From 1bd8d1870fe68ad6830450426de55c043eafe03f Mon Sep 17 00:00:00 2001
From: Xuan Son Nguyen <son@huggingface.co>
Date: Thu, 18 Dec 2025 23:39:00 +0100
Subject: [PATCH] clean up

---
 tools/mtmd/clip-model.h   | 6 +++---
 tools/mtmd/mtmd-audio.cpp | 2 +-
 tools/mtmd/mtmd-cli.cpp   | 8 ++++++--
 3 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/tools/mtmd/clip-model.h b/tools/mtmd/clip-model.h
index 9dc32cf928..b4c31cdde6 100644
--- a/tools/mtmd/clip-model.h
+++ b/tools/mtmd/clip-model.h
@@ -311,11 +311,11 @@ struct clip_model {
     ggml_tensor * mm_boi = nullptr;
     ggml_tensor * mm_eoi = nullptr;
 
-    // lfm2
+    // lfm2 audio
     std::array<ggml_tensor *, 7> pre_encode_conv_X_w = {nullptr};
     std::array<ggml_tensor *, 7> pre_encode_conv_X_b = {nullptr};
-    ggml_tensor * pre_encode_out_w     = nullptr;
-    ggml_tensor * pre_encode_out_b     = nullptr;
+    ggml_tensor * pre_encode_out_w = nullptr;
+    ggml_tensor * pre_encode_out_b = nullptr;
 
     bool audio_has_avgpool() const {
         return proj_type == PROJECTOR_TYPE_QWEN2A
diff --git a/tools/mtmd/mtmd-audio.cpp b/tools/mtmd/mtmd-audio.cpp
index bf68847da4..e99101184b 100644
--- a/tools/mtmd/mtmd-audio.cpp
+++ b/tools/mtmd/mtmd-audio.cpp
@@ -565,7 +565,7 @@ bool mtmd_audio_preprocessor_conformer::preprocess(
     params.hop_length       = hparams.audio_hop_len;
     params.sample_rate      = hparams.audio_sample_rate;
     params.center_padding   = true;
-    params.preemph          = 0.97f; // disabled
+    params.preemph          = 0.97f;
     params.use_natural_log  = true;
     params.norm_per_feature = true;
 
diff --git a/tools/mtmd/mtmd-cli.cpp b/tools/mtmd/mtmd-cli.cpp
index 00f62d673e..1ba02a5233 100644
--- a/tools/mtmd/mtmd-cli.cpp
+++ b/tools/mtmd/mtmd-cli.cpp
@@ -323,7 +323,9 @@ int main(int argc, char ** argv) {
     LOG_WRN("WARN: This is an experimental CLI for testing multimodal capability.\n");
     LOG_WRN("      For normal use cases, please use the standard llama-cli\n");
 
-    eval_system_prompt_if_present();
+    if (eval_system_prompt_if_present()) {
+        return 1;
+    }
 
     if (is_single_turn) {
         g_is_generating = true;
@@ -383,8 +385,10 @@ int main(int argc, char ** argv) {
                 ctx.n_past = 0;
                 ctx.chat_history.clear();
                 llama_memory_clear(llama_get_memory(ctx.lctx), true);
+                if (eval_system_prompt_if_present()) {
+                    return 1;
+                }
                 LOG("Chat history cleared\n\n");
-                eval_system_prompt_if_present();
                 continue;
             }
             g_is_generating = true;