server: refactor batch construction (#24843)

* server: refactor batch construction

* wip

* wip 2

* wip 3

* wip 4

* add abort_all_slots

* handle batch full more carefully

* fix assert

* rm debug log

* small nits

* (debug) add timings

* debug: force llama_synchronize for accurate timings

* address comments

* disable DEBUG_TIMINGS

common/jinja/runtime.cpp

@@ -686,59 +686,62 @@ value set_statement::execute_impl(context & ctx) {
     return mk_val<value_undefined>();
 }
 
+static inline void bind_parameters(const std::string & name, const statements & this_args, const func_args & args, context & ctx) {
+    const size_t expected_count = this_args.size();
+    const size_t input_count = args.count();
+
+    JJ_DEBUG("Invoking '%s' with %zu input arguments (expected %zu)", name.c_str(), input_count, expected_count);
+    for (size_t i = 0; i < expected_count; ++i) {
+        if (i < input_count) {
+            if (is_stmt<identifier>(this_args[i])) {
+                // normal parameter
+                std::string param_name = cast_stmt<identifier>(this_args[i])->val;
+                value param_value = args.get_kwarg_or_pos(param_name, i);
+                JJ_DEBUG("  Binding parameter '%s' to argument of type %s", param_name.c_str(), param_value->type().c_str());
+                ctx.set_val(param_name, param_value);
+            } else if (is_stmt<keyword_argument_expression>(this_args[i])) {
+                // default argument used as normal parameter
+                auto kwarg = cast_stmt<keyword_argument_expression>(this_args[i]);
+                if (!is_stmt<identifier>(kwarg->key)) {
+                    throw std::runtime_error("Keyword argument key must be an identifier in '" + name + "'");
+                }
+                std::string param_name = cast_stmt<identifier>(kwarg->key)->val;
+                value param_value = args.get_kwarg_or_pos(param_name, i);
+                JJ_DEBUG("  Binding parameter '%s' to argument of type %s", param_name.c_str(), param_value->type().c_str());
+                ctx.set_val(param_name, param_value);
+            } else {
+                throw std::runtime_error("Invalid parameter type in '" + name + "'");
+            }
+        } else {
+            auto & default_arg = this_args[i];
+            if (is_stmt<keyword_argument_expression>(default_arg)) {
+                auto kwarg = cast_stmt<keyword_argument_expression>(default_arg);
+                if (!is_stmt<identifier>(kwarg->key)) {
+                    throw std::runtime_error("Keyword argument key must be an identifier in '" + name + "'");
+                }
+                std::string param_name = cast_stmt<identifier>(kwarg->key)->val;
+                JJ_DEBUG("  Binding parameter '%s' to default argument of type %s", param_name.c_str(), kwarg->val->type().c_str());
+                ctx.set_val(param_name, kwarg->val->execute(args.ctx));
+            } else {
+                throw std::runtime_error("Not enough arguments provided to '" + name + "'");
+            }
+            //std::string param_name = cast_stmt<identifier>(default_args[i])->val;
+            //JJ_DEBUG("  Binding parameter '%s' to default", param_name.c_str());
+            //ctx.var[param_name] = default_args[i]->execute(ctx);
+        }
+    }
+}
+
 value macro_statement::execute_impl(context & ctx) {
     if (!is_stmt<identifier>(this->name)) {
         throw std::runtime_error("Macro name must be an identifier");
     }
     std::string name = cast_stmt<identifier>(this->name)->val;
 
-    const func_handler func = [this, name, &ctx](const func_args & args) -> value {
-        size_t expected_count = this->args.size();
-        size_t input_count = args.count();
+    const func_handler func = [this, name](const func_args & args) -> value {
+        context macro_ctx(args.ctx); // new scope for macro execution
 
-        JJ_DEBUG("Invoking macro '%s' with %zu input arguments (expected %zu)", name.c_str(), input_count, expected_count);
-        context macro_ctx(ctx); // new scope for macro execution
-
-        // bind parameters
-        for (size_t i = 0; i < expected_count; ++i) {
-            if (i < input_count) {
-                if (is_stmt<identifier>(this->args[i])) {
-                    // normal parameter
-                    std::string param_name = cast_stmt<identifier>(this->args[i])->val;
-                    value param_value = args.get_kwarg_or_pos(param_name, i);
-                    JJ_DEBUG("  Binding parameter '%s' to argument of type %s", param_name.c_str(), param_value->type().c_str());
-                    macro_ctx.set_val(param_name, param_value);
-                } else if (is_stmt<keyword_argument_expression>(this->args[i])) {
-                    // default argument used as normal parameter
-                    auto kwarg = cast_stmt<keyword_argument_expression>(this->args[i]);
-                    if (!is_stmt<identifier>(kwarg->key)) {
-                        throw std::runtime_error("Keyword argument key must be an identifier in macro '" + name + "'");
-                    }
-                    std::string param_name = cast_stmt<identifier>(kwarg->key)->val;
-                    value param_value = args.get_kwarg_or_pos(param_name, i);
-                    JJ_DEBUG("  Binding parameter '%s' to argument of type %s", param_name.c_str(), param_value->type().c_str());
-                    macro_ctx.set_val(param_name, param_value);
-                } else {
-                    throw std::runtime_error("Invalid parameter type in macro '" + name + "'");
-                }
-            } else {
-                auto & default_arg = this->args[i];
-                if (is_stmt<keyword_argument_expression>(default_arg)) {
-                    auto kwarg = cast_stmt<keyword_argument_expression>(default_arg);
-                    if (!is_stmt<identifier>(kwarg->key)) {
-                        throw std::runtime_error("Keyword argument key must be an identifier in macro '" + name + "'");
-                    }
-                    std::string param_name = cast_stmt<identifier>(kwarg->key)->val;
-                    JJ_DEBUG("  Binding parameter '%s' to default argument of type %s", param_name.c_str(), kwarg->val->type().c_str());
-                    macro_ctx.set_val(param_name, kwarg->val->execute(ctx));
-                } else {
-                    throw std::runtime_error("Not enough arguments provided to macro '" + name + "'");
-                }
-                //std::string param_name = cast_stmt<identifier>(default_args[i])->val;
-                //JJ_DEBUG("  Binding parameter '%s' to default", param_name.c_str());
-                //macro_ctx.var[param_name] = default_args[i]->execute(ctx);
-            }
-        }
+        bind_parameters(name, this->args, args, macro_ctx);
 
         // execute macro body
         JJ_DEBUG("Executing macro '%s' body with %zu statements", name.c_str(), this->body.size());
@@ -752,6 +755,46 @@ value macro_statement::execute_impl(context & ctx) {
     return mk_val<value_undefined>();
 }
 
+value call_statement::execute_impl(context & ctx) {
+    auto call_expr = cast_stmt<call_expression>(this->call);
+    if (!call_expr) {
+        throw std::runtime_error("Call statement requires a valid call expression");
+    }
+
+    value callee_val = call_expr->callee->execute(ctx);
+    if (!is_val<value_func>(callee_val)) {
+        throw std::runtime_error("Callee is not a function: got " + callee_val->type());
+    }
+    auto * callee_func = cast_val<value_func>(callee_val);
+
+    context caller_ctx(ctx); // new scope for caller execution
+
+    const func_handler func = [this, caller_ctx = std::move(caller_ctx)](const func_args & args) -> value {
+        context block_ctx(caller_ctx); // new scope for block execution
+
+        bind_parameters("caller", this->caller_args, args, block_ctx);
+
+        JJ_DEBUG("Executing call body with %zu statements", this->body.size());
+        auto res = exec_statements(this->body, block_ctx);
+        JJ_DEBUG("Call body execution complete, result: %s", res->val_str.str().c_str());
+        return res;
+    };
+
+    context call_ctx(ctx);
+    call_ctx.set_val("caller", mk_val<value_func>("caller", func));
+
+    func_args args(call_ctx);
+
+    for (const auto & arg_expr : call_expr->args) {
+        auto arg_val = arg_expr->execute(ctx);
+        JJ_DEBUG("  Argument type: %s", arg_val->type().c_str());
+        args.push_back(arg_val);
+    }
+
+    JJ_DEBUG("Calling macro '%s' with %zu arguments", callee_func->name.c_str(), args.count());
+    return callee_func->invoke(args);
+}
+
 value member_expression::execute_impl(context & ctx) {
     value object = this->object->execute(ctx);
 

common/jinja/runtime.h

@@ -552,6 +552,7 @@ struct call_statement : public statement {
         for (const auto & arg : this->caller_args) chk_type<expression>(arg);
     }
     std::string type() const override { return "CallStatement"; }
+    value execute_impl(context & ctx) override;
 };
 
 struct ternary_expression : public expression {

common/peg-parser.cpp

@@ -6,13 +6,14 @@
 #include "unicode.h"
 
