add lora in minicpmv-cli

OpenBMB · Sep 29, 2024 · f21723a · f21723a
1 parent 936faf8
commit f21723a
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 0 deletions.
diff --git a/convert_lora_to_gguf.py b/convert_lora_to_gguf.py
@@ -225,6 +225,9 @@ def get_base_tensor_name(lora_tensor_name: str) -> str:
     base_name = lora_tensor_name.replace("base_model.model.", "")
     base_name = base_name.replace(".lora_A.weight", ".weight")
     base_name = base_name.replace(".lora_B.weight", ".weight")
+    # base_name = lora_tensor_name.replace("base_model.model.llm.", "")
+    # base_name = base_name.replace(".lora_A.default.weight", ".weight")
+    # base_name = base_name.replace(".lora_B.default.weight", ".weight")
     return base_name
 
 
@@ -340,6 +343,8 @@ def get_tensors(self) -> Iterator[tuple[str, Tensor]]:
                     base_name = get_base_tensor_name(name)
                     is_lora_a = ".lora_A.weight" in name
                     is_lora_b = ".lora_B.weight" in name
+                    # is_lora_a = ".lora_A.default.weight" in name
+                    # is_lora_b = ".lora_B.default.weight" in name
                     if not is_lora_a and not is_lora_b:
                         if ".base_layer.weight" in name:
                             continue

diff --git a/examples/llava/minicpmv-cli.cpp b/examples/llava/minicpmv-cli.cpp
@@ -211,6 +211,25 @@ static struct llava_context * llava_init_context(gpt_params * params) {
         return NULL;
     }
 
+    // // load and optionally apply lora adapters
+    // for (auto & la : params->lora_adapters) {
+    //     llama_lora_adapter_container loaded_la;
+    //     loaded_la.path = la.path;
+    //     loaded_la.scale = la.scale;
+    //     loaded_la.adapter = llama_lora_adapter_init(model, la.path.c_str());
+    //     if (loaded_la.adapter == nullptr) {
+    //         fprintf(stderr, "%s: error: failed to apply lora adapter '%s'\n", __func__, la.path.c_str());
+    //         // llama_free(lctx);
+    //         // llama_free_model(model);
+    //         // return iparams;
+    //         return NULL;
+    //     }
+    //     iparams.lora_adapters.push_back(loaded_la); // copy to list of loaded adapters
+    // }
+    // if (!params->lora_init_without_apply) {
+    //     llama_lora_adapters_apply(ctx_llama, iparams.lora_adapters);
+    // }
+
     auto ctx_llava = (struct llava_context *)malloc(sizeof(llava_context));
 
     ctx_llava->ctx_llama = ctx_llama;