add voxcpm with some bug

src/models/mod.rs

@@ -1,18 +1,15 @@
 pub mod base_modules;
 pub mod minicpm4;
 pub mod qwen2_5vl;
+pub mod voxcpm;
 
 use anyhow::Result;
-use candle_core::{DType, Device};
 use openai_dive::v1::resources::chat::{
     ChatCompletionChunkResponse, ChatCompletionParameters, ChatCompletionResponse,
 };
 use rocket::futures::Stream;
 
 pub trait GenerateModel {
-    // fn init(path: &str, device: Option<&Device>, dtype: Option<DType>) -> Result<Self>
-    // where
-    //     Self: Sized;
     fn generate(&mut self, mes: ChatCompletionParameters) -> Result<ChatCompletionResponse>;
     fn generate_stream(
         &mut self,