Merge branch 'gguf' into gguf-convert

Author: ggerganov

examples/gguf/gguf.cpp

@@ -233,16 +233,13 @@ int main(int argc, char ** argv) {
     const std::string fname(argv[1]);
     const std::string mode (argv[2]);
 
-    GGML_ASSERT((mode == "r" || mode == "w" || mode == "q") && "mode must be r, w or q");
+    GGML_ASSERT((mode == "r" || mode == "w") && "mode must be r or w");
 
     if (mode == "w") {
         GGML_ASSERT(gguf_ex_write(fname) && "failed to write gguf file");
     } else if (mode == "r") {
         GGML_ASSERT(gguf_ex_read_0(fname) && "failed to read gguf file");
         GGML_ASSERT(gguf_ex_read_1(fname) && "failed to read gguf file");
-    } else if (mode == "q") {
-        llama_model_quantize_params params = llama_model_quantize_default_params();
-        llama_model_quantize(fname.c_str(), "quant.gguf", &params);
     }
 
     return 0;

llama.cpp

@@ -1022,6 +1022,7 @@ struct llama_model_loader {
     int n_kv      = 0;
     int n_tensors = 0;
     int n_created = 0;
+    size_t n_tot_elements = 0;
 
     bool use_mmap = false;
 
@@ -1046,6 +1047,12 @@ struct llama_model_loader {
 
         file_version = (enum llama_file_version) gguf_get_version(ctx_gguf);
 
+        for (int i = 0; i < n_tensors; i++) {
+            const char * name = gguf_get_tensor_name(ctx_gguf, i);
+            struct ggml_tensor * t = ggml_get_tensor(ctx_meta, name);
+            n_tot_elements += ggml_nelements(t);
+        }
+        
         // print meta data
         // TODO: make optional
         {
@@ -1416,7 +1423,8 @@ static void llama_model_load_internal(
         LLAMA_LOG_INFO("%s: n_ff        = %u\n",   __func__, hparams.n_ff);
         LLAMA_LOG_INFO("%s: freq_base   = %.1f\n", __func__, hparams.rope_freq_base);
         LLAMA_LOG_INFO("%s: freq_scale  = %g\n",   __func__, hparams.rope_freq_scale);
-        LLAMA_LOG_INFO("%s: model size  = %s\n",   __func__, llama_model_type_name(model.type));
+        LLAMA_LOG_INFO("%s: model type  = %s\n",   __func__, llama_model_type_name(model.type));
+        LLAMA_LOG_INFO("%s: model size = %.2f B\n",   __func__, ml->n_tot_elements*1e-9);
 
         // TODO: print number of tensors for each quantization
     }