 #include <algorithm>
+#include <deque>
 #include <initializer_list>
 #include <map>
 #include <memory>
 #include <nlohmann/json.hpp>
 #include <regex>
+#include <set>
 #include <stdexcept>
-#include <unordered_set>
 
 // Trick to catch missing branches
 template <typename T>
@@ -88,40 +89,7 @@ struct trie {
         return match_result{match_result::NO_MATCH};
     }
 
-    struct prefix_and_next {
-        std::vector<uint32_t> prefix;
-        std::vector<uint32_t> next_chars;
-    };
-
-    std::vector<prefix_and_next> collect_prefix_and_next() {
-        std::vector<uint32_t>        prefix;
-        std::vector<prefix_and_next> result;
-        collect_prefix_and_next(0, prefix, result);
-        return result;
-    }
-
   private:
-    void collect_prefix_and_next(size_t index, std::vector<uint32_t> & prefix, std::vector<prefix_and_next> & out) {
-        if (!nodes[index].is_word) {
-            if (!nodes[index].children.empty()) {
-                std::vector<uint32_t> chars;
-                chars.reserve(nodes[index].children.size());
-                for (const auto & p : nodes[index].children) {
-                    chars.push_back(p.first);
-                }
-                out.emplace_back(prefix_and_next{prefix, chars});
-            }
-        }
-
-        for (const auto & p : nodes[index].children) {
-            uint32_t ch = p.first;
-            auto child = p.second;
-            prefix.push_back(ch);
-            collect_prefix_and_next(child, prefix, out);
-            prefix.pop_back();
-        }
-    }
-
     size_t create_node() {
         size_t index = nodes.size();
         nodes.emplace_back();
@@ -153,6 +121,65 @@ struct trie {
     }
 };
 
+// Aho-Corasick automaton
+struct aho_corasick {
+    trie                t;
+    std::vector<size_t> fail;      // failure links
+    std::vector<size_t> order;     // states in BFS order
+    std::vector<bool>   terminal;  // match states (directly or via a suffix link)
+    std::set<uint32_t>  alphabet;  // every character with a transition
+
+    aho_corasick(const std::vector<std::string> & strings) : t(strings) {
+        const auto & nodes = t.nodes;
+        const size_t n = nodes.size();
+
+        fail.assign(n, 0);
+        order.reserve(n);
+
+        std::deque<size_t> queue{ 0 };
+        while (!queue.empty()) {
+            size_t u = queue.front();
+            queue.pop_front();
+            order.push_back(u);
+            for (const auto & [ch, v] : nodes[u].children) {
+                if (u != 0) {
+                    size_t f = fail[u];
+                    while (f && nodes[f].children.find(ch) == nodes[f].children.end()) {
+                        f = fail[f];
+                    }
+                    auto it = nodes[f].children.find(ch);
+                    fail[v] = (it != nodes[f].children.end() && it->second != v) ? it->second : 0;
+                }
+                queue.push_back(v);
+            }
+        }
+
+        terminal.assign(n, false);
+        for (size_t u : order) {
+            terminal[u] = nodes[u].is_word || (u != 0 && terminal[fail[u]]);
+        }
+
+        for (const auto & node : nodes) {
+            for (const auto & [ch, v] : node.children) {
+                alphabet.insert(ch);
+            }
+        }
+    }
+
+    size_t num_states()          const { return t.nodes.size(); }
+    bool   is_terminal(size_t s) const { return terminal[s]; }
+
+    // follow failure links until a transition on `ch` exists.
+    size_t next(size_t state, uint32_t ch) const {
+        const auto & nodes = t.nodes;
+        while (state && nodes[state].children.find(ch) == nodes[state].children.end()) {
+            state = fail[state];
+        }
+        auto it = nodes[state].children.find(ch);
+        return it != nodes[state].children.end() ? it->second : 0;
+    }
+};
+
 static std::pair<uint32_t, size_t> parse_hex_escape(const std::string & str, size_t pos, int hex_count) {
     if (pos + hex_count > str.length()) {
         return {0, 0};
@@ -992,12 +1019,12 @@ void common_peg_arena::resolve_refs() {
 }
 
 std::string common_peg_arena::dump(common_peg_parser_id id) const {
-    std::unordered_set<common_peg_parser_id> visited;
+    std::set<common_peg_parser_id> visited;
     return dump_impl(id, visited);
 }
 
 std::string common_peg_arena::dump_impl(common_peg_parser_id                       id,
-                                        std::unordered_set<common_peg_parser_id> & visited) const {
+                                        std::set<common_peg_parser_id> & visited) const {
     // Check for cycles
     if (visited.count(id)) {
         return "[cycle]";
@@ -1502,61 +1529,74 @@ static std::string gbnf_escape_char_class(uint32_t c) {
     return std::string(buf);
 }
 
-static std::string gbnf_excluding_pattern(const std::vector<std::string> & strings) {
-    trie matcher(strings);
-    auto pieces = matcher.collect_prefix_and_next();
+// GBNF grammar matching strings that contain no string in `strings` as a
+// substring. Emits the complement of an Aho-Corasick automaton DFA and returns
+// the start state rule name.
+//
+// ref: https://github.com/ggml-org/llama.cpp/pull/24839
+static std::string gbnf_excluding_grammar(const common_grammar_builder & builder,
+                                          const std::string &            prefix,
+                                          const std::vector<std::string> & strings) {
+    aho_corasick ac(strings);
 
-    std::string pattern;
-    std::string trailing;  // optional proper-prefix of a delimiter, allowed only at the very end
-    for (size_t i = 0; i < pieces.size(); ++i) {
-        if (i > 0) {
-            pattern += " | ";
+    auto state_name = [&](size_t s) -> std::string {
+        if (s == 0) {
+            return prefix;
         }
+        std::string num = std::to_string(s);
+        num = num.size() == 1 ? ("0" + num) : num;
+        return prefix + "-" + num;
+    };
 
-        const auto & pre = pieces[i].prefix;
-        const auto & chars = pieces[i].next_chars;
-
-        std::string cls;
-        cls.reserve(chars.size());
+    auto char_class = [](const std::vector<uint32_t> & chars, bool negate) {
+        std::string s = negate ? "[^" : "[";
         for (uint32_t ch : chars) {
-            cls += gbnf_escape_char_class(ch);
+            s += gbnf_escape_char_class(ch);
+        }
+        return s + "]";
+    };
+
+    for (size_t q = 0; q < ac.num_states(); q++) {
+        if (ac.is_terminal(q)) {
+            continue; // match states are dropped
         }
 
-        if (!pre.empty()) {
-            std::string pre_literal = gbnf_format_literal(common_unicode_cpts_to_utf8(pre));
-            pattern += pre_literal + " [^" + cls + "]";
-            // Each interior alternative consumes a delimiter-prefix plus a disambiguating
-            // char, so the repetition alone cannot match a value that *ends* on a proper
-            // prefix of a delimiter (e.g. a trailing "\n" when the delimiter is
-            // "\n</parameter>\n"). The runtime until() (greedy first-match) accepts such
-            // values, so without this the grammar would reject input the parser accepts.
-            // Allow the value to terminate on any proper prefix as an optional tail.
-            // This makes the grammar a slight superset of the runtime language (a value
-            // may end on the longest prefix, which greedy first-match would not itself
-            // produce); harmless for constrained generation, which only needs to admit
-            // every runtime-valid string.
-            if (!trailing.empty()) {
-                trailing += " | ";
+        std::map<size_t, std::vector<uint32_t>> buckets;
+        std::vector<uint32_t> excluded;
+        for (uint32_t c : ac.alphabet) {
+            size_t d = ac.next(q, c);
+            if (ac.is_terminal(d)) {
+                excluded.push_back(c); // completes a forbidden string -> omit
+            } else if (d != 0) {
+                buckets[d].push_back(c); // specific non-root destination
+                excluded.push_back(c);
             }
-            trailing += pre_literal;
-        } else {
-            pattern += "[^" + cls + "]";
         }
+
+        std::string rhs = "|"; // every state is accepting
+        for (const auto & [d, chars] : buckets) {
+            rhs += " " + char_class(chars, false) + " " + state_name(d) + " |";
+        }
+        rhs += " " + char_class(excluded, true) + " " + state_name(0);
+
+        builder.add_rule(state_name(q), rhs);
     }
 
