sequence<String> = readContents("parse_errors.json"),
   fixedSnippets: Sequence<String> = readContents("parse_fixes.json")
diff --git a/src/main/kotlin/edu/mcgill/cstk/utils/ParseUtils.kt b/src/main/kotlin/edu/mcgill/cstk/utils/ParseUtils.kt
index 0c0d0071..354e7026 100644
--- a/src/main/kotlin/edu/mcgill/cstk/utils/ParseUtils.kt
+++ b/src/main/kotlin/edu/mcgill/cstk/utils/ParseUtils.kt
@@ -19,6 +19,23 @@ val errorListener =
     ) { throw Exception("$msg") }
   }
 
+// Reports the index of the token that caused the error
+val indexReportingListener =
+  object: BaseErrorListener() {
+    override fun syntaxError(
+      recognizer: Recognizer<*, *>?,
+      offendingSymbol: Any?,
+      line: Int,
+      charPositionInLine: Int,
+      msg: Σᐩ?,
+      e: RecognitionException?
+    ) {
+      (offendingSymbol as? Token)
+        ?.let { throw Exception(it.tokenIndex.toString()) }
+        ?: throw Exception("")
+    }
+  }
+
 // Exhaustive tokenization includes whitespaces
 fun Σᐩ.tokenizeAsPython(exhaustive: Boolean = false): List<Σᐩ> =
   if (!exhaustive) lexAsPython().allTokens.map { it.text }
@@ -32,6 +49,17 @@ fun Σᐩ.tokenizeAsPython(exhaustive: Boolean = false): List<Σᐩ> =
       else throw Exception("Could not find token $t in ${toSplit.map { it.code }}").also { println("\n\n$this\n\n") }
   }
 
+fun List<Int>.getIndexOfFirstPythonError(): Int {
+  val tokenSource = ListTokenSource(map { CommonToken(it) })
+  val tokens = CommonTokenStream(tokenSource)
+  return try {
+    Python3Parser(tokens)
+      .apply { removeErrorListeners(); addErrorListener(indexReportingListener) }
+      .file_input()
+    -1
+  } catch (e: Exception) { e.message?.toIntOrNull() ?: -1 }
+}
+
 fun List<Int>.isValidPython(): Boolean {
   val tokenSource = ListTokenSource(map { CommonToken(it) })
   val tokens = CommonTokenStream(tokenSource)
@@ -104,6 +132,7 @@ fun Σᐩ.javac(): Σᐩ =
 
 fun Σᐩ.isValidJava() = javac().isEmpty()
 
+
 fun Σᐩ.isValidPython(onErrorAction: (Σᐩ?) -> Unit = {}): Boolean =
   try {
     Python3Parser(
diff --git a/src/main/kotlin/edu/mcgill/cstk/utils/RepairUtils.kt b/src/main/kotlin/edu/mcgill/cstk/utils/RepairUtils.kt
index 856c80b2..3dfb412e 100644
--- a/src/main/kotlin/edu/mcgill/cstk/utils/RepairUtils.kt
+++ b/src/main/kotlin/edu/mcgill/cstk/utils/RepairUtils.kt
@@ -114,6 +114,7 @@ fun <T> deltaDebug(elements: List<T>, n: Int = 2, checkValid: (List<T>) -> Boole
 fun parallelRepair(
   prompt: Σᐩ,
   fillers: Collection<Σᐩ>,
+  hints: List<Int> = emptyList(),
   maxEdits: Int = 2,
   admissibilityFilter: List<Σᐩ>.() -> Boolean,
   scoreEdit: ((List<Σᐩ>) -> Double)? = null,
@@ -123,7 +124,10 @@ fun parallelRepair(
   println("$delim\nBest repairs so far:\n$delim")
   // We intersperse the prompt with empty strings to enable the repair of the first and last token
   // as well as insertion of tokens by the repair algorithm, which only considers substitutions
-  val promptTokens = prompt.tokenizeByWhitespace().intersperse(maxEdits.coerceAtMost(2))
+  val spacingLength = maxEdits.coerceAtMost(2)
+  val promptTokens = prompt.tokenizeByWhitespace().intersperse(spacingLength)
+  // Remap the hints to the new indices in the interspersed prompt tokens
+  val remappedHints = hints.map { (spacingLength + 1) * it + 2 }
 
   val deck = fillers + promptTokens.toSet() - "\""
 
@@ -131,6 +135,7 @@ fun parallelRepair(
   return bijectiveRepair(
     promptTokens = promptTokens,
     deck = deck,
+    hints = remappedHints,
     maxEdits = maxEdits,
     takeMoreWhile = { clock.elapsedNow().inWholeMilliseconds < TIMEOUT_MS },
     admissibilityFilter = admissibilityFilter,