ggml update to b7108 (#12992)

* Revert "vulkan: temporary cary of vulkan fixes (#12971)"

This reverts commit 3a9e8e9fd4.

* ggml update to b7087

* fix argsort on metal

* update to b7108

* fix bakllava regression

This model lacks the metadata for the projector type.

* update to b7209

* fix TopK perf

* only build arm code on arm

llama/llama.cpp/tools/mtmd/clip-impl.h

@@ -39,6 +39,7 @@
 #define KEY_FEATURE_LAYER       "clip.vision.feature_layer"
 #define KEY_PROJ_SCALE_FACTOR   "clip.vision.projector.scale_factor"
 #define KEY_SPATIAL_MERGE_SIZE  "clip.vision.spatial_merge_size"
+#define KEY_IS_DEEPSTACK_LAYERS "clip.vision.is_deepstack_layers"
 
 #define KEY_MM_PATCH_MERGE_TYPE   "clip.vision.mm_patch_merge_type"
 #define KEY_IMAGE_GRID_PINPOINTS  "clip.vision.image_grid_pinpoints"
@@ -63,6 +64,7 @@
 #define TN_PATCH_EMBD      "v.patch_embd.weight"  // not rename tensor with ".0" postfix for backwrad compat
 #define TN_PATCH_EMBD_1    "v.patch_embd.weight.1"
 #define TN_PATCH_BIAS      "v.patch_embd.bias"
+#define TN_ATTN_QKV        "%s.blk.%d.attn_qkv.%s"
 #define TN_ATTN_K          "%s.blk.%d.attn_k.%s"
 #define TN_ATTN_Q          "%s.blk.%d.attn_q.%s"
 #define TN_ATTN_V          "%s.blk.%d.attn_v.%s"
@@ -93,6 +95,9 @@
 #define TN_TOK_IMG_BREAK   "v.token_embd.img_break"     // pixtral
 #define TN_TOK_GLM_BOI     "adapter.boi"                // glm-edge (these embeddings are not in text model)
 #define TN_TOK_GLM_EOI     "adapter.eoi"                // glm-edge (these embeddings are not in text model)
+#define TN_DEEPSTACK_NORM  "v.deepstack.%d.norm.%s"     // qwen3vl deepstack
+#define TN_DEEPSTACK_FC1   "v.deepstack.%d.fc1.%s"      // qwen3vl deepstack
+#define TN_DEEPSTACK_FC2   "v.deepstack.%d.fc2.%s"      // qwen3vl deepstack
 
 // mimicpmv
 #define TN_MINICPMV_POS_EMBD_K "resampler.pos_embed_k"
@@ -116,6 +121,14 @@
 #define TN_MM_NORM_PRE  "mm.a.norm_pre.%s"
 #define TN_MM_NORM_MID  "mm.a.norm_mid.%s"
 
+// cogvlm
+#define TN_MM_POST_FC_NORM "mm.post_fc_norm.%s"
+#define TN_MM_H_TO_4H      "mm.up.%s"
+#define TN_MM_GATE         "mm.gate.%s"
+#define TN_MM_4H_TO_H      "mm.down.%s"
+#define TN_TOK_BOI         "v.boi"
+#define TN_TOK_EOI         "v.eoi"
+
 // align x to upper multiple of n
 #define CLIP_ALIGN(x, n) ((((x) + (n) - 1) / (n)) * (n))
 
@@ -127,6 +140,7 @@ enum projector_type {
     PROJECTOR_TYPE_MINICPMV,
     PROJECTOR_TYPE_GLM_EDGE,
     PROJECTOR_TYPE_QWEN2VL,
+    PROJECTOR_TYPE_QWEN3VL,
     PROJECTOR_TYPE_GEMMA3,
     PROJECTOR_TYPE_IDEFICS3,
     PROJECTOR_TYPE_PIXTRAL,
@@ -139,6 +153,9 @@ enum projector_type {
     PROJECTOR_TYPE_VOXTRAL,
     PROJECTOR_TYPE_LFM2,
     PROJECTOR_TYPE_KIMIVL,
+    PROJECTOR_TYPE_LIGHTONOCR,
+    PROJECTOR_TYPE_COGVLM,
+    PROJECTOR_TYPE_JANUS_PRO,
     PROJECTOR_TYPE_UNKNOWN,
 };
 