-    std::string result = "(" + pattern + ")*";
-    if (!trailing.empty()) {
-        result += " (" + trailing + ")?";
+    // An empty delimiter makes the start state terminal. Emit an entry rule
+    // that matches nothing so the returned reference stays valid.
+    if (ac.is_terminal(0)) {
+        builder.add_rule(prefix, "|");
     }
-    return result;
+
+    return state_name(0);
 }
 
-static std::unordered_set<std::string> collect_reachable_rules(
+static std::set<std::string> collect_reachable_rules(
     const common_peg_arena & arena,
     const common_peg_parser_id & rule
 ) {
-    std::unordered_set<std::string> reachable;
-    std::unordered_set<std::string> visited;
+    std::set<std::string> reachable;
+    std::set<std::string> visited;
 
     std::function<void(common_peg_parser_id)> visit = [&](common_peg_parser_id id) {
         const auto & parser = arena.get(id);
@@ -1765,7 +1805,7 @@ void common_peg_arena::build_grammar(const common_grammar_builder & builder, boo
                 if (p.delimiters.empty()) {
                     return ".*";
                 }
-                return gbnf_excluding_pattern(p.delimiters);
+                return gbnf_excluding_grammar(builder, "until-" + std::to_string(id), p.delimiters);
             } else if constexpr (std::is_same_v<T, common_peg_schema_parser>) {
                 if (schema_delegates(p)) {
                     return to_gbnf(p.child);
@@ -1789,7 +1829,7 @@ void common_peg_arena::build_grammar(const common_grammar_builder & builder, boo
     };
 
     // Collect reachable rules
-    std::unordered_set<std::string> reachable_rules;
+    std::set<std::string> reachable_rules;
 
     if (lazy) {
         // Collect rules reachable from trigger rules

common/peg-parser.h

@@ -3,8 +3,8 @@
 #include <nlohmann/json_fwd.hpp>
 
 #include <memory>
+#include <set>
 #include <unordered_map>
-#include <unordered_set>
 #include <string>
 #include <string_view>
 #include <functional>
@@ -335,7 +335,7 @@ class common_peg_arena {
     friend class common_peg_parser_builder;
 
   private:
-    std::string dump_impl(common_peg_parser_id id, std::unordered_set<common_peg_parser_id> & visited) const;
+    std::string dump_impl(common_peg_parser_id id, std::set<common_peg_parser_id> & visited) const;
 
     common_peg_parser_id add_parser(common_peg_parser_variant parser);
     void add_rule(const std::string & name, common_peg_parser_id id);

common/speculative.cpp

@@ -905,7 +905,13 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
 
     int32_t n_embd = 0;
 
-    bool is_mem_shared = false;
+    // One MTP draft driver, three modes (set once in the ctor):
+    //   is_mem_shared (gemma4): shares the target KV, runs all heads in one graph.
+    //   chain_heads (step35): n_mtp_layers trained heads, one per draft step.
+    //   neither (qwen35 / qwen35moe): a single trained MTP head.
+    int32_t n_mtp_layers  = 1;
+    bool    is_mem_shared = false;   // gemma4
+    bool    chain_heads   = false;   // derived in the ctor: n_mtp_layers > 1 && !is_mem_shared
 
     // Per-sequence cross-batch carryover: pair (h_p, x_{p+1}) at MTP pos p+1.
     // The last h-row of one process() call needs the first token of the NEXT
@@ -920,10 +926,8 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
     std::vector<std::vector<float>> verify_h;
     std::vector<int32_t> verify_h_rows;
 
-    // Per-seq draft length from the last draft() call, used in accept() to
-    // roll back ctx_dft's recurrent state past the AR draft's redundant
-    // pre-advancement before process() mirrored the verify batch.
-    std::vector<uint16_t> last_n_drafted;
+    std::vector<int>                i_last;
+    std::vector<std::vector<float>> chain_h;
 
     common_speculative_impl_draft_mtp(const common_params_speculative & params, uint32_t n_seq)
         : common_speculative_impl(COMMON_SPECULATIVE_TYPE_DRAFT_MTP, n_seq)
@@ -936,6 +940,7 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
         n_embd = llama_model_n_embd_out(llama_get_model(ctx_dft));
         GGML_ASSERT(n_embd == llama_model_n_embd(llama_get_model(ctx_tgt)) &&
                 "MTP input row width must match the target h_nextn width");
+        n_mtp_layers = std::max(1, (int) llama_model_n_layer_nextn(llama_get_model(ctx_dft)));
 
         LOG_INF("%s: adding speculative implementation 'draft-mtp'\n", __func__);
         LOG_INF("%s: - n_max=%d, n_min=%d, p_min=%.2f, n_embd=%d, backend_sampling=%d\n", __func__, this->params.n_max, this->params.n_min, this->params.p_min, n_embd, (int) this->params.backend_sampling);
@@ -982,16 +987,25 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
         llama_set_embeddings_nextn(ctx_dft, true, /*masked*/ true);
 
         is_mem_shared = llama_get_ctx_other(ctx_dft) == ctx_tgt;
+        chain_heads   = n_mtp_layers > 1 && !is_mem_shared;
+
+        if (chain_heads) {
+            this->params.n_max = std::min(this->params.n_max, n_mtp_layers);
+
+            chain_h.assign(n_seq, {});
+            for (auto & c : chain_h) {
+                c.reserve((size_t) (this->params.n_max + 1) * n_embd);
+            }
+        }
 
         pending_h.assign(n_seq, std::vector<float>(n_embd, 0.0f));
 
+        i_last.assign(n_seq, -1);
         i_batch_beg.assign(n_seq, -1);
         i_batch_end.assign(n_seq, -1);
 
         verify_h.assign(n_seq, {});
         verify_h_rows.assign(n_seq, 0);
-
-        last_n_drafted.assign(n_seq, 0);
     }
 
     ~common_speculative_impl_draft_mtp() override {
@@ -1097,9 +1111,34 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
                 set_h(i_batch_beg[seq_id], pending_h[seq_id].data());
             }
 
-            const int32_t rc = llama_decode(ctx_dft, batch);
-            if (rc != 0) {
-                LOG_ERR("%s: llama_decode(ctx_dft) failed rc=%d (pos=%d)\n", __func__, (int) rc, (int) batch_in.pos[0]);
+            auto * mem_dft = llama_get_memory(ctx_dft);
+
+            bool ok = true;
+            for (int head = 0; head < n_mtp_layers; ++head) {
+                if (chain_heads) {
+                    // ref: https://github.com/ggml-org/llama.cpp/pull/24340/changes#r3413498544
+                    for (llama_seq_id seq_id = 0; seq_id < (llama_seq_id) n_seq; ++seq_id) {
+                        if (i_batch_beg[seq_id] < 0) {
+                            continue;
+                        }
+                        llama_memory_seq_rm(mem_dft, seq_id, batch_in.pos[i_batch_beg[seq_id]], -1);
+                    }
+                    llama_set_nextn_layer_offset(ctx_dft, head);
+                }
+
+                const int32_t rc = llama_decode(ctx_dft, batch);
+                if (rc != 0) {
+                    LOG_ERR("%s: llama_decode(ctx_dft) head=%d failed rc=%d (pos=%d)\n",
+                            __func__, head, (int) rc, (int) batch_in.pos[0]);
+                    ok = false;
+                    break;
+                }
+            }
+
+            if (chain_heads) {
+                llama_set_nextn_layer_offset(ctx_dft, 0); // restore default for non-draft decodes
+            }
+            if (!ok) {
                 return false;
             }
         }
@@ -1134,7 +1173,6 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
         int n_drafting = 0;
         std::vector<bool> drafting(n_seq);
 
-        const float * h_row = nullptr;
         const size_t row_bytes = (size_t) n_embd * sizeof(float);
 
         for (llama_seq_id seq_id = 0; seq_id < (llama_seq_id) n_seq; ++seq_id) {
@@ -1149,22 +1187,43 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
             common_sampler_reset(smpls[seq_id].get());
 
             common_batch_add(batch, dp.id_last, dp.n_past, { seq_id }, true);
+            std::memcpy(batch.embd + (size_t) (batch.n_tokens - 1) * n_embd, pending_h[seq_id].data(), row_bytes);
 
-            h_row = pending_h[seq_id].data();
-            std::memcpy(batch.embd + n_embd*(batch.n_tokens - 1), h_row, row_bytes);
-        }
+            i_last[seq_id] = batch.n_tokens - 1;
 
