@fugood/llama.node 1.4.11 → 1.4.13

package/src/llama.cpp/src/llama-context.h

@@ -70,6 +70,18 @@ struct llama_context {
     float * get_embeddings_ith(int32_t i);
     float * get_embeddings_seq(llama_seq_id seq_id);
 
+    llama_token * get_sampled_tokens() const;
+    llama_token   get_sampled_token_ith(int32_t idx);
+
+    float * get_sampled_logits_ith(int32_t idx);
+    size_t  get_sampled_logits_count(int32_t idx);
+
+    float * get_sampled_probs_ith(int32_t idx);
+    size_t  get_sampled_probs_count(int32_t idx);
+
+    const llama_token * get_sampled_candidates_ith(int32_t idx);
+    size_t get_sampled_candidates_count(int32_t idx);
+
     void attach_threadpool(
             ggml_threadpool_t threadpool,
             ggml_threadpool_t threadpool_batch);
@@ -192,10 +204,13 @@ private:
 
     // Make sure enough space is available for outputs.
     // Returns max number of outputs for which space was reserved.
-    uint32_t output_reserve(int32_t n_outputs);
+    uint32_t output_reserve(int32_t n_outputs, const llama_batch & batch);
 
     void output_reorder();
 
+    // map the output row index `i` to batch index
+    int64_t output_resolve_row(int32_t i) const;
+
     //
     // graph
     //
@@ -213,6 +228,8 @@ public:
     ggml_cgraph * graph_reserve(
         uint32_t n_tokens, uint32_t n_seqs, uint32_t n_outputs, const llama_memory_context_i * mctx, bool split_only = false, size_t * sizes = nullptr);
 
+    bool set_sampler(llama_seq_id seq_id, llama_sampler * sampler);
+
 private:
     llm_graph_params graph_params(
                         llm_graph_result * res,
@@ -252,6 +269,31 @@ private:
     size_t  embd_size = 0; // capacity (of floats) for embeddings
     float * embd      = nullptr;
 
+    // TODO: simplify
+    struct sampling_info {
+        std::map<llama_seq_id, llama_sampler *> samplers;
+
+        float       * logits      = nullptr;
+        size_t        logits_size = 0;
+
+        llama_token * sampled      = nullptr;
+        size_t        sampled_size = 0;
+
+        float       * probs        = nullptr;
+        size_t        probs_size   = 0;
+
+        llama_token * candidates   = nullptr;
+        size_t        candidates_size = 0;
+
+        std::vector<uint32_t> logits_count;
+        std::vector<uint32_t> probs_count;
+        std::vector<uint32_t> candidates_count;
+
+        std::vector<llama_token> token_ids_full_vocab;
+    };
+
+    sampling_info sampling;
+
     // sequence embeddings output (map of [n_embd] vectors)
     // populated only when pooling_type != LLAMA_POOLING_TYPE_NONE
     std::map<llama_seq_id, std::vector<float>> embd_seq;

package/src/llama.cpp/src/llama-grammar.cpp

@@ -369,6 +369,44 @@ static void print_rule(
     fprintf(file, "\n");
 }
 
+//
+// Regex utilities
+//
+
+size_t llama_grammar_trigger_pattern::find(const std::string & input) const {
+    auto find_start_pos = [](const std::smatch & match) {
+        // get from the first matched capturing group to the end of the string
+        size_t start = std::string::npos;
+        for (auto i = 1u; i < match.size(); i++) {
+            if (match.length(i) > 0) {
+                start = match.position(i);
+                break;
+            }
+        }
+        if (start == std::string::npos) {
+            start = match.position(0);
+        }
+        return start;
+    };
+
+    if (!pattern.empty() && pattern.front() == '^' && pattern.back() == '$') {
+        // match against the entire input
+        std::smatch match;
+        if (std::regex_match(input, match, regex)) {
+            return find_start_pos(match);
+        }
+    }
+
+    // search anywhere
+    std::smatch match;
+    if (std::regex_search(input, match, regex)) {
+        return find_start_pos(match);
+    }
+
+    return std::string::npos;
+}
+
+
 //
 // implementation
 //
@@ -1312,21 +1350,10 @@ void llama_grammar_accept_impl(struct llama_grammar & grammar, llama_token token
             grammar.trigger_buffer_positions.push_back(std::make_pair(token, position));
             grammar.trigger_buffer += piece;
 