@@ -150,6 +167,7 @@ static std::map<projector_type, std::string> PROJECTOR_TYPE_NAMES = {
     { PROJECTOR_TYPE_GLM_EDGE,  "adapter"},
     { PROJECTOR_TYPE_QWEN2VL,   "qwen2vl_merger"},
     { PROJECTOR_TYPE_QWEN25VL,  "qwen2.5vl_merger"},
+    { PROJECTOR_TYPE_QWEN3VL,   "qwen3vl_merger"},
     { PROJECTOR_TYPE_GEMMA3,    "gemma3"},
     { PROJECTOR_TYPE_IDEFICS3,  "idefics3"},
     { PROJECTOR_TYPE_PIXTRAL,   "pixtral"},
@@ -161,6 +179,9 @@ static std::map<projector_type, std::string> PROJECTOR_TYPE_NAMES = {
     { PROJECTOR_TYPE_VOXTRAL,   "voxtral"},
     { PROJECTOR_TYPE_LFM2,      "lfm2"},
     { PROJECTOR_TYPE_KIMIVL,    "kimivl"},
+    { PROJECTOR_TYPE_LIGHTONOCR,"lightonocr"},
+    { PROJECTOR_TYPE_COGVLM,    "cogvlm"},
+    { PROJECTOR_TYPE_JANUS_PRO, "janus_pro"},
 };
 
 static projector_type clip_projector_type_from_string(const std::string & str) {
@@ -203,7 +224,6 @@ static void clip_log_callback_default(enum ggml_log_level level, const char * te
 }
 
 struct clip_logger_state {
-    ggml_log_level verbosity_thold;
     ggml_log_callback log_callback;
     void * log_callback_user_data;
 };
@@ -237,17 +257,11 @@ static void clip_log_internal(enum ggml_log_level level, const char * format, ..
     va_end(args);
 }
 
-#define LOG_TMPL(level, ...) \
-    do { \
-        if ((level) >= g_logger_state.verbosity_thold) { \
-            clip_log_internal((level), __VA_ARGS__); \
-        } \
-    } while (0)
-#define LOG_INF(...) LOG_TMPL(GGML_LOG_LEVEL_INFO,  __VA_ARGS__)
-#define LOG_WRN(...) LOG_TMPL(GGML_LOG_LEVEL_WARN,  __VA_ARGS__)
-#define LOG_ERR(...) LOG_TMPL(GGML_LOG_LEVEL_ERROR, __VA_ARGS__)
-#define LOG_DBG(...) LOG_TMPL(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
-#define LOG_CNT(...) LOG_TMPL(GGML_LOG_LEVEL_CONT,  __VA_ARGS__)
+#define LOG_INF(...) clip_log_internal(GGML_LOG_LEVEL_INFO,  __VA_ARGS__)
+#define LOG_WRN(...) clip_log_internal(GGML_LOG_LEVEL_WARN,  __VA_ARGS__)
+#define LOG_ERR(...) clip_log_internal(GGML_LOG_LEVEL_ERROR, __VA_ARGS__)
+#define LOG_DBG(...) clip_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
+#define LOG_CNT(...) clip_log_internal(GGML_LOG_LEVEL_CONT,  __VA_ARGS__)
 
 //
 // cpp wrappers

llama/llama.cpp/tools/mtmd/clip.h

@@ -1,6 +1,7 @@
 #pragma once
 
 #include "ggml.h"
+
 #include <stddef.h>
 #include <stdint.h>
 
@@ -22,9 +23,17 @@ enum clip_modality {
     CLIP_MODALITY_AUDIO,
 };
 
+enum clip_flash_attn_type {
+    CLIP_FLASH_ATTN_TYPE_AUTO     = -1,
+    CLIP_FLASH_ATTN_TYPE_DISABLED = 0,
+    CLIP_FLASH_ATTN_TYPE_ENABLED  = 1,
+};
+
 struct clip_context_params {
     bool use_gpu;
-    enum ggml_log_level verbosity;
+    enum clip_flash_attn_type flash_attn_type;
+    int image_min_tokens;
+    int image_max_tokens;
 };
 
 struct clip_init_result {

llama/llama.cpp/tools/mtmd/mtmd-helper.cpp

@@ -32,8 +32,65 @@
 #define STB_IMAGE_IMPLEMENTATION
 #include "stb/stb_image.h"
 