-        int ret = llama_decode(ctx_dft, batch);
-        if (ret != 0) {
-            LOG_WRN("%s: llama_decode returned %d\n", __func__, ret);
-            return;
+            if (chain_heads) {
+                chain_h[seq_id].assign(pending_h[seq_id].begin(), pending_h[seq_id].end());
+            }
         }
 
         int i = 0;
 
         while (n_drafting > 0) {
-            int i_batch = 0;
+            // each step decodes under a different head, i.e. a different decoder layer, and
+            // KV is per layer. process() filled this layer's KV only for positions < n_past
+            // (prompt + accepted prefix) — nothing in the draft region yet. so reset the
+            // draft region (the seq_rm lower bound is n_past, leaving the prompt KV intact)
+            // and select head i so it rebuilds its own layer's KV there; decoding just the
+            // latest token would leave its attention reading cells only another head wrote.
+            if (chain_heads) {
+                auto * mem_dft = llama_get_memory(ctx_dft);
+                for (llama_seq_id seq_id = 0; seq_id < (llama_seq_id) n_seq; ++seq_id) {
+                    if (drafting[seq_id]) {
+                        llama_memory_seq_rm(mem_dft, seq_id, dparams[seq_id].n_past, -1);
+                    }
+                }
+                llama_set_nextn_layer_offset(ctx_dft, i);
+            }
 
+            int ret = llama_decode(ctx_dft, batch);
+            if (ret != 0) {
+                LOG_WRN("%s: llama_decode[%d] returned %d\n", __func__, i, ret);
+                break;
+            }
+
+            // rebuild the batch for the next step: the growing-KV paths re-add only the
+            // new token (the KV already holds the prefix), while chained heads re-add the
+            // whole prefix at the next head. dropped sequences are simply not re-added.
             common_batch_clear(batch);
 
             for (llama_seq_id seq_id = 0; seq_id < (llama_seq_id) n_seq; ++seq_id) {
@@ -1174,9 +1233,8 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
 
                 auto * smpl = smpls[seq_id].get();
 
-                common_sampler_sample(smpl, ctx_dft, i_batch, true);
-                h_row = llama_get_embeddings_nextn_ith(ctx_dft, i_batch);
-                ++i_batch;
+                common_sampler_sample(smpl, ctx_dft, i_last[seq_id], true);
+                const float * h_row = llama_get_embeddings_nextn_ith(ctx_dft, i_last[seq_id]);
 
                 const auto * cur_p = common_sampler_get_candidates(smpl, true);
 
@@ -1210,30 +1268,41 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
                     continue;
                 }
 
-                if (is_mem_shared) {
+                if (chain_heads) {
+                    // ref: https://github.com/ggml-org/llama.cpp/pull/24340#discussion_r3448031546
+                    chain_h[seq_id].insert(chain_h[seq_id].end(), h_row, h_row + n_embd);
+
+                    const int n_rows = (int) result.size() + 1; // id_last + tokens drafted so far
+                    for (int t = 0; t < n_rows; ++t) {
+                        const llama_token tok = (t == 0) ? dp.id_last : result[t - 1];
+                        common_batch_add(batch, tok, dp.n_past + t, { seq_id }, t == n_rows - 1);
+                        std::memcpy(batch.embd + (size_t) (batch.n_tokens - 1) * n_embd,
+                                    chain_h[seq_id].data() + (size_t) t * n_embd, row_bytes);
+                    }
+                } else if (is_mem_shared) {
                     // note: with shared memory (e.g. Gemma4 assistants) we use the same position for all draft tokens
                     // ref: https://github.com/huggingface/transformers/blob/effde20942e3f82a1b97449f60b3a48c5ff96145/docs/source/en/model_doc/gemma4_assistant.md?plain=1#L36-L37
                     common_batch_add(batch, id, dp.n_past, { seq_id }, true);
+                    std::memcpy(batch.embd + (size_t) (batch.n_tokens - 1) * n_embd, h_row, row_bytes);
                 } else {
                     common_batch_add(batch, id, dp.n_past + i + 1, { seq_id }, true);
+                    std::memcpy(batch.embd + (size_t) (batch.n_tokens - 1) * n_embd, h_row, row_bytes);
                 }
-                std::memcpy(batch.embd + n_embd*(batch.n_tokens - 1), h_row, row_bytes);
+
+                i_last[seq_id] = batch.n_tokens - 1;
             }
 
             if (batch.n_tokens == 0) {
                 break;
             }
 
-            // evaluate the drafted tokens on the draft model
-            ret = llama_decode(ctx_dft, batch);
-            if (ret != 0) {
-                LOG_WRN("%s: llama_decode[%d] returned %d\n", __func__, i, ret);
-                break;
-            }
-
             ++i;
         }
 
+        if (chain_heads) {
+            llama_set_nextn_layer_offset(ctx_dft, 0); // restore default for non-draft decodes
+        }
+
         for (llama_seq_id seq_id = 0; seq_id < (llama_seq_id) n_seq; ++seq_id) {
             auto & dp = dparams[seq_id];
             if (!dp.drafting) {
@@ -1243,8 +1312,6 @@ struct common_speculative_impl_draft_mtp : public common_speculative_impl {
             if (dp.result->size() < (size_t) params.n_min) {
                 dp.result->clear();
             }
-
-            last_n_drafted[seq_id] = (uint16_t) dp.result->size();
         }
     }
 
@@ -1857,7 +1924,7 @@ common_speculative * common_speculative_init(common_params_speculative & params,
 
         bool has_draft_simple = (enabled_configs & (1u << COMMON_SPECULATIVE_TYPE_DRAFT_SIMPLE));
         bool has_draft_eagle3 = (enabled_configs & (1u << COMMON_SPECULATIVE_TYPE_DRAFT_EAGLE3)) && params.draft.ctx_dft != nullptr;
-        bool has_mtp = (enabled_configs & (1u << COMMON_SPECULATIVE_TYPE_DRAFT_MTP)) && params.draft.ctx_dft != nullptr;
+        bool has_draft_mtp    = (enabled_configs & (1u << COMMON_SPECULATIVE_TYPE_DRAFT_MTP))    && params.draft.ctx_dft != nullptr;
 
 
 
@@ -1895,7 +1962,7 @@ common_speculative * common_speculative_init(common_params_speculative & params,
         if (has_draft_eagle3) {
             configs.push_back(common_speculative_config(COMMON_SPECULATIVE_TYPE_DRAFT_EAGLE3, params));
         }
-        if (has_mtp) {
+        if (has_draft_mtp) {
             configs.push_back(common_speculative_config(COMMON_SPECULATIVE_TYPE_DRAFT_MTP, params));
         }
     }

include/llama.h

@@ -558,14 +558,15 @@ extern "C" {
     LLAMA_API const struct llama_vocab * llama_model_get_vocab(const struct llama_model * model);
     LLAMA_API enum llama_rope_type       llama_model_rope_type(const struct llama_model * model);
 
-    LLAMA_API int32_t llama_model_n_ctx_train(const struct llama_model * model);
-    LLAMA_API int32_t llama_model_n_embd     (const struct llama_model * model);
-    LLAMA_API int32_t llama_model_n_embd_inp (const struct llama_model * model);
-    LLAMA_API int32_t llama_model_n_embd_out (const struct llama_model * model);
-    LLAMA_API int32_t llama_model_n_layer    (const struct llama_model * model);
-    LLAMA_API int32_t llama_model_n_head     (const struct llama_model * model);
-    LLAMA_API int32_t llama_model_n_head_kv  (const struct llama_model * model);
-    LLAMA_API int32_t llama_model_n_swa      (const struct llama_model * model);
+    LLAMA_API int32_t llama_model_n_ctx_train  (const struct llama_model * model);
+    LLAMA_API int32_t llama_model_n_embd       (const struct llama_model * model);
+    LLAMA_API int32_t llama_model_n_embd_inp   (const struct llama_model * model);
+    LLAMA_API int32_t llama_model_n_embd_out   (const struct llama_model * model);
+    LLAMA_API int32_t llama_model_n_layer      (const struct llama_model * model);
+    LLAMA_API int32_t llama_model_n_layer_nextn(const struct llama_model * model);
+    LLAMA_API int32_t llama_model_n_head       (const struct llama_model * model);
+    LLAMA_API int32_t llama_model_n_head_kv    (const struct llama_model * model);
+    LLAMA_API int32_t llama_model_n_swa        (const struct llama_model * model);
 
     // Get the model's RoPE frequency scaling factor
     LLAMA_API float llama_model_rope_freq_scale_train(const struct llama_model * model);

src/llama-context.cpp

@@ -1156,6 +1156,10 @@ void llama_context::set_embeddings_layer_inp(uint32_t lid, bool enable) {
     sched_need_reserve = true;
 }
 
