minor : warning fixes

Author: ggerganov

examples/main/main.cpp

@@ -354,7 +354,7 @@ int main(int argc, char ** argv) {
             if ((int)embd.size() > max_embd_size) {
                 auto skipped_tokens = embd.size() - max_embd_size;
                 console_set_color(con_st, CONSOLE_COLOR_ERROR);
-                printf("<<input too long: skipped %" PRIu64 "  token%s>>", skipped_tokens, skipped_tokens != 1 ? "s" : "");
+                printf("<<input too long: skipped %zu token%s>>", skipped_tokens, skipped_tokens != 1 ? "s" : "");
                 console_set_color(con_st, CONSOLE_COLOR_DEFAULT);
                 fflush(stdout);
                 embd.resize(max_embd_size);

ggml-metal.m

@@ -256,10 +256,10 @@ bool ggml_metal_add_buffer(
         if (ctx->buffers[ctx->n_buffers].metal == nil) {
             fprintf(stderr, "%s: failed to allocate '%-16s' buffer, size = %8.2f MB\n", __func__, name, aligned_size / 1024.0 / 1024.0);
             return false;
-        } else {
-            fprintf(stderr, "%s: allocated '%-16s' buffer, size = %8.2f MB\n", __func__, name, aligned_size / 1024.0 / 1024.0);
         }
 
+        fprintf(stderr, "%s: allocated '%-16s' buffer, size = %8.2f MB\n", __func__, name, aligned_size / 1024.0 / 1024.0);
+
         ++ctx->n_buffers;
     }
 
@@ -765,18 +765,23 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_ALIBI:
                         {
+                            if (encoder == nil) {
+                                encoder = [command_buffer computeCommandEncoder];
+                            }
+
                             GGML_ASSERT((src0t == GGML_TYPE_F32));
-                            const int   n_past   = ((int32_t *) src1->data)[0];
+
+                            const int   n_past   = ((int32_t *) src1->data)[0]; UNUSED(n_past);
                             const int   n_head   = ((int32_t *) src1->data)[1];
                             const float max_bias = ((float *)   src1->data)[2];
+
                             if (__builtin_popcount(n_head) != 1) {
                                 GGML_ASSERT(false && "only power-of-two n_head implemented");
                             }
+
                             const int n_heads_log2_floor = 1 << (int) floor(log2(n_head));
                             const float m0 = powf(2.0f, -(max_bias) / n_heads_log2_floor);
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoder];
-                            }
+
                             [encoder setComputePipelineState:ctx->pipeline_alibi_f32];
                             [encoder setBuffer:id_src0 offset:offs_src0 atIndex:0];
                             [encoder setBuffer:id_dst  offset:offs_dst  atIndex:1];