-#define LOG_INF(...) fprintf(stdout, __VA_ARGS__)
-#define LOG_ERR(...) fprintf(stderr, __VA_ARGS__)
+//
+// internal logging functions
+//
+
+struct mtmd_helper_logger {
+    ggml_log_callback default_callback = [](ggml_log_level level, const char * text, void * user_data) {
+        (void) level;
+        (void) user_data;
+        fputs(text, stderr);
+        fflush(stderr);
+    };
+
+    ggml_log_callback log_callback = default_callback;
+    void * log_callback_user_data;
+
+    void log_v(enum ggml_log_level level, const char * format, va_list args) {
+        if (format == NULL) {
+            return;
+        }
+        va_list args_copy;
+        va_copy(args_copy, args);
+        char buffer[128];
+        int len = vsnprintf(buffer, 128, format, args);
+        if (len < 128) {
+            log_callback(level, buffer, log_callback_user_data);
+        } else {
+            char * buffer2 = (char *) calloc(len + 1, sizeof(char));
+            vsnprintf(buffer2, len + 1, format, args_copy);
+            buffer2[len] = 0;
+            log_callback(level, buffer2, log_callback_user_data);
+            free(buffer2);
+        }
+        va_end(args_copy);
+    }
+
+    void log(enum ggml_log_level level, const char * format, ...) {
+        va_list args;
+        va_start(args, format);
+        log_v(level, format, args);
+        va_end(args);
+    }
+} g_logger;
+
+#define LOG_INF(...) g_logger.log(GGML_LOG_LEVEL_INFO,  __VA_ARGS__)
+#define LOG_WRN(...) g_logger.log(GGML_LOG_LEVEL_WARN,  __VA_ARGS__)
+#define LOG_ERR(...) g_logger.log(GGML_LOG_LEVEL_ERROR, __VA_ARGS__)
+
+void mtmd_helper_log_set(ggml_log_callback log_callback, void * user_data) {
+    if (log_callback == nullptr) {
+        log_callback = g_logger.default_callback;
+    }
+    g_logger.log_callback = log_callback;
+    g_logger.log_callback_user_data = user_data;
+    mtmd_log_set(log_callback, user_data);
+}
+
+//
+// helper functions
+//
 
 size_t mtmd_helper_get_n_tokens(const mtmd_input_chunks * chunks) {
     size_t n_tokens = 0;
@@ -182,7 +239,7 @@ int32_t mtmd_helper_decode_image_chunk(
     }
 
     const llama_model * model = llama_get_model(lctx);
-    int n_mmproj_embd = llama_model_n_embd(model);
+    int n_mmproj_embd = llama_model_n_embd_inp(model);
     int n_pos_per_embd = mtmd_decode_use_mrope(ctx) ? 4 : 1;
 
     int32_t n_tokens = mtmd_input_chunk_get_n_tokens(chunk);
@@ -325,7 +382,7 @@ int32_t mtmd_helper_eval_chunks(mtmd_context * ctx,
                                 llama_pos * new_n_past) {
     size_t n_chunks = mtmd_input_chunks_size(chunks);
     if (n_chunks == 0) {
-        LOG_ERR("no chunks to eval\n");
+        LOG_WRN("no chunks to eval\n");
         return 0;
     }
 

llama/llama.cpp/tools/mtmd/mtmd-helper.h

@@ -20,6 +20,11 @@ extern "C" {
 // BREAKING CHANGES are expected.
 //
 
+// Set callback for all future logging events.
+// If this is not called, or NULL is supplied, everything is output on stderr.
+// Note: this also call mtmd_log_set() internally
+MTMD_API void mtmd_helper_log_set(ggml_log_callback log_callback, void * user_data);
+
 // helper function to construct a mtmd_bitmap from a file
 // it calls mtmd_helper_bitmap_init_from_buf() internally
 // returns nullptr on failure

llama/llama.cpp/tools/mtmd/mtmd.cpp

@@ -5,12 +5,20 @@
 
 #include "llama.h"
 
+// fix problem with std::min and std::max
+#if defined(_WIN32)
+#define WIN32_LEAN_AND_MEAN
+#ifndef NOMINMAX
+#   define NOMINMAX
+#endif
+#include <windows.h>
+#endif
+
 #include <algorithm>
 #include <cerrno>
 #include <cstdio>
 #include <cstdlib>
 #include <cstring>
-#include <limits>
 #include <vector>
 