+void llama_context::set_nextn_layer_offset(int32_t offset) {
+    cparams.nextn_layer_offset = offset;
+}
+
 void llama_context::set_causal_attn(bool value) {
     LLAMA_LOG_DEBUG("%s: value = %d\n", __func__, value);
 
@@ -3699,6 +3703,10 @@ void llama_set_embeddings_layer_inp(llama_context * ctx, uint32_t lid, bool valu
     ctx->set_embeddings_layer_inp(lid, value);
 }
 
+void llama_set_nextn_layer_offset(llama_context * ctx, int32_t offset) {
+    ctx->set_nextn_layer_offset(offset);
+}
+
 llama_memory_t llama_get_memory(const struct llama_context * ctx) {
     if (!ctx) {
         return nullptr;

src/llama-context.h

@@ -115,6 +115,7 @@ struct llama_context {
     void set_embeddings (bool value);
     void set_embeddings_nextn(bool value, bool masked);
     void set_embeddings_layer_inp(uint32_t lid, bool enable);
+    void set_nextn_layer_offset(int32_t offset);
     void set_causal_attn(bool value);
     void set_warmup(bool value);
 

src/llama-cparams.h

@@ -18,6 +18,8 @@ struct llama_cparams {
     int32_t  n_threads;       // number of threads to use for generation
     int32_t  n_threads_batch; // number of threads to use for batch processing
 
+    int32_t  nextn_layer_offset = 0;
+
     float rope_freq_base;
     float rope_freq_scale;
 

src/llama-ext.h

@@ -95,6 +95,11 @@ LLAMA_API llama_memory_breakdown llama_get_memory_breakdown(const struct llama_c
 // If masked == false, output the embeddings for all tokens in the batch regardless of batch.logits
 LLAMA_API void llama_set_embeddings_nextn(struct llama_context * ctx, bool value, bool masked);
 
+// Select which appended NextN block the DECODER_MTP graph runs (offset past
+// the trunk: il = n_layer() + offset). Used by the speculative NextN driver to
+// chain multiple trained NextN heads. Default 0 (first head).
+LLAMA_API void llama_set_nextn_layer_offset(struct llama_context * ctx, int32_t offset);
+
 // mirrors:
 // LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
 LLAMA_API float * llama_get_embeddings_nextn(struct llama_context * ctx);

src/llama-graph.h

@@ -682,9 +682,16 @@ struct llm_graph_params {
             }
         }
 
+        // TODO: https://github.com/ggml-org/llama.cpp/pull/24340#discussion_r3448035248
+        if (cparams.nextn_layer_offset != other.cparams.nextn_layer_offset) {
+            return false;
+        }
+
         return
-            cparams.embeddings  == other.cparams.embeddings  &&
-            cparams.causal_attn == other.cparams.causal_attn &&
+            cparams.embeddings              == other.cparams.embeddings              &&
+            cparams.embeddings_nextn        == other.cparams.embeddings_nextn        &&
+            cparams.embeddings_nextn_masked == other.cparams.embeddings_nextn_masked &&
+            cparams.causal_attn             == other.cparams.causal_attn             &&
             arch  == other.arch  &&
             gtype == other.gtype &&
             cvec  == other.cvec  &&

src/llama-model.cpp

@@ -2312,6 +2312,10 @@ int32_t llama_model_n_layer(const llama_model * model) {
     return model->hparams.n_layer();
 }
 
+int32_t llama_model_n_layer_nextn(const llama_model * model) {
+    return model->hparams.n_layer_nextn;
+}
+
 int32_t llama_model_n_head(const llama_model * model) {
     return model->hparams.n_head();
 }

src/models/step35.cpp

@@ -112,7 +112,7 @@ void llama_model_step35::load_arch_tensors(llama_model_loader & ml) {
         layer.ffn_down_shexp = create_tensor(tn(LLM_TENSOR_FFN_DOWN_SHEXP, "weight", i), {hparams.n_ff_shexp, n_embd}, TENSOR_NOT_REQUIRED);
     };
 
-    auto load_block_mtp = [&](int i, bool is_first_mtp) {
+    auto load_block_mtp = [&](int i) {
         auto & layer = layers[i];
 
         const uint32_t n_head_l      = hparams.n_head(i);
@@ -121,15 +121,12 @@ void llama_model_step35::load_arch_tensors(llama_model_loader & ml) {
 
         // The MTP block is a full Step3p5 decoder layer (mtp_block) plus the
         // NextN-specific wiring (enorm/hnorm/eh_proj + optional shared head).
-        // `mtp_flags` becomes NOT_REQUIRED when the GGUF is trunk-only.
-        //
-        // Only the FIRST MTP block (i == n_main) is required for the
-        // single-block MTP runtime; trailing MTP blocks are always tolerated
-        // as missing so pruned GGUFs (block 0 only) load cleanly. Override
-        // mtp_flags to NOT_REQUIRED for those.
-        const int eff_mtp_flags = is_first_mtp ? mtp_flags : (mtp_flags | TENSOR_NOT_REQUIRED);
+        // Multi-block MTP: every declared MTP block is required (the draft chain
+        // runs all n_layer_nextn heads), so each block uses the captured
+        // `mtp_flags` directly — already NOT_REQUIRED for a trunk-only GGUF,
+        // which keeps that path correct.
 
-        layer.attn_norm   = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, eff_mtp_flags);
+        layer.attn_norm   = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, mtp_flags);
         layer.attn_q_norm = create_tensor(tn(LLM_TENSOR_ATTN_Q_NORM, "weight", i), {n_embd_head_k}, TENSOR_NOT_REQUIRED);
         layer.attn_k_norm = create_tensor(tn(LLM_TENSOR_ATTN_K_NORM, "weight", i), {n_embd_head_k}, TENSOR_NOT_REQUIRED);
 
@@ -140,12 +137,12 @@ void llama_model_step35::load_arch_tensors(llama_model_loader & ml) {
             layer.rope_freqs = create_tensor(tn(LLM_TENSOR_ROPE_FREQS, "weight", i), {n_rot_max/2}, TENSOR_NOT_REQUIRED | TENSOR_DUPLICATED);
         }
 
-        create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head_l, n_embd_k_gqa, n_embd_v_gqa, eff_mtp_flags);
-        layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd_head_v * n_head_l, n_embd}, eff_mtp_flags);
+        create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head_l, n_embd_k_gqa, n_embd_v_gqa, mtp_flags);
+        layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd_head_v * n_head_l, n_embd}, mtp_flags);
 
         layer.wqkv_gate = create_tensor(tn(LLM_TENSOR_ATTN_GATE, "weight", i), {n_embd, n_head_l}, TENSOR_NOT_REQUIRED);
 
-        layer.ffn_norm = create_tensor(tn(LLM_TENSOR_FFN_NORM, "weight", i), {n_embd}, eff_mtp_flags);
+        layer.ffn_norm = create_tensor(tn(LLM_TENSOR_FFN_NORM, "weight", i), {n_embd}, mtp_flags);
 
         // dense MLP (leading dense blocks) — present if the MTP block isn't MoE
         layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd,   n_ff}, TENSOR_NOT_REQUIRED);
@@ -165,9 +162,9 @@ void llama_model_step35::load_arch_tensors(llama_model_loader & ml) {
         layer.ffn_down_shexp = create_tensor(tn(LLM_TENSOR_FFN_DOWN_SHEXP, "weight", i), {hparams.n_ff_shexp, n_embd}, TENSOR_NOT_REQUIRED);
 
