rename functions to conformer

tools/mtmd/CMakeLists.txt

@@ -15,9 +15,9 @@ add_library(mtmd
             clip-graph.h
             models/models.h
             models/cogvlm.cpp
+            models/conformer.cpp
             models/internvl.cpp
             models/kimivl.cpp
-            models/lfm2-audio-enc.cpp
             models/llama4.cpp
             models/llava.cpp
             models/minicpmv.cpp

tools/mtmd/clip-impl.h

@@ -133,7 +133,7 @@
 #define TN_TOK_BOI         "v.boi"
 #define TN_TOK_EOI         "v.eoi"
 
-// lfm2
+// (conformer) lfm2
 #define TN_PRE_ENCODE_OUT  "a.pre_encode.out.%s"
 #define TN_FFN_NORM        "%s.blk.%d.ffn_norm.%s"
 #define TN_FFN_NORM_1      "%s.blk.%d.ffn_norm_1.%s"

tools/mtmd/clip.cpp

@@ -844,7 +844,7 @@ static ggml_cgraph * clip_image_build_graph(clip_ctx * ctx, const clip_image_f32
             } break;
         case PROJECTOR_TYPE_LFM2A:
             {
-                builder = std::make_unique<clip_graph_lfm2_audio_enc>(ctx, img);
+                builder = std::make_unique<clip_graph_conformer>(ctx, img);
             } break;
         default:
             GGML_ABORT("missing cgraph builder");

tools/mtmd/models/conformer.cpp

@@ -1,6 +1,6 @@
 #include "models.h"
 
-ggml_cgraph * clip_graph_lfm2_audio_enc::build() {
+ggml_cgraph * clip_graph_conformer::build() {
     const int n_frames   = img.nx;
     const int n_pos      = n_frames / 2;
     const int n_pos_embd = (((((n_frames + 1) / 2) + 1) / 2 + 1) / 2) * 2 - 1;

tools/mtmd/models/models.h

@@ -57,7 +57,7 @@ struct clip_graph_whisper_enc : clip_graph {
     ggml_cgraph * build() override;
 };
 
-struct clip_graph_lfm2_audio_enc : clip_graph {
-    clip_graph_lfm2_audio_enc(clip_ctx * ctx, const clip_image_f32 & img) : clip_graph(ctx, img) {}
+struct clip_graph_conformer : clip_graph {
+    clip_graph_conformer(clip_ctx * ctx, const clip_image_f32 & img) : clip_graph(ctx, img) {}
     ggml_cgraph * build() override;
 };

tools/mtmd/mtmd-audio.cpp

@@ -537,9 +537,10 @@ bool mtmd_audio_preprocessor_whisper::preprocess(
 }
 
 //
-// mtmd_audio_preprocessor_lfm2
+// mtmd_audio_preprocessor_conformer
 //
-void mtmd_audio_preprocessor_lfm2::initialize() {
+
+void mtmd_audio_preprocessor_conformer::initialize() {
     g_cache.fill_sin_cos_table(hparams.audio_n_fft);
     g_cache.fill_hann_window(hparams.audio_window_len, true);
     g_cache.fill_mel_filterbank_matrix(
@@ -548,7 +549,7 @@ void mtmd_audio_preprocessor_lfm2::initialize() {
         hparams.audio_sample_rate);
 }
 
-bool mtmd_audio_preprocessor_lfm2::preprocess(
+bool mtmd_audio_preprocessor_conformer::preprocess(
         const float * samples,
         size_t n_samples,
         std::vector<mtmd_audio_mel> & output) {

tools/mtmd/mtmd-audio.h

@@ -33,8 +33,8 @@ struct mtmd_audio_preprocessor_whisper : mtmd_audio_preprocessor {
     bool preprocess(const float * samples, size_t n_samples, std::vector<mtmd_audio_mel> & output) override;
 };
 
-struct mtmd_audio_preprocessor_lfm2: mtmd_audio_preprocessor {
-    mtmd_audio_preprocessor_lfm2(const clip_ctx * ctx) : mtmd_audio_preprocessor(ctx) {}
+struct mtmd_audio_preprocessor_conformer : mtmd_audio_preprocessor {
+    mtmd_audio_preprocessor_conformer(const clip_ctx * ctx) : mtmd_audio_preprocessor(ctx) {}
     void initialize() override;
     bool preprocess(const float * samples, size_t n_samples, std::vector<mtmd_audio_mel> & output) override;
 };

tools/mtmd/mtmd.cpp

@@ -328,7 +328,7 @@ struct mtmd_context {
                 audio_preproc = std::make_unique<mtmd_audio_preprocessor_whisper>(ctx_a);
                 break;
             case PROJECTOR_TYPE_LFM2A:
-                audio_preproc = std::make_unique<mtmd_audio_preprocessor_lfm2>(ctx_a);
+                audio_preproc = std::make_unique<mtmd_audio_preprocessor_conformer>(ctx_a);
                 break;
             default:
                 GGML_ABORT("unsupported audio projector type");