Merge branch 'upstream' into concedo_experimental

# Conflicts:
#	.github/ISSUE_TEMPLATE/020-enhancement.yml
#	.github/ISSUE_TEMPLATE/030-research.yml
#	.github/ISSUE_TEMPLATE/040-refactor.yml
#	.github/workflows/build.yml
#	Makefile
#	common/CMakeLists.txt
#	examples/CMakeLists.txt
#	examples/infill/infill.cpp
#	examples/lookahead/lookahead.cpp
#	examples/lookup/lookup-stats.cpp
#	examples/lookup/lookup.cpp
#	examples/parallel/parallel.cpp
#	examples/retrieval/retrieval.cpp
#	examples/save-load-state/save-load-state.cpp
#	examples/speculative/speculative.cpp
#	flake.lock
#	ggml/src/ggml-cann/CMakeLists.txt
#	ggml/src/ggml-cann/aclnn_ops.cpp
#	ggml/src/ggml-cann/kernels/CMakeLists.txt
#	ggml/src/ggml-cann/kernels/dup.cpp
#	ggml/src/ggml-cann/kernels/get_row_f16.cpp
#	ggml/src/ggml-cann/kernels/get_row_f32.cpp
#	ggml/src/ggml-cann/kernels/get_row_q4_0.cpp
#	tests/test-arg-parser.cpp
#	tests/test-backend-ops.cpp

examples/server/server.cpp

@@ -176,7 +176,7 @@ struct server_slot {
     // sampling
     json json_schema;
 
-    struct common_sampler_params sparams;
+    struct common_params_sampling sparams;
     struct common_sampler * smpl = nullptr;
 
     llama_token sampled;
@@ -688,7 +688,7 @@ struct server_context {
 
             SLT_INF(slot, "new slot n_ctx_slot = %d\n", slot.n_ctx);
 
-            slot.sparams = params.sparams;
+            slot.sparams = params.sampling;
 
             slot.callback_on_release = [this](int) {
                 queue_tasks.pop_deferred_task();
@@ -744,7 +744,7 @@ struct server_context {
                 }
 
                 // length of the Longest Common Subsequence between the current slot's prompt and the input prompt
-                int cur_lcs_len = longest_common_subsequence(slot.cache_tokens, task.prompt_tokens);
+                int cur_lcs_len = common_lcs(slot.cache_tokens, task.prompt_tokens);
 
                 // fraction of the common subsequence length compared to the current slot's prompt length
                 float cur_similarity = static_cast<float>(cur_lcs_len) / static_cast<int>(slot.cache_tokens.size());
@@ -789,7 +789,7 @@ struct server_context {
     bool launch_slot_with_task(server_slot & slot, const server_task & task) {
         slot_params default_params;
         // Sampling parameter defaults are loaded from the global server context (but individual requests can still override them)
-        auto default_sparams = params.sparams;
+        auto default_sparams = params.sampling;
         const auto & data = task.data;
 
         if (data.count("__oaicompat") != 0) {
@@ -1961,7 +1961,7 @@ struct server_context {
 
                             if (slot.params.cache_prompt) {
                                 // reuse any previously computed tokens that are common with the new prompt
-                                slot.n_past = longest_common_prefix(slot.cache_tokens, prompt_tokens);
+                                slot.n_past = common_lcp(slot.cache_tokens, prompt_tokens);
 
                                 // reuse chunks from the cached prompt by shifting their KV cache in the new position
                                 if (params.n_cache_reuse > 0) {

examples/server/utils.hpp

@@ -24,7 +24,6 @@
 #define DEFAULT_OAICOMPAT_MODEL "gpt-3.5-turbo-0613"
 
 using json = nlohmann::ordered_json;
-using llama_tokens = std::vector<llama_token>;
 
 #define SLT_INF(slot, fmt, ...) LOG_INF("slot %12.*s: id %2d | task %d | " fmt, 12, __func__, (slot).id, (slot).id_task, __VA_ARGS__)
 #define SLT_WRN(slot, fmt, ...) LOG_WRN("slot %12.*s: id %2d | task %d | " fmt, 12, __func__, (slot).id, (slot).id_task, __VA_ARGS__)
@@ -439,62 +438,6 @@ static std::string gen_chatcmplid() {
 // other common utils
 //
 
-static size_t longest_common_prefix(const llama_tokens & a, const llama_tokens & b) {
-    size_t i;
-    for (i = 0; i < a.size() && i < b.size() && a[i] == b[i]; i++) {}
-
-    return i;
-}
-
-static size_t longest_common_subsequence(const llama_tokens & a, const llama_tokens & b) {
-    // check for empty sequences
-    if (a.empty() || b.empty()) {
-        return 0;
-    }
-
-    // get the lengths of the input sequences
-    size_t a_len = a.size();
-    size_t b_len = b.size();
-
-    // initialize the maximum length of the longest common subsequence (LCS)
-    size_t max_length = 0;
-
-    // use two rows instead of a 2D matrix to optimize space
-    std::vector<size_t> prev_row(b_len + 1, 0);
-    std::vector<size_t> curr_row(b_len + 1, 0);
-
-    // iterate through the elements of a
-    for (size_t i = 1; i <= a_len; i++) {
-        // iterate through the elements of b
-        for (size_t j = 1; j <= b_len; j++) {
-            // if elements at the current positions match
-            if (a[i - 1] == b[j - 1]) {
-                // if it's the first element of either sequences, set LCS length to 1
-                if (i == 1 || j == 1) {
-                    curr_row[j] = 1;
-                } else {
-                    // increment LCS length by 1 compared to the previous element
-                    curr_row[j] = prev_row[j - 1] + 1;
-                }
-
-                // update max_length if necessary
-                if (curr_row[j] > max_length) {
-                    max_length = curr_row[j];
-                }
-            } else {
-                // reset LCS length if elements don't match
-                curr_row[j] = 0;
-            }
-        }
-
-        // update the previous row for the next iteration
-        prev_row = curr_row;
-    }
-
-    // return the maximum length of the LCS
-    return max_length;
-}
-
 static bool ends_with(const std::string & str, const std::string & suffix) {
     return str.size() >= suffix.size() && 0 == str.compare(str.size() - suffix.size(), suffix.size(), suffix);
 }