chore(代码变更): 1. 清理无用代码 2. 更新依赖库 3. 优化项目结构 4. 修复小型bug 5. 增加注释以提高可读性

crates/minicpm-inference/src/loader/mod.rs

@@ -1,5 +1,3 @@
-pub mod q8;
-
 use burn::module::Module;
 use burn::record::{FullPrecisionSettings, NamedMpkFileRecorder};
 use burn::tensor::backend::Backend;
@@ -7,8 +5,6 @@ use burn::tensor::backend::Backend;
 use minicpm_core::config::LlamaConfig;
 use minicpm_core::model::LlamaForCausalLM;
 
-use self::q8::load_model_q8;
-
 pub fn load_model<B: Backend>(
     model_path: &str,
     config: &LlamaConfig,
@@ -19,11 +15,3 @@ pub fn load_model<B: Backend>(
     let model = model.load_file(model_path, &recorder, device)?;
     Ok(model)
 }
-
-pub fn load_model_q8_from_dir<B: Backend>(
-    model_dir: &str,
-    config: &LlamaConfig,
-    device: &B::Device,
-) -> anyhow::Result<LlamaForCausalLM<B>> {
-    load_model_q8(model_dir, config, device)
-}