         // NextN-specific tensors that define the MTP block.
-        layer.nextn.eh_proj          = create_tensor(tn(LLM_TENSOR_NEXTN_EH_PROJ,          "weight", i), { 2 * n_embd, n_embd }, eff_mtp_flags);
-        layer.nextn.enorm            = create_tensor(tn(LLM_TENSOR_NEXTN_ENORM,            "weight", i), { n_embd },              eff_mtp_flags);
-        layer.nextn.hnorm            = create_tensor(tn(LLM_TENSOR_NEXTN_HNORM,            "weight", i), { n_embd },              eff_mtp_flags);
+        layer.nextn.eh_proj          = create_tensor(tn(LLM_TENSOR_NEXTN_EH_PROJ,          "weight", i), { 2 * n_embd, n_embd }, mtp_flags);
+        layer.nextn.enorm            = create_tensor(tn(LLM_TENSOR_NEXTN_ENORM,            "weight", i), { n_embd },              mtp_flags);
+        layer.nextn.hnorm            = create_tensor(tn(LLM_TENSOR_NEXTN_HNORM,            "weight", i), { n_embd },              mtp_flags);
         layer.nextn.embed_tokens     = create_tensor(tn(LLM_TENSOR_NEXTN_EMBED_TOKENS,     "weight", i), { n_embd, n_vocab },     TENSOR_NOT_REQUIRED);
         layer.nextn.shared_head_head = create_tensor(tn(LLM_TENSOR_NEXTN_SHARED_HEAD_HEAD, "weight", i), { n_embd, n_vocab },     TENSOR_NOT_REQUIRED);
         layer.nextn.shared_head_norm = create_tensor(tn(LLM_TENSOR_NEXTN_SHARED_HEAD_NORM, "weight", i), { n_embd },              TENSOR_NOT_REQUIRED);
@@ -176,13 +173,11 @@ void llama_model_step35::load_arch_tensors(llama_model_loader & ml) {
     for (int i = 0; i < n_layer; ++i) {
         load_block_trunk(i, trunk_flags);
     }
-    // Only the first MTP block (i == n_main) is required at runtime — the
-    // single-block-MTP graph in build_arch_graph always uses that one.
-    // Trailing MTP blocks are loaded if present (so an un-pruned GGUF with
-    // all MTP layers still works) but tolerated when absent via the pruning
-    // path. See scripts/prune_step35_extra_mtp.py for the pruner.
+    // All n_layer_nextn MTP blocks are required — the multi-block draft chain
+    // runs every head (head k at offset k). The GGUF declares the count via
+    // step35.nextn_predict_layers.
     for (int i = n_layer; i < n_layer_all; ++i) {
-        load_block_mtp(i, /*is_first_mtp=*/ i == n_layer);
+        load_block_mtp(i);
     }
 }
 
@@ -372,13 +367,14 @@ llama_model_step35::graph_mtp::graph_mtp(const llama_model & model, const llm_gr
     : llm_graph_context(params) {
     GGML_ASSERT(hparams.n_layer_nextn > 0 && "STEP35 MTP requires n_layer_nextn > 0");
 
-    // Single-block MTP only: always run the first trained MTP block (Qwen
-    // MTP / vLLM single-MTP-layer style). Multi-block round-robin proved to
-    // be a much deeper refactor than this PR justifies; the trailing MTP
-    // blocks are loaded with TENSOR_NOT_REQUIRED so pruned GGUFs (with just
-    // block 0) also work — see load_arch_tensors below and
-    // scripts/prune_step35_extra_mtp.py.
-    const int il = hparams.n_layer();
+    // Multi-block MTP: the DECODER_MTP graph runs the MTP head selected by
+    // cparams.nextn_layer_offset (0 = first trained head). The speculative driver
+    // bumps the offset per draft step to chain heads 45->46->47. offset 0 keeps
+    // single-block behavior identical to before.
+    const int il = hparams.n_layer() + cparams.nextn_layer_offset;
+    GGML_ASSERT(cparams.nextn_layer_offset >= 0 &&
+                cparams.nextn_layer_offset < (int) hparams.n_layer_nextn &&
+                "nextn_layer_offset out of range [0, n_layer_nextn)");
     const auto & layer = model.layers[il];
 
     GGML_ASSERT(layer.nextn.eh_proj && "MTP block missing nextn.eh_proj");
@@ -536,6 +532,9 @@ llama_model_step35::graph_mtp::graph_mtp(const llama_model & model, const llm_gr
     cur = ggml_add(ctx0, cur, ffn_inp);
     cb(cur, "mtp_post_ffn", il);
 
+    ggml_tensor * inp_out_ids = build_inp_out_ids();
+    cur = ggml_get_rows(ctx0, cur, inp_out_ids);
+
     // Pre-norm hidden state: used by the AR draft loop to seed the next MTP step.
     cb(cur, "h_nextn", -1);
     res->t_h_nextn = cur;

tests/peg-parser/test-gbnf-generation.cpp

@@ -129,8 +129,86 @@ void test_gbnf_generation(testing &t) {
         });
 
         assert_gbnf_equal(t, R"""(
-            root ::= ([^<] | "<" [^/] | "</" [^t] | "</t" [^a] | "</ta" [^g] | "</tag" [^>])* ("<" | "</" | "</t" | "</ta" | "</tag")?
+            root ::= until-0
             space ::= | " " | "\n"{1,2} [ \t]{0,20}
+            until-0 ::= | [<] until-0-01 | [^<] until-0
+            until-0-01 ::= | [<] until-0-01 | [/] until-0-02 | [^/<] until-0
+            until-0-02 ::= | [<] until-0-01 | [t] until-0-03 | [^<t] until-0
+            until-0-03 ::= | [<] until-0-01 | [a] until-0-04 | [^<a] until-0
+            until-0-04 ::= | [<] until-0-01 | [g] until-0-05 | [^<g] until-0
+            until-0-05 ::= | [<] until-0-01 | [^<>] until-0
+        )""", gbnf);
+    });
+
+    t.test("until grammar overlapping delimiter", [](testing &t) {
+        auto parser = build_peg_parser([](common_peg_parser_builder & p)  {
+            return p.until("\n</parameter>\n");
+        });
+
+        auto gbnf = build_grammar([&](const common_grammar_builder & builder) {
+            parser.build_grammar(builder);
+        });
+
+        assert_gbnf_equal(t, R"""(
+            root ::= until-0
+            space ::= | " " | "\n"{1,2} [ \t]{0,20}
+            until-0 ::= | [\n] until-0-01 | [^\n] until-0
+            until-0-01 ::= | [\n] until-0-01 | [<] until-0-02 | [^\n<] until-0
+            until-0-02 ::= | [\n] until-0-01 | [/] until-0-03 | [^\n/] until-0
+            until-0-03 ::= | [\n] until-0-01 | [p] until-0-04 | [^\np] until-0
+            until-0-04 ::= | [\n] until-0-01 | [a] until-0-05 | [^\na] until-0
+            until-0-05 ::= | [\n] until-0-01 | [r] until-0-06 | [^\nr] until-0
+            until-0-06 ::= | [\n] until-0-01 | [a] until-0-07 | [^\na] until-0
+            until-0-07 ::= | [\n] until-0-01 | [m] until-0-08 | [^\nm] until-0
+            until-0-08 ::= | [\n] until-0-01 | [e] until-0-09 | [^\ne] until-0
+            until-0-09 ::= | [\n] until-0-01 | [t] until-0-10 | [^\nt] until-0
+            until-0-10 ::= | [\n] until-0-01 | [e] until-0-11 | [^\ne] until-0
+            until-0-11 ::= | [\n] until-0-01 | [r] until-0-12 | [^\nr] until-0
+            until-0-12 ::= | [\n] until-0-01 | [>] until-0-13 | [^\n>] until-0
+            until-0-13 ::= | [^\n] until-0
+        )""", gbnf);
+    });
+
+    // DeepSeek-V3.2 tag prefix. The DSML token (｜DSML｜) embeds U+FF5C,
+    // so the delimiter mixes ASCII and multi-byte codepoints.
+    t.test("until grammar unicode delimiter", [](testing &t) {
+        auto parser = build_peg_parser([](common_peg_parser_builder & p)  {
+            return p.until("<｜DSML｜");
+        });
+
+        auto gbnf = build_grammar([&](const common_grammar_builder & builder) {
+            parser.build_grammar(builder);
+        });
+
+        assert_gbnf_equal(t, R"""(
+            root ::= until-0
+            space ::= | " " | "\n"{1,2} [ \t]{0,20}
+            until-0 ::= | [<] until-0-01 | [^<] until-0
+            until-0-01 ::= | [<] until-0-01 | [\uFF5C] until-0-02 | [^<\uFF5C] until-0
+            until-0-02 ::= | [<] until-0-01 | [D] until-0-03 | [^<D] until-0
+            until-0-03 ::= | [<] until-0-01 | [S] until-0-04 | [^<S] until-0
+            until-0-04 ::= | [<] until-0-01 | [M] until-0-05 | [^<M] until-0
+            until-0-05 ::= | [<] until-0-01 | [L] until-0-06 | [^<L] until-0
+            until-0-06 ::= | [<] until-0-01 | [^<\uFF5C] until-0
+        )""", gbnf);
+    });
+
+    t.test("until grammar multiple delimiters", [](testing &t) {
+        auto parser = build_peg_parser([](common_peg_parser_builder & p)  {
+            return p.until_one_of({"ab", "cd", "ef"});
+        });
+
+        auto gbnf = build_grammar([&](const common_grammar_builder & builder) {
+            parser.build_grammar(builder);
+        });
+
+        assert_gbnf_equal(t, R"""(
+            root ::= until-0
+            space ::= | " " | "\n"{1,2} [ \t]{0,20}
+            until-0 ::= | [a] until-0-01 | [c] until-0-03 | [e] until-0-05 | [^ace] until-0
+            until-0-01 ::= | [a] until-0-01 | [c] until-0-03 | [e] until-0-05 | [^abce] until-0
+            until-0-03 ::= | [a] until-0-01 | [c] until-0-03 | [e] until-0-05 | [^acde] until-0
+            until-0-05 ::= | [a] until-0-01 | [c] until-0-03 | [e] until-0-05 | [^acef] until-0
         )""", gbnf);
     });
 