-            std::smatch match;
             for (const auto & trigger_pattern : grammar.trigger_patterns) {
-                if (std::regex_match(grammar.trigger_buffer, match, trigger_pattern.regex)) {
+                auto start = trigger_pattern.find(grammar.trigger_buffer);
+                if (start != std::string::npos) {
                     grammar.awaiting_trigger = false;
-                    // get from the first matched capturing group to the end of the string
-                    size_t start = std::string::npos;
-                    for (auto i = 1u; i < match.size(); i++) {
-                        if (match.length(i) > 0) {
-                            start = match.position(i);
-                            break;
-                        }
-                    }
-                    if (start == std::string::npos) {
-                        start = match.position(0);
-                    }
 
                     // replay tokens that overlap with [start, end)
                     for (const auto & [tok, tok_pos] : grammar.trigger_buffer_positions) {

package/src/llama.cpp/src/llama-grammar.h

@@ -119,6 +119,8 @@ struct llama_grammar_parser {
 struct llama_grammar_trigger_pattern {
     std::string pattern;
     std::regex  regex;
+
+    size_t find(const std::string & input) const;
 };
 
 struct llama_grammar {

package/src/llama.cpp/src/llama-graph.cpp

@@ -12,6 +12,7 @@
 #include <cassert>
 #include <cmath>
 #include <cstring>
+#include <unordered_set>
 
 void llm_graph_input_embd::set_input(const llama_ubatch * ubatch) {
     if (ubatch->token) {
@@ -32,7 +33,7 @@ bool llm_graph_input_embd::can_reuse(const llm_graph_params & params) {
     bool res = true;
 
     res &= (!tokens && !params.ubatch.token) || (tokens && tokens->ne[0] == params.ubatch.n_tokens);
-    res &= (!embd   && !params.ubatch.embd)  || (embd   &&   embd->ne[0] == params.ubatch.n_tokens);
+    res &= (!embd   && !params.ubatch.embd)  || (embd   &&   embd->ne[1] == params.ubatch.n_tokens);
 
     return res;
 }
@@ -62,7 +63,7 @@ void llm_graph_input_pos::set_input(const llama_ubatch * ubatch) {
 bool llm_graph_input_pos::can_reuse(const llm_graph_params & params) {
     bool res = true;
 
-    res &= pos->ne[0] == params.ubatch.n_tokens;
+    res &= pos->ne[0] == params.ubatch.n_tokens*n_pos_per_embd;
 
     return res;
 }
@@ -521,6 +522,43 @@ bool llm_graph_input_mem_hybrid::can_reuse(const llm_graph_params & params) {
     return res;
 }
 
+void llm_graph_input_sampling::set_input(const llama_ubatch * ubatch) {
+    // set the inputs only for the active samplers in the current ubatch
+    std::unordered_set<llama_seq_id> active_samplers;
+    for (uint32_t i = 0; i < ubatch->n_tokens; i++) {
+        if (ubatch->output[i]) {
+            llama_seq_id seq_id = ubatch->seq_id[i][0];
+            active_samplers.insert(seq_id);
+        }
+    }
+
+    for (auto seq_id : active_samplers) {
+        if (samplers.find(seq_id) == samplers.end()) {
+            continue;
+        }
+
+        auto & sampler = samplers[seq_id];
+
+        if (sampler->iface->backend_set_input) {
+            sampler->iface->backend_set_input(sampler);
+        }
+    }
+}
+
+bool llm_graph_input_sampling::can_reuse(const llm_graph_params & params) {
+    if (samplers.size() != params.samplers.size()) {
+        return false;
+    }
+
+    for (const auto & [seq_id, sampler] : params.samplers) {
+        if (samplers[seq_id] != sampler) {
+            return false;
+        }
+    }
+
+    return true;
+}
+
 //
 // llm_graph_result
 //
@@ -541,6 +579,10 @@ void llm_graph_result::reset() {
     t_logits      = nullptr;
     t_embd        = nullptr;
     t_embd_pooled = nullptr;
+    t_sampled.clear();
+    t_sampled_probs.clear();
+    t_sampled_logits.clear();
+    t_candidates.clear();
 
