update fmt

src/exec/mod.rs

@@ -8,6 +8,7 @@ pub mod fun_asr_nano;
 pub mod glm_asr_nano;
 pub mod glm_ocr;
 pub mod hunyuan_ocr;
+pub mod lfm2;
 pub mod minicpm4;
 pub mod paddleocr_vl;
 pub mod qwen2_5vl;
@@ -18,7 +19,6 @@ pub mod qwen3vl;
 pub mod rmbg2_0;
 pub mod voxcpm;
 pub mod voxcpm1_5;
-pub mod lfm2;
 
 use anyhow::Result;
 

src/models/lfm2/config.rs

@@ -1,5 +1,5 @@
+use anyhow::{Result, anyhow};
 use serde::{Deserialize, Serialize};
-use anyhow::{anyhow, Result};
 #[derive(Debug, PartialEq, Deserialize, Serialize)]
 pub struct Lfm2Config {
     pub architectures: Vec<String>,
@@ -43,7 +43,6 @@ pub struct Lfm2Config {