 // represents raw image data, layout is RGBRGBRGB...
@@ -93,14 +101,26 @@ const char * mtmd_default_marker() {
     return "<__media__>";
 }
 
+static clip_flash_attn_type mtmd_get_clip_flash_attn_type(enum llama_flash_attn_type flash_attn_type) {
+    switch (flash_attn_type) {
+        case LLAMA_FLASH_ATTN_TYPE_AUTO:     return CLIP_FLASH_ATTN_TYPE_AUTO;
+        case LLAMA_FLASH_ATTN_TYPE_DISABLED: return CLIP_FLASH_ATTN_TYPE_DISABLED;
+        case LLAMA_FLASH_ATTN_TYPE_ENABLED:  return CLIP_FLASH_ATTN_TYPE_ENABLED;
+    }
+    return CLIP_FLASH_ATTN_TYPE_AUTO;
+}
+
 mtmd_context_params mtmd_context_params_default() {
-    mtmd_context_params params;
-    params.use_gpu = true;
-    params.print_timings = true;
-    params.n_threads = 4;
-    params.verbosity = GGML_LOG_LEVEL_INFO;
-    params.image_marker = MTMD_DEFAULT_IMAGE_MARKER;
-    params.media_marker = mtmd_default_marker();
+    mtmd_context_params params {
+        /* use_gpu           */ true,
+        /* print_timings     */ true,
+        /* n_threads         */ 4,
+        /* image_marker      */ MTMD_DEFAULT_IMAGE_MARKER,
+        /* media_marker      */ mtmd_default_marker(),
+        /* flash_attn_type   */ LLAMA_FLASH_ATTN_TYPE_AUTO,
+        /* image_min_tokens  */ -1,
+        /* image_max_tokens  */ -1,
+    };
     return params;
 }
 
@@ -152,7 +172,7 @@ struct mtmd_context {
         print_timings(ctx_params.print_timings),
         n_threads    (ctx_params.n_threads),
         media_marker (ctx_params.media_marker),
-        n_embd_text  (llama_model_n_embd(text_model))
+        n_embd_text  (llama_model_n_embd_inp(text_model))
     {
         if (std::string(ctx_params.image_marker) != MTMD_DEFAULT_IMAGE_MARKER) {
             throw std::runtime_error("custom image_marker is not supported anymore, use media_marker instead");
@@ -162,9 +182,13 @@ struct mtmd_context {
             throw std::runtime_error("media_marker must not be empty");
         }
 
-        clip_context_params ctx_clip_params;
-        ctx_clip_params.use_gpu   = ctx_params.use_gpu;
-        ctx_clip_params.verbosity = ctx_params.verbosity;
+        clip_context_params ctx_clip_params {
+            /* use_gpu           */ ctx_params.use_gpu,
+            /* flash_attn_type   */ CLIP_FLASH_ATTN_TYPE_AUTO,
+            /* image_min_tokens  */ ctx_params.image_min_tokens,
+            /* image_max_tokens  */ ctx_params.image_max_tokens,
+        };
+
         auto res = clip_init(mmproj_fname, ctx_clip_params);
         ctx_v = res.ctx_v;
         ctx_a = res.ctx_a;
@@ -268,7 +292,7 @@ struct mtmd_context {
             // https://github.com/huggingface/transformers/blob/1cd110c6cb6a6237614130c470e9a902dbc1a4bd/docs/source/en/model_doc/pixtral.md
             img_end = "[IMG_END]";
 
-        } else if (proj == PROJECTOR_TYPE_QWEN2VL || proj == PROJECTOR_TYPE_QWEN25VL) {
+        } else if (proj == PROJECTOR_TYPE_QWEN2VL || proj == PROJECTOR_TYPE_QWEN25VL || proj == PROJECTOR_TYPE_QWEN3VL) {
             // <|vision_start|> ... (image embeddings) ... <|vision_end|>
             img_beg = "<|vision_start|>";
             img_end = "<|vision_end|>";
@@ -285,6 +309,11 @@ struct mtmd_context {
             img_beg = "<img>";
             img_end = "</img>";
 
+        } else if (proj == PROJECTOR_TYPE_LIGHTONOCR) {
+            // <|im_start|> ... (image embeddings) ... <|im_end|>
+            img_beg = "<|im_start|>";
+            img_end = "<|im_end|>";
+
         }
     }
 
