visual_model warmup (technically) works

Author: sfallah

tools/mtmd/clip.cpp

@@ -5412,6 +5412,7 @@ bool clip_image_batch_encode(clip_ctx * ctx, const int n_threads, const clip_ima
         case PROJECTOR_TYPE_VOXTRAL:
         case PROJECTOR_TYPE_JANUS_PRO:
         case PROJECTOR_TYPE_COGVLM:
+        case PROJECTOR_TYPE_DEEPSEEKOCR:
             {
                 // do nothing
             } break;
@@ -5554,6 +5555,10 @@ bool clip_is_gemma3(const struct clip_ctx * ctx) {
     return ctx->proj_type() == PROJECTOR_TYPE_GEMMA3;
 }
 
+bool clip_is_deepseekocr(const struct clip_ctx * ctx) {
+    return ctx->proj_type() == PROJECTOR_TYPE_DEEPSEEKOCR;
+}
+
 bool clip_has_vision_encoder(const struct clip_ctx * ctx) {
     return ctx->model.modality == CLIP_MODALITY_VISION;
 }

tools/mtmd/clip.h

@@ -105,6 +105,8 @@ bool clip_is_glm(const struct clip_ctx * ctx);
 bool clip_is_qwen2vl(const struct clip_ctx * ctx);
 bool clip_is_llava(const struct clip_ctx * ctx);
 bool clip_is_gemma3(const struct clip_ctx * ctx);
+bool clip_is_deepseekocr(const struct clip_ctx * ctx);
+
 
 bool clip_encode_float_image (struct clip_ctx * ctx, int n_threads, float * img, int h, int w, float * vec);
 

tools/mtmd/mtmd.cpp

@@ -810,7 +810,8 @@ int32_t mtmd_encode(mtmd_context * ctx, const mtmd_image_tokens * image_tokens)
 
     if (clip_is_llava(ctx_clip)
         || clip_is_minicpmv(ctx_clip)
-        || clip_is_glm(ctx_clip)) {
+        || clip_is_glm(ctx_clip)
+        || clip_is_deepseekocr(ctx_clip)) {
         // TODO @ngxson : llava does not support batched encoding ; this should be fixed inside clip_image_batch_encode()
         const auto & entries = image_tokens->batch_f32.entries;
         for (size_t i = 0; i < entries.size(); i++) {