tests/test-jinja.cpp

@@ -995,6 +995,32 @@ static void test_macros(testing & t) {
         json::object(),
         "Hello, John Smith,Hi, Jane Doe"
     );
+
+    test_template(t, "macro with caller",
+        "\
+{%- macro nest_dict(o, i, ff='') %}\n\
+  {{- caller(ff) }}\n\
+  {%- for k, v in o|items %}\n\
+    {{- i + k + ': ' }}\n\
+    {%- if v is mapping %}\n\
+      {{- '{' }}\n\
+      {% call(f) nest_dict(v, i + '    ') %}\n\
+        {{- 'fail' if ff is undefined }}\n\
+      {%- endcall %}\n\
+      {{- i + '}' }}\n\
+    {% else %}\n\
+      {{- v|string }}\n\
+    {% endif %}\n\
+  {%- endfor %}\n\
+{%- endmacro %}\n\
+{%- call(f) nest_dict({'root1': 1, 'root2': {'nest1': 1, 'nest2': {'nest3': 2}}}, '    ', 'Dict') %}\n\
+  {{- 'fail' if ff is defined }}\n\
+  {{- f + ' {' }}\n\
+{% endcall %}\n\
+{{- '}' }}",
+        json::object(),
+        "Dict {\n    root1: 1\n    root2: {\n        nest1: 1\n        nest2: {\n            nest3: 2\n        }\n    }\n}"
+    );
 }
 
 static void test_namespace(testing & t) {

tools/mtmd/clip.cpp

@@ -1045,8 +1045,17 @@ struct clip_model_loader {
     bool has_vision = false;
     bool has_audio  = false;
 
+    mtmd_progress_callback progress_callback = nullptr;
+    void * progress_callback_user_data = nullptr;
+
     // TODO @ngxson : we should not pass clip_ctx here, it should be clip_model
-    clip_model_loader(const char * fname, bool skip_tensors = false) : fname(fname) {
+    clip_model_loader(const char * fname,
+            bool skip_tensors = false,
+            mtmd_progress_callback progress_cb = nullptr,
+            void * progress_user_data = nullptr)
+        : fname(fname),
+          progress_callback(progress_cb),
+          progress_callback_user_data(progress_user_data) {
         struct ggml_context * meta = nullptr;
 
         struct gguf_init_params params = {
@@ -2787,37 +2796,60 @@ struct clip_model_loader {
         }
 
         // load data
-        if (!ctx_clip.no_alloc) {
+        {
             std::vector<uint8_t> read_buf;
 
+            // start loading event
+            if (progress_callback){
+                progress_callback(0.0, progress_callback_user_data);
+            }
+
+            // compute total tensor data size for progress reporting
+            size_t total_data_size = 0;
+            for (auto & t : tensors_to_load) {
+                total_data_size += ggml_nbytes(t);
+            }
+
             // alloc memory and offload data
             ggml_backend_buffer_type_t buft = ggml_backend_get_default_buffer_type(ctx_clip.backend);
             ctx_clip.buf.reset(ggml_backend_alloc_ctx_tensors_from_buft(ctx_clip.ctx_data.get(), buft));
             ggml_backend_buffer_set_usage(ctx_clip.buf.get(), GGML_BACKEND_BUFFER_USAGE_WEIGHTS);
-            for (auto & t : tensors_to_load) {
-                ggml_tensor * cur = ggml_get_tensor(ctx_clip.ctx_data.get(), t->name);
-                GGML_ASSERT(cur && "tensor not found in ctx_data");
-                auto it_off = tensor_offset.find(t->name);
-                GGML_ASSERT(it_off != tensor_offset.end() && "no offset for tensor");
-                const size_t offset = it_off->second;
-                fin.seekg(offset, std::ios::beg);
-                if (!fin) {
-                    throw std::runtime_error(string_format("%s: failed to seek for tensor %s\n", __func__, t->name));
-                }
-                size_t num_bytes = ggml_nbytes(cur);
-                if (ggml_backend_buft_is_host(buft)) {
-                    // for the CPU and Metal backend, we can read directly into the tensor
-                    fin.read(reinterpret_cast<char *>(cur->data), num_bytes);
-                } else {
-                    // read into a temporary buffer first, then copy to device memory
-                    read_buf.resize(num_bytes);
-                    fin.read(reinterpret_cast<char *>(read_buf.data()), num_bytes);
-                    ggml_backend_tensor_set(cur, read_buf.data(), 0, num_bytes);
+            // read the weight from file
+            if (!ctx_clip.no_alloc) {
+                size_t data_loaded = 0;
+                for (auto & t : tensors_to_load) {
+                    ggml_tensor * cur = ggml_get_tensor(ctx_clip.ctx_data.get(), t->name);
+                    GGML_ASSERT(cur && "tensor not found in ctx_data");
+                    auto it_off = tensor_offset.find(t->name);
+                    GGML_ASSERT(it_off != tensor_offset.end() && "no offset for tensor");
+                    const size_t offset = it_off->second;
+                    fin.seekg(offset, std::ios::beg);
+                    if (!fin) {
+                        throw std::runtime_error(string_format("%s: failed to seek for tensor %s\n", __func__, t->name));
+                    }
+                    size_t num_bytes = ggml_nbytes(cur);
+                    if (ggml_backend_buft_is_host(buft)) {
+                        // for the CPU and Metal backend, we can read directly into the tensor
+                        fin.read(reinterpret_cast<char *>(cur->data), num_bytes);
+                    } else {
+                        // read into a temporary buffer first, then copy to device memory
+                        read_buf.resize(num_bytes);
+                        fin.read(reinterpret_cast<char *>(read_buf.data()), num_bytes);
+                        ggml_backend_tensor_set(cur, read_buf.data(), 0, num_bytes);
+                    }
+                    data_loaded += num_bytes;
+                    if (progress_callback && total_data_size > 0) {
+                        const float progress = (float)data_loaded / (float)total_data_size;
+                        if (!progress_callback(progress, progress_callback_user_data)) {
+                            throw std::runtime_error(string_format("%s: model loading cancelled by progress_callback\n", __func__));
+                        }
+                    }
                 }
+                LOG_DBG("%s: loaded %zu tensors from %s\n", __func__, tensors_to_load.size(), fname.c_str());
+            } else {
+                LOG_DBG("%s: no_alloc is set, skipping tensor data loading (%zu tensors)\n", __func__, tensors_to_load.size());
             }
             fin.close();
-
-            LOG_DBG("%s: loaded %zu tensors from %s\n", __func__, tensors_to_load.size(), fname.c_str());
         }
 
     }
@@ -3105,7 +3137,10 @@ struct clip_init_result clip_init(const char * fname, struct clip_context_params
     clip_ctx * ctx_audio = nullptr;
 
     try {
-        clip_model_loader loader(fname);
+        clip_model_loader loader(fname,
+            /* skip_tensors */ false,
+            ctx_params.progress_callback,
+            ctx_params.progress_callback_user_data);
         bool skip_audio = false;
 
         if (loader.has_vision) {

tools/mtmd/clip.h

@@ -54,6 +54,8 @@ struct clip_context_params {
     ggml_backend_sched_eval_callback cb_eval;
     void * cb_eval_user_data;
     bool no_alloc;
+    mtmd_progress_callback progress_callback;
+    void * progress_callback_user_data;
 };
 
 struct clip_init_result {

tools/mtmd/mtmd.cpp

@@ -251,6 +251,8 @@ mtmd_context_params mtmd_context_params_default() {
         /* cb_eval           */ nullptr,
         /* cb_eval_user_data */ nullptr,
         /* batch_max_tokens  */ 1024,
+        /* progress_callback */ nullptr,
+        /* progress_callback_user_data */ nullptr,
     };
     return params;
 }
@@ -345,6 +347,8 @@ struct mtmd_context {
             /* cb_eval           */ ctx_params.cb_eval,
             /* cb_eval_user_data */ ctx_params.cb_eval_user_data,
             /* no_alloc          */ no_alloc,
+            /* progress_callback */ ctx_params.progress_callback,
+            /* progress_callback_user_data */ ctx_params.progress_callback_user_data,
         };
 
         auto res = clip_init(mmproj_fname, ctx_clip_params);
@@ -2133,9 +2137,12 @@ std::map<ggml_backend_dev_t, size_t> mtmd_get_memory_usage(const char * mmproj_f
     mtmd::context_ptr ctx;
     auto saved_log_callback = g_logger_state.log_callback;
     auto saved_log_user_data = g_logger_state.log_callback_user_data;
+
+    ctx_params.progress_callback = nullptr;
+
     try {
         mtmd_log_set(stub_log_callback, nullptr); // suppress logging
-        ctx.reset(new mtmd_context(mmproj_fname, nullptr, ctx_params));
+        ctx.reset(new mtmd_context(mmproj_fname, nullptr, ctx_params, true));
         mtmd_log_set(saved_log_callback, saved_log_user_data); // restore log callback
         std::map<ggml_backend_dev_t, size_t> total_mem;
         auto merge = [&](const struct clip_ctx * c) {

tools/mtmd/mtmd.h

@@ -83,6 +83,8 @@ typedef struct mtmd_input_chunks mtmd_input_chunks;
 typedef struct mtmd_input_text   mtmd_input_text;
 typedef struct mtmd_batch        mtmd_batch;
 