@@ -374,9 +403,7 @@ mtmd_context * mtmd_init_from_file(const char * mmproj_fname,
 }
 
 void mtmd_free(mtmd_context * ctx) {
-    if (ctx) {
-        delete ctx;
-    }
+    delete ctx;
 }
 
 struct mtmd_tokenizer {
@@ -1036,7 +1063,9 @@ const char * mtmd_image_tokens_get_id(const mtmd_image_tokens * image_tokens) {
 
 llama_pos mtmd_image_tokens_get_n_pos(const mtmd_image_tokens * image_tokens) {
     if (image_tokens->use_mrope_pos) {
-        return 1; // for M-RoPE, the whole image is 1 in temporal dimension
+        // for M-RoPE, temporal dimension = max(t,h,w)
+        // t is omitted as we don't support video input
+        return std::max(image_tokens->nx, image_tokens->ny);
     }
     return image_tokens->n_tokens();
 }
@@ -1075,3 +1104,8 @@ mtmd_input_chunks * mtmd_test_create_input_chunks() {
 
     return chunks;
 }
+
+void mtmd_log_set(ggml_log_callback log_callback, void * user_data) {
+    g_logger_state.log_callback = log_callback ? log_callback : clip_log_callback_default;
+    g_logger_state.log_callback_user_data = user_data;
+}

llama/llama.cpp/tools/mtmd/mtmd.h

@@ -82,9 +82,13 @@ struct mtmd_context_params {
     bool use_gpu;
     bool print_timings;
     int n_threads;
-    enum ggml_log_level verbosity;
     const char * image_marker; // deprecated, use media_marker instead
     const char * media_marker;
+    enum llama_flash_attn_type flash_attn_type;
+
+    // limit number of image tokens, only for vision models with dynamic resolution
+    int image_min_tokens; // minimum number of tokens for image input (default: read from metadata)
+    int image_max_tokens; // maximum number of tokens for image input (default: read from metadata)
 };
 
 MTMD_API const char * mtmd_default_marker(void);
@@ -156,7 +160,7 @@ MTMD_API const mtmd_image_tokens *  mtmd_input_chunk_get_tokens_image(const mtmd
 MTMD_API size_t                     mtmd_input_chunk_get_n_tokens    (const mtmd_input_chunk * chunk);
 // returns nullptr for ID on text chunk
 MTMD_API const char *               mtmd_input_chunk_get_id          (const mtmd_input_chunk * chunk);
-// number of temporal positions (always 1 for M-RoPE, n_tokens otherwise)
+// number of temporal positions (equals to max(t,h,w) for M-RoPE; equals to n_tokens otherwise)
 MTMD_API llama_pos                  mtmd_input_chunk_get_n_pos       (const mtmd_input_chunk * chunk);
 
 // in case you want to use custom logic to handle the chunk (i.e. KV cache management)
@@ -174,7 +178,7 @@ MTMD_API size_t       mtmd_image_tokens_get_n_tokens(const mtmd_image_tokens * i
 MTMD_API size_t       mtmd_image_tokens_get_nx      (const mtmd_image_tokens * image_tokens);
 MTMD_API size_t       mtmd_image_tokens_get_ny      (const mtmd_image_tokens * image_tokens);
 MTMD_API const char * mtmd_image_tokens_get_id      (const mtmd_image_tokens * image_tokens); // TODO: deprecate
-// number of temporal positions (always 1 for M-RoPE, n_tokens otherwise)
+// number of temporal positions (equals to max(t,h,w) for M-RoPE; equals to n_tokens otherwise)
 MTMD_API llama_pos    mtmd_image_tokens_get_n_pos   (const mtmd_image_tokens * image_tokens); // TODO: deprecate
 
 // tokenize an input text prompt and a list of bitmaps (images/audio)
@@ -213,6 +217,10 @@ MTMD_API int32_t mtmd_encode_chunk(mtmd_context * ctx,
 // llama_model_n_embd(model) * mtmd_input_chunk_get_n_tokens(chunk) * sizeof(float)
 MTMD_API float * mtmd_get_output_embd(mtmd_context * ctx);
 
+// Set callback for all future logging events.
+// If this is not called, or NULL is supplied, everything is output on stderr.
+MTMD_API void mtmd_log_set(ggml_log_callback log_callback, void * user_data);
+
 /////////////////////////////////////////
 
 // test function, to be used in test-mtmd-c-api.c