     params = {};
 
@@ -565,6 +607,38 @@ void llm_graph_result::set_inputs(const llama_ubatch * ubatch) {
     }
 }
 
+void llm_graph_result::set_outputs() {
+    if (t_logits != nullptr) {
+        ggml_set_output(t_logits);
+    }
+    if (t_embd != nullptr) {
+        ggml_set_output(t_embd);
+    }
+    if (t_embd_pooled != nullptr) {
+        ggml_set_output(t_embd_pooled);
+    }
+    for (auto & [seq_id, t] : t_sampled) {
+        if (t != nullptr) {
+            ggml_set_output(t);
+        }
+    }
+    for (auto & [seq_id, t] : t_sampled_probs) {
+        if (t != nullptr) {
+            ggml_set_output(t);
+        }
+    }
+    for (auto & [seq_id, t] : t_sampled_logits) {
+        if (t != nullptr) {
+            ggml_set_output(t);
+        }
+    }
+    for (auto & [seq_id, t] : t_candidates) {
+        if (t != nullptr) {
+            ggml_set_output(t);
+        }
+    }
+}
+
 bool llm_graph_result::can_reuse(const llm_graph_params & params) {
     if (!this->params.allow_reuse(params)) {
         if (debug > 1) {
@@ -646,6 +720,7 @@ llm_graph_context::llm_graph_context(const llm_graph_params & params) :
     loras            (params.loras),
     mctx             (params.mctx),
     cross            (params.cross),
+    samplers         (params.samplers),
     cb_func          (params.cb),
     res              (params.res),
     ctx0             (res->get_ctx()),
@@ -1251,6 +1326,10 @@ ggml_tensor * llm_graph_context::build_inp_embd(ggml_tensor * tok_embd) const {
 
     res->add_input(std::move(inp));
 
+    // make sure the produced embeddings are immediately materialized in the ggml graph
+    // ref: https://github.com/ggml-org/llama.cpp/pull/18599
+    ggml_build_forward_expand(gf, cur);
+
     return cur;
 }
 
@@ -1834,8 +1913,10 @@ llm_graph_input_attn_kv_iswa * llm_graph_context::build_attn_inp_kv_iswa() const
 
         inp->self_kq_mask = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_kv, n_tokens/n_stream, 1, n_stream);
         ggml_set_input(inp->self_kq_mask);
+        ggml_set_name(inp->self_kq_mask, "self_kq_mask");
 
         inp->self_kq_mask_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask, GGML_TYPE_F16) : inp->self_kq_mask;
+        ggml_set_name(inp->self_kq_mask_cnv, "self_kq_mask_cnv");
     }
 
