Revert "grammars: 1.5x faster inference w/ complex grammars (vector r…

…eserves / reuses) (ggerganov#6609)"

This reverts commit cbaadc9.

examples/gbnf-validator/gbnf-validator.cpp

@@ -17,7 +17,7 @@ static bool llama_sample_grammar_string(struct llama_grammar * grammar, const st
     size_t pos = 0;
     for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
         auto prev_stacks = grammar->stacks;
-        llama_grammar_accept(grammar->rules, prev_stacks, *it, grammar->stacks);
+        grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
         if (grammar->stacks.empty()) {
             error_pos = pos;
             error_msg = "Unexpected character '" + unicode_cpt_to_utf8(*it) + "'";

llama.cpp

@@ -12202,13 +12202,12 @@ static void llama_grammar_advance_stack(
 // be positioned at a character range (see `llama_grammar_advance_stack`), and
 // produces the N possible stacks if the given char is accepted at those
 // positions
-void llama_grammar_accept(
+std::vector<std::vector<const llama_grammar_element *>> llama_grammar_accept(
         const std::vector<std::vector<llama_grammar_element>>         & rules,
         const std::vector<std::vector<const llama_grammar_element *>> & stacks,
-        const uint32_t                                                  chr,
-        std::vector<std::vector<const llama_grammar_element *>>       & new_stacks) {
+        const uint32_t                                                  chr) {
 
-    new_stacks.clear();
+    std::vector<std::vector<const llama_grammar_element *>> new_stacks;
 
     for (const auto & stack : stacks) {
         if (stack.empty()) {
@@ -12227,6 +12226,8 @@ void llama_grammar_accept(
             llama_grammar_advance_stack(rules, new_stack, new_stacks);
         }
     }
+
+    return new_stacks;
 }
 
 static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
@@ -12240,7 +12241,6 @@ static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates_for_
         const std::vector<llama_grammar_candidate>            & candidates) {
 
     std::vector<llama_grammar_candidate> rejects;
-    rejects.reserve(candidates.size());
 
     if (stack.empty()) {
         for (const auto & tok : candidates) {
@@ -12254,8 +12254,6 @@ static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates_for_
     const llama_grammar_element * stack_pos = stack.back();
 
     std::vector<llama_grammar_candidate> next_candidates;
-    next_candidates.reserve(candidates.size());
-
     for (const auto & tok : candidates) {
         if (*tok.code_points == 0) {
             // reached end of full codepoints in token, reject iff it ended in a partial sequence
@@ -13077,10 +13075,8 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
     // Note terminating 0 in decoded string
     const auto   decoded     = decode_utf8(piece, grammar->partial_utf8);
     const auto & code_points = decoded.first;
-    std::vector<std::vector<const llama_grammar_element *>> tmp_new_stacks;
     for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
-        llama_grammar_accept(grammar->rules, grammar->stacks, *it, tmp_new_stacks);
-        grammar->stacks = tmp_new_stacks;
+        grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
     }
     grammar->partial_utf8 = decoded.second;
     GGML_ASSERT(!grammar->stacks.empty());

llama.h

@@ -1101,11 +1101,10 @@ const std::vector<std::pair<std::string, struct ggml_tensor *>> & llama_internal
     struct llama_context * ctx
 );
 
-void llama_grammar_accept(
+std::vector<std::vector<const llama_grammar_element *>> llama_grammar_accept(
         const std::vector<std::vector<llama_grammar_element>>         & rules,
         const std::vector<std::vector<const llama_grammar_element *>> & stacks,
-        const uint32_t                                                  chr,
-        std::vector<std::vector<const llama_grammar_element *>>       & new_stacks);
+        const uint32_t                                                  chr);
 
 std::pair<std::vector<uint32_t>, llama_partial_utf8> decode_utf8(
         const std::string & src,

tests/test-grammar-integration.cpp

@@ -38,7 +38,7 @@ number ::= [0-9]+)""";
 
     for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
         auto prev_stacks = grammar->stacks;
-        llama_grammar_accept(grammar->rules, prev_stacks, *it, grammar->stacks);
+        grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
         assert(!grammar->stacks.empty());
     }
 
@@ -138,7 +138,7 @@ ws ::= [ \t\n\r]?)""";
         for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
             ++pos;
             auto prev_stacks = grammar->stacks;
-            llama_grammar_accept(grammar->rules, prev_stacks, *it, grammar->stacks);
+            grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
 
             // Expect that each code point will not cause the grammar to fail
             if (grammar->stacks.empty()) {
@@ -173,7 +173,7 @@ ws ::= [ \t\n\r]?)""";
 
         for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
             auto prev_stacks = grammar->stacks;
-            llama_grammar_accept(grammar->rules, prev_stacks, *it, grammar->stacks);
+            grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
             if (grammar->stacks.empty()) {
                 parse_failed = true;
                 break;