+typedef bool (*mtmd_progress_callback)(float progress, void * user_data);
+
 struct mtmd_context_params {
     bool use_gpu;
     bool print_timings;
@@ -104,6 +106,12 @@ struct mtmd_context_params {
     int32_t batch_max_tokens; // maximum number of output tokens in a batch
                               // (note: this is not a hard-limit, the first image will always be added even if it exceeds this limit)
                               // (default: 1024)
+
+    // Called with a progress value between 0.0 and 1.0. Pass NULL to disable.
+    // If the provided progress_callback returns true, model loading continues.
+    // If it returns false, model loading is immediately aborted.
+    mtmd_progress_callback progress_callback;
+    void * progress_callback_user_data;
 };
 
 MTMD_API const char * mtmd_default_marker(void);

tools/server/README.md

@@ -1859,9 +1859,33 @@ Example events:
 
 {
   "model": "...",
-  "event": "download_finished",
+  "event": "model_status",
   "data": {
-    "status": "loading"
+    "status": "loading",
+    "progress": {
+      "stage": "fit_params",
+      "value": 0.5 // from 0.0 to 1.0 ; note: not all stages have this "value"
+    }
+  }
+}
+
+{
+  "model": "...",
+  "event": "model_status",
+  "data": {
+    "status": "loaded",
+    "info": {
+      // note: only include info on first load
+      // waking up from sleep doesn't have this
+    }
+  }
+}
+
+{
+  "model": "...",
+  "event": "model_status",
+  "data": {
+    "status": "sleeping"
   }
 }
 

tools/server/server-models.cpp

@@ -442,6 +442,7 @@ void server_models::load_models() {
                 /* last_used     */ 0,
                 /* args          */ std::vector<std::string>(),
                 /* loaded_info   */ {},
+                /* progress      */ {},
                 /* exit_code     */ 0,
                 /* stop_timeout  */ DEFAULT_STOP_TIMEOUT,
                 /* multimodal    */ mtmd_caps{false, false},
@@ -608,6 +609,7 @@ void server_models::load_models() {
                     /* last_used     */ 0,
                     /* args          */ std::vector<std::string>(),
                     /* loaded_info   */ {},
+                    /* progress      */ {},
                     /* exit_code     */ 0,
                     /* stop_timeout  */ DEFAULT_STOP_TIMEOUT,
                     /* multimodal    */ mtmd_caps{false, false},
@@ -1140,6 +1142,9 @@ void server_models::update_status(const std::string & name, const update_status_
         if (!args.loaded_info.is_null()) {
             meta.loaded_info = args.loaded_info;
         }
+        if (!args.progress.is_null()) {
+            meta.progress = args.progress;
+        }
     }
     // broadcast status change to SSE
     {
@@ -1152,6 +1157,9 @@ void server_models::update_status(const std::string & name, const update_status_
         if (!args.loaded_info.is_null()) {
             data["info"] = args.loaded_info;
         }
+        if (!args.progress.is_null()) {
+            data["progress"] = args.progress;
+        }
         // note: notify_sse doesn't acquire the lock, so no deadlock here
         notify_sse("status_change", name, data);
     }
@@ -1322,8 +1330,12 @@ void server_models::handle_child_state(const std::string & name, const std::stri
     switch (state) {
         case SERVER_STATE_LOADING:
             {
-                // do nothing for now
-                // TODO: report loading progress for first load and wakeup from sleep
+                update_status(name, {
+                    SERVER_MODEL_STATUS_LOADING,
+                    0,
+                    nullptr, // no loaded_info yet
+                    payload,
+                });
             } break;
         case SERVER_STATE_READY:
             {
@@ -1331,7 +1343,8 @@ void server_models::handle_child_state(const std::string & name, const std::stri
                     SERVER_MODEL_STATUS_LOADED,
                     0,
                     // note: payload can be empty if this is a wakeup from sleep
-                    payload.size() > 0 ? payload : nullptr
+                    payload.size() > 0 ? payload : nullptr,
+                    {}, // reset progress info
                 });
             } break;
         case SERVER_STATE_SLEEPING:
@@ -1384,6 +1397,7 @@ void server_child::notify_to_router(const std::string & state, const json & payl
         {"state", state},
         {"payload", payload},
     };
+    std::lock_guard<std::mutex> lk(mtx_stdout);
     common_log_pause(common_log_main());
     fflush(stdout);
     fprintf(stdout, "%s%s\n", CMD_CHILD_TO_ROUTER_STATE, safe_json_to_str(data).c_str());

tools/server/server-models.h

@@ -72,6 +72,7 @@ struct server_model_meta {
     int64_t last_used = 0; // for LRU unloading
     std::vector<std::string> args; // args passed to the model instance, will be populated by render_args()
     json loaded_info; // info to be reflected via /v1/models endpoint ; if in DOWNLOADING state, it should contain download progress info
+    json progress; // reflect load or download progress info, if any
     int exit_code = 0; // exit code of the model instance process (only valid if status == FAILED)
     int stop_timeout = 0; // seconds to wait before force-killing the model instance during shutdown
     mtmd_caps multimodal; // multimodal capabilities
@@ -170,12 +171,14 @@ public:
     // to stop the download, call unload()
     void download(common_params_model && model, common_download_opts && opts);
 
-    // update the status of a model instance (thread-safe)
     struct update_status_args {
         server_model_status status;
         int exit_code = 0; // only valid if status == UNLOADED
         json loaded_info = nullptr;
+        json progress = nullptr;
     };
+    // update the status of a model instance (thread-safe)
+    // also send SSE notification to /models/sse endpoint
     void update_status(const std::string & name, const update_status_args & args);
     void update_download_progress(const std::string & name, const common_download_progress & progress, bool done, bool ok = true);
 
@@ -208,6 +211,9 @@ public:
 };
 
 struct server_child {
+    // serializes the notify_to_router writes
+    std::mutex mtx_stdout;
+
     // return true if the current process is a child server instance
     bool is_child();
 

tools/server/server-schema.cpp

@@ -14,6 +14,9 @@ std::vector<std::unique_ptr<field>> make_llama_cmpl_schema(const common_params &
         fields.emplace_back(f);
     };
 
+    add((new field_bool("verbose", params.verbose))
+        ->set_desc("Include __verbose field in the response with additional debug information"));
+
     add((new field_bool("timings_per_token", params.timings_per_token))
         ->set_desc("Include prompt processing and text generation speed information in each response"));
 

tools/server/server-tools.cpp

@@ -11,6 +11,7 @@
 #include <cstring>
 #include <climits>
 #include <algorithm>
+#include <unordered_set>
 
 namespace fs = std::filesystem;
 

tools/server/tests/unit/test_chat_completion.py

@@ -603,3 +603,23 @@ def test_chat_completions_token_count():
         })
         assert res.status_code == 200
         assert res.body["input_tokens"] > 5
+
+
+def test_verbose_debug():
+    global server
+    server.start()
+    for verbose in [True, False]:
+        res = server.make_request("POST", "/chat/completions", data={
+            "max_tokens": 2,
+            "messages": [
+                {"role": "system", "content": "Book"},
+                {"role": "user", "content": "What is the best book"},
+            ],
+            "verbose": verbose,
+        })
+        assert res.status_code == 200
+        if verbose:
+            assert "__verbose" in res.body
+            assert "Book" in res.body["__verbose"]["prompt"]
+        else:
+            assert "__verbose" not in res.body