Back out "Better NaN handling"

faiss/impl/HNSW.cpp

@@ -9,7 +9,6 @@
 
 #include <faiss/impl/HNSW.h>
 
-#include <cmath>
 #include <string>
 
 #include <faiss/impl/AuxIndexStructures.h>
@@ -543,11 +542,12 @@ int search_from_candidates(
     for (int i = 0; i < candidates.size(); i++) {
         idx_t v1 = candidates.ids[i];
         float d = candidates.dis[i];
-        assert(v1 >= 0);
+        FAISS_ASSERT(v1 >= 0);
         if (!sel || sel->is_member(v1)) {
-            if (d < D[0]) {
-                faiss::maxheap_replace_top(k, D, I, d, v1);
-                nres++;
+            if (nres < k) {
+                faiss::maxheap_push(++nres, D, I, d, v1);
+            } else if (d < D[0]) {
+                faiss::maxheap_replace_top(nres, D, I, d, v1);
             }
         }
         vt.set(v1);
@@ -612,9 +612,10 @@ int search_from_candidates(
 
         auto add_to_heap = [&](const size_t idx, const float dis) {
             if (!sel || sel->is_member(idx)) {
-                if (dis < D[0]) {
-                    faiss::maxheap_replace_top(k, D, I, dis, idx);
-                    nres++;
+                if (nres < k) {
+                    faiss::maxheap_push(++nres, D, I, dis, idx);
+                } else if (dis < D[0]) {
+                    faiss::maxheap_replace_top(nres, D, I, dis, idx);
                 }
             }
             candidates.push(idx, dis);
@@ -667,7 +668,7 @@ int search_from_candidates(
         stats.n3 += ndis;
     }
 
-    return std::min(nres, k);
+    return nres;
 }
 
 std::priority_queue<HNSW::Node> search_from_candidate_unbounded(
@@ -815,11 +816,6 @@ HNSWStats HNSW::search(
         //  greedy search on upper levels
         storage_idx_t nearest = entry_point;
         float d_nearest = qdis(nearest);
-        if (!std::isfinite(d_nearest)) {
-            // means either the query or the entry point are NaN: in
-            // both cases we can only return -1 as a result
-            return stats;
-        }
 
         for (int level = max_level; level >= 1; level--) {
             greedy_update_nearest(*this, qdis, level, nearest, d_nearest);
@@ -830,6 +826,7 @@ HNSWStats HNSW::search(
             MinimaxHeap candidates(ef);
 
             candidates.push(nearest, d_nearest);
+
             search_from_candidates(
                     *this, qdis, k, I, D, candidates, vt, stats, 0, 0, params);
         } else {

faiss/impl/ResultHandler.h

@@ -445,8 +445,8 @@ struct SingleBestResultHandler {
         /// begin results for query # i
         void begin(const size_t current_idx) {
             this->current_idx = current_idx;
-            min_dis = C::neutral();
-            min_idx = -1;
+            min_dis = HUGE_VALF;
+            min_idx = 0;
         }
 
         /// add one result for query i
@@ -472,8 +472,7 @@ struct SingleBestResultHandler {
         this->i1 = i1;
 
         for (size_t i = i0; i < i1; i++) {
-            this->dis_tab[i] = C::neutral();
-            this->ids_tab[i] = -1;
+            this->dis_tab[i] = HUGE_VALF;
         }
     }
 

faiss/impl/ScalarQuantizer.cpp

@@ -1075,11 +1075,6 @@ void ScalarQuantizer::set_derived_sizes() {
 }
 
 void ScalarQuantizer::train(size_t n, const float* x) {
-    for (size_t i = 0; i < n * d; i++) {
-        FAISS_THROW_IF_NOT_MSG(
-                std::isfinite(x[i]), "training data contains NaN or Inf");
-    }
-
     int bit_per_dim = qtype == QT_4bit_uniform ? 4
             : qtype == QT_4bit                 ? 4
             : qtype == QT_6bit                 ? 6