fix free + swift

Author: ggerganov

examples/batched.swift/Sources/main.swift

@@ -214,7 +214,7 @@ let t_main_end = ggml_time_us()
 
 print("decoded \(n_decode) tokens in \(String(format: "%.2f", Double(t_main_end - t_main_start) / 1_000_000.0)) s, speed: \(String(format: "%.2f", Double(n_decode) / (Double(t_main_end - t_main_start) / 1_000_000.0))) t/s\n")
 
-llama_print_timings(context, smpl, nil)
+llama_print_timings(context, smpl)
 
 private func tokenize(text: String, add_bos: Bool) -> [llama_token] {
     let utf8Count = text.utf8.count

examples/parallel/parallel.cpp

@@ -51,7 +51,6 @@ static std::vector<std::string> k_prompts = {
 struct client {
     ~client() {
         if (ctx_sampling) {
-            llama_sampling_free(ctx_sampling->smpl);
             llama_sampling_free(ctx_sampling);
         }
     }

examples/server/server.cpp

@@ -663,7 +663,6 @@ struct server_context {
         // Clear any sampling context
         for (server_slot & slot : slots) {
             if (slot.ctx_sampling != nullptr) {
-                llama_sampling_free(slot.ctx_sampling->smpl);
                 llama_sampling_free(slot.ctx_sampling);
             }
         }
@@ -1088,7 +1087,6 @@ struct server_context {
 
         {
             if (slot.ctx_sampling != nullptr) {
-                llama_sampling_free(slot.ctx_sampling->smpl);
                 llama_sampling_free(slot.ctx_sampling);
             }
 

examples/speculative/speculative.cpp

@@ -600,7 +600,6 @@ int main(int argc, char ** argv) {
 
     llama_sampling_free(ctx_sampling);
     for (int s = 0; s < n_seq_dft; ++s) {
-        llama_sampling_free(drafts[s].ctx_sampling->smpl);
         llama_sampling_free(drafts[s].ctx_sampling);
     }