Refactor: fix controlnet and tae

Author: stduhpf

control.hpp

@@ -317,12 +317,10 @@ struct ControlNet : public GGMLRunner {
     bool guided_hint_cached         = false;
 
     ControlNet(ggml_backend_t backend,
-               SDVersion version = VERSION_SD1)
+               std::map<std::string, enum ggml_type>& tensor_types,
+               SDVersion version        = VERSION_SD1)
         : GGMLRunner(backend), control_net(version) {
-    }
-
-    void init_params(std::map<std::string, enum ggml_type>& tensor_types, const std::string prefix) {
-        control_net.init(params_ctx, tensor_types, prefix);
+        control_net.init(params_ctx, tensor_types, "");
     }
 
     ~ControlNet() {

stable-diffusion.cpp

@@ -358,7 +358,7 @@ class StableDiffusionGGML {
                 first_stage_model->alloc_params_buffer();
                 first_stage_model->get_param_tensors(tensors, "first_stage_model");
             } else {
-                tae_first_stage = std::make_shared<TinyAutoEncoder>(backend, vae_decode_only);
+                tae_first_stage = std::make_shared<TinyAutoEncoder>(backend, model_loader.tensor_storages_types, "decoder.layers", vae_decode_only);
             }
             // first_stage_model->get_param_tensors(tensors, "first_stage_model.");
 
@@ -370,7 +370,7 @@ class StableDiffusionGGML {
                 } else {
                     controlnet_backend = backend;
                 }
-                control_net = std::make_shared<ControlNet>(controlnet_backend, version);
+                control_net = std::make_shared<ControlNet>(controlnet_backend, model_loader.tensor_storages_types, version);
             }
 
             if (id_embeddings_path.find("v2") != std::string::npos) {

tae.hpp

@@ -188,14 +188,15 @@ struct TinyAutoEncoder : public GGMLRunner {
     bool decode_only = false;
 
     TinyAutoEncoder(ggml_backend_t backend,
+                    std::map<std::string, enum ggml_type>& tensor_types,
+                    const std::string prefix,
                     bool decoder_only = true)
         : decode_only(decoder_only),
           taesd(decode_only),
           GGMLRunner(backend) {
-    }
-    void init_params(std::map<std::string, enum ggml_type>& tensor_types, const std::string prefix) {
         taesd.init(params_ctx, tensor_types, prefix);
     }
+    
     std::string get_desc() {
         return "taesd";
     }