     {
@@ -1848,8 +1929,10 @@ llm_graph_input_attn_kv_iswa * llm_graph_context::build_attn_inp_kv_iswa() const
 
         inp->self_kq_mask_swa = ggml_new_tensor_4d(ctx0, GGML_TYPE_F32, n_kv, n_tokens/n_stream, 1, n_stream);
         ggml_set_input(inp->self_kq_mask_swa);
+        ggml_set_name(inp->self_kq_mask_swa, "self_kq_mask_swa");
 
         inp->self_kq_mask_swa_cnv = cparams.flash_attn ? ggml_cast(ctx0, inp->self_kq_mask_swa, GGML_TYPE_F16) : inp->self_kq_mask_swa;
+        ggml_set_name(inp->self_kq_mask_swa_cnv, "self_kq_mask_swa_cnv");
     }
 
     return (llm_graph_input_attn_kv_iswa *) res->add_input(std::move(inp));
@@ -1988,14 +2071,18 @@ llm_graph_input_mem_hybrid * llm_graph_context::build_inp_mem_hybrid() const {
 void llm_graph_context::build_dense_out(
     ggml_tensor * dense_2,
     ggml_tensor * dense_3) const {
-    if (!cparams.embeddings || dense_2 == nullptr || dense_3 == nullptr) {
+    if (!cparams.embeddings || !(dense_2 || dense_3)) {
         return;
     }
     ggml_tensor * cur = res->t_embd_pooled != nullptr ? res->t_embd_pooled : res->t_embd;
     GGML_ASSERT(cur != nullptr && "missing t_embd_pooled/t_embd");
 
-    cur = ggml_mul_mat(ctx0, dense_2, cur);
-    cur = ggml_mul_mat(ctx0, dense_3, cur);
+    if (dense_2) {
+        cur = ggml_mul_mat(ctx0, dense_2, cur);
+    }
+    if (dense_3) {
+        cur = ggml_mul_mat(ctx0, dense_3, cur);
+    }
     cb(cur, "result_embd_pooled", -1);
     res->t_embd_pooled = cur;
     ggml_build_forward_expand(gf, cur);
@@ -2086,6 +2173,87 @@ void llm_graph_context::build_pooling(
     ggml_build_forward_expand(gf, cur);
 }
 
+void llm_graph_context::build_sampling() const {
+    if (samplers.empty() || !res->t_logits) {
+        return;
+    }
+
+    auto inp_sampling = std::make_unique<llm_graph_input_sampling>(samplers);
+    res->add_input(std::move(inp_sampling));
+
+    std::map<llama_seq_id, int32_t> seq_to_logit_row;
+    int32_t logit_row_idx = 0;
+
+    for (uint32_t i = 0; i < ubatch.n_tokens; i++) {
+        if (ubatch.output[i]) {
+            llama_seq_id seq_id = ubatch.seq_id[i][0];
+            seq_to_logit_row[seq_id] = logit_row_idx;
+            logit_row_idx++;
+        }
+    }
+
+    // res->t_logits will contain logits for all tokens that want the logits calculated (logits=1 or output=1)
+    GGML_ASSERT(res->t_logits != nullptr && "missing t_logits tensor");
+
+    // add a dummy row of logits
+    // this trick makes the graph static, regardless of which samplers are activated
+    // this is important in order to minimize graph reallocations
+    // TODO: use `ggml_build_forward_select()` when available (https://github.com/ggml-org/llama.cpp/pull/18550)
+    ggml_tensor * logits_t = ggml_pad(ctx0, res->t_logits, 0, 1, 0, 0);
+
+    for (const auto & [seq_id, sampler] : samplers) {
+        const auto it = seq_to_logit_row.find(seq_id);
+
+        // inactive samplers always work on the first row
+        const auto row_idx = seq_to_logit_row.find(seq_id) != seq_to_logit_row.end() ? it->second : 0;
+
+        ggml_tensor * logits_seq = ggml_view_1d(ctx0, logits_t, logits_t->ne[0], row_idx * logits_t->nb[1]);
+        ggml_format_name(logits_seq, "logits_seq_%d", seq_id);
+
+        struct llama_sampler_data data = {
+            /*.logits      =*/ logits_seq,
+            /*.probs       =*/ nullptr,
+            /*.sampled     =*/ nullptr,
+            /*.candidates  =*/ nullptr,
+        };
+
+        assert(sampler->iface->backend_apply);
+        sampler->iface->backend_apply(sampler, ctx0, gf, &data);
+
+        if (data.sampled != nullptr) {
+            res->t_sampled[seq_id] = data.sampled;
+            ggml_build_forward_expand(gf, data.sampled);
+        }
+
+        if (data.probs != nullptr) {
+            res->t_sampled_probs[seq_id] = data.probs;
+            ggml_build_forward_expand(gf, data.probs);
+        }
+
+        if (data.logits != nullptr) {
+            res->t_sampled_logits[seq_id] = data.logits;
+            ggml_build_forward_expand(gf, data.logits);
+        }
+
+        if (data.candidates != nullptr) {
+            res->t_candidates[seq_id] = data.candidates;
+            ggml_build_forward_expand(gf, data.candidates);
+        }
+    }
+
+    // TODO: Call llama_sampler_accept_ggml after all samplers have been applied.
+    /*
+    for (const auto & [seq_id, sampler] : samplers) {
+        if (auto it = res->t_sampled.find(seq_id); it != res->t_sampled.end()) {
+            ggml_tensor * selected_token = it->second;
+            if (selected_token != nullptr) {
+                llama_sampler_accept_ggml(sampler, ctx0, gf, selected_token);
+            }
+        }
+    }
+    */
+}
+
 int32_t llama_relative_position_bucket(llama_pos x, llama_pos y, uint64_t n_buckets, bool bidirectional) {
     // TODO move to hparams if a T5 variant appears that uses a different value
     const int64_t max_distance = 128;

package/src/llama.cpp/src/llama-graph.h

@@ -10,6 +10,7 @@
 #include <memory>
 #include <set>
 #include <functional>
+#include <map>
 
 struct ggml_cgraph;
 struct ggml_context;
@@ -396,6 +397,18 @@ public:
     const llama_memory_hybrid_context * mctx;
 };
 
+class llm_graph_input_sampling : public llm_graph_input_i {
+public:
+    llm_graph_input_sampling(std::map<llama_seq_id, llama_sampler *> samplers) :
+        samplers(std::move(samplers)) { }
+    virtual ~llm_graph_input_sampling() = default;
+
+    void set_input(const llama_ubatch * ubatch) override;
+    bool can_reuse(const llm_graph_params & params) override;
+
+    std::map<llama_seq_id, llama_sampler *> samplers;
+};
+
 //
 // llm_graph_result
 //
@@ -429,6 +442,23 @@ struct llm_graph_params {
     const llama_memory_context_i * mctx;
     const llama_cross            * cross;
 
+    std::map<llama_seq_id, llama_sampler *> samplers;
+
+    static bool samplers_equal(
+          const std::map<llama_seq_id, llama_sampler *> & lhs,
+          const std::map<llama_seq_id, llama_sampler *> & rhs) {
+        if (lhs.size() != rhs.size()) {
+            return false;
+        }
+        for (const auto & [seq_id, sampler] : lhs) {
+            auto it = rhs.find(seq_id);
+            if (it == rhs.end() || it->second != sampler) {
+                return false;
+            }
+        }
+        return true;
+    }
+
     uint32_t n_outputs;
 
     llm_graph_cb cb;
@@ -468,15 +498,36 @@ struct llm_graph_params {
             return false;
         }
 
+        if (n_outputs != other.n_outputs) {
+            return false;
+        }
+
+        if (!samplers_equal(samplers, other.samplers)) {
+            return false;
+        }
+
+        if (samplers.size() > 0) {
+            if (!ubatch.data || !other.ubatch.data) {
+                return false;
+            }
+
+            // check that the outputs are the same for all samplers
+            for (uint32_t i = 0; i < ubatch.n_tokens; ++i) {
+                if (ubatch.output[i]    != other.ubatch.output[i] ||
+                    ubatch.seq_id[i][0] != other.ubatch.seq_id[i][0]) {
+                    return false;
+                }
+            }
+        }
+
         return
             cparams.embeddings  == other.cparams.embeddings  &&
             cparams.causal_attn == other.cparams.causal_attn &&
-            arch      == other.arch  &&
-            gtype     == other.gtype &&
-            cvec      == other.cvec  &&
-            loras     == other.loras &&
-            cross     == other.cross &&
-            n_outputs == other.n_outputs;
+            arch  == other.arch  &&
+            gtype == other.gtype &&
+            cvec  == other.cvec  &&
+            loras == other.loras &&
+            cross == other.cross;
     }
 };
 
@@ -499,6 +550,7 @@ public:
     void reset();
 
     void set_inputs(const llama_ubatch * ubatch);
+    void set_outputs();
 
     // try to update the existing graph result using the new graph parameters in order to reuse it
     // this can only be done if we determine that the resulting graph using the new graph parameters
@@ -517,6 +569,11 @@ public:
     ggml_tensor * t_embd        = nullptr;
     ggml_tensor * t_embd_pooled = nullptr;
 
+    std::map<llama_seq_id, ggml_tensor*> t_sampled_logits;
+    std::map<llama_seq_id, ggml_tensor*> t_candidates;
+    std::map<llama_seq_id, ggml_tensor*> t_sampled;
+    std::map<llama_seq_id, ggml_tensor*> t_sampled_probs;
+
     std::vector<llm_graph_input_ptr> inputs;
 
     ggml_context_ptr ctx_compute;
@@ -592,6 +649,8 @@ struct llm_graph_context {
     const llama_memory_context_i * mctx;
     const llama_cross            * cross;
 
+    std::map<llama_seq_id, llama_sampler *> samplers;
+
     const llm_graph_cb & cb_func;
 
     llm_graph_result * res;
@@ -832,6 +891,12 @@ struct llm_graph_context {
             ggml_tensor * cls_out,
             ggml_tensor * cls_out_b) const;
 
+    //
+    // sampling (backend sampling)
+    //
+
+    void build_sampling() const;
+
     //
     // dense (out)
     //

package/src/llama.cpp/src/llama-hparams.cpp

@@ -72,6 +72,10 @@ uint32_t llama_hparams::n_embd_inp() const {
     return n_embd_inp;
 }
 
+uint32_t llama_hparams::get_n_embd_out() const {
+    return n_embd_out > 0 ? n_embd_out : n_embd;
+}
+
 uint32_t llama_hparams::n_embd_k_gqa(uint32_t il) const {
     const uint32_t n_head_kv = this->n_head_kv(il);
 

package/src/llama.cpp/src/llama-hparams.h

@@ -105,9 +105,9 @@ struct llama_hparams {
 
     float    rope_attn_factor = 1.0f;
     float    rope_freq_base_train;
-    float    rope_freq_base_train_swa;
+    float    rope_freq_base_train_swa  = 10000.0f;
     float    rope_freq_scale_train;
-    float    rope_freq_scale_train_swa;
+    float    rope_freq_scale_train_swa = 1.0f;
 
     uint32_t n_ctx_orig_yarn;
     float    rope_yarn_log_mul = 0.0f;
@@ -123,10 +123,11 @@ struct llama_hparams {
     llama_swa_type swa_type = LLAMA_SWA_TYPE_NONE;
     // the size of the sliding window (0 - no SWA)
     uint32_t n_swa = 0;
-    // if swa_layers[il] == true, then layer il is SWA
-    // if swa_layers[il] == false, then layer il is dense (i.e. non-SWA)
+    // if swa_layers[il] == 1, then layer il is SWA
+    // if swa_layers[il] == 0, then layer il is dense (i.e. non-SWA)
     // by default, all layers are dense
-    std::array<bool, LLAMA_MAX_LAYERS> swa_layers;
+    // note: using uint32_t type for compatibility reason
+    std::array<uint32_t, LLAMA_MAX_LAYERS> swa_layers;
 
     // for State Space Models
     uint32_t ssm_d_conv  = 0;
@@ -161,6 +162,9 @@ struct llama_hparams {
     // for Classifiers
     uint32_t n_cls_out = 1;
 
+    // output embedding dimension (0 = use n_embd)
+    uint32_t n_embd_out = 0;
+
     // llama4 smallthinker
     uint32_t n_moe_layer_step        = 0;
     uint32_t n_no_rope_layer_step    = 4;
@@ -233,6 +237,9 @@ struct llama_hparams {
     // dimension of main + auxiliary input embeddings
     uint32_t n_embd_inp() const;
 
+    // dimension of output embeddings
+    uint32_t get_n_embd_out() const;
+
     // dimension of key embeddings across all k-v heads
     uint32_t n_embd_k_gqa(uint32_t il = 0) const;
 

package/src/llama.cpp/src/llama-kv-cache.h

@@ -305,7 +305,7 @@ public:
             bool do_shift,
             stream_copy_info sc_info);
 
-    // used to create a batch procesing context from a batch
+    // used to create a batch processing context from a batch
     llama_kv_cache_context(
             llama_kv_cache * kv,
             slot_info_vec_t sinfos,

package/src/llama.cpp/src/llama-mmap.cpp

@@ -240,9 +240,10 @@ struct llama_file::impl {
                 throw std::runtime_error("unexpectedly reached end of file");
             }
         } else {
-            bool successful = false;
-            while (!successful) {
-                off_t ret = read(fd, ptr, len);
+            size_t bytes_read = 0;
+            while (bytes_read < len) {
+                const size_t to_read = len - bytes_read;
+                ssize_t ret = ::read(fd, reinterpret_cast<char *>(ptr) + bytes_read, to_read);
 
                 if (ret == -1) {
                     if (errno == EINTR) {
@@ -251,10 +252,16 @@ struct llama_file::impl {
                     throw std::runtime_error(format("read error: %s", strerror(errno)));
                 }
                 if (ret == 0) {
+                    // EOF: allow if this read was only pulling alignment padding past file end
+                    off_t pos = lseek(fd, 0, SEEK_CUR);
+                    if (pos != -1 && (size_t) pos == size) {
+                        std::memset(reinterpret_cast<char *>(ptr) + bytes_read, 0, len - bytes_read);
+                        return;
+                    }
                     throw std::runtime_error("unexpectedly reached end of file");
                 }
 
-                successful = true;
+                bytes_read += (size_t) ret;
             }
         }
     }

package/src/llama.cpp/src/llama-model-loader.cpp

@@ -462,6 +462,29 @@ namespace GGUFMeta {
         return get_key_or_arr(llm_kv(kid), result, n, required);
     }
 
+    bool llama_model_loader::get_key_or_arr(enum llm_kv kid, uint32_t & result, bool required) {
+        const std::string key = llm_kv(kid);
+
+        const int id = gguf_find_key(meta.get(), key.c_str());
+
+        if (id < 0) {
+            if (required) {
+                throw std::runtime_error(format("key not found in model: %s", key.c_str()));
+            }
+            return false;
+        }
+
+        // throw and error if type is an array
+        if (gguf_get_kv_type(meta.get(), id) == GGUF_TYPE_ARRAY) {
+            if (required) {
+                throw std::runtime_error(format("expected scalar, found array for key: %s", key.c_str()));
+            }
+            return false;
+        }
+
+        return get_key(key, result, required);
+    }
+
     // TODO: this is not very clever - figure out something better
     template bool llama_model_loader::get_key_or_arr<std::array<int, 4>>(enum llm_kv kid, std::array<int, 4> & result, uint32_t n, bool required);
     template bool llama_model_loader::get_key_or_arr<std::array<uint32_t, 512>>(enum llm_kv kid, std::array<uint32_t, 512> & result, uint32_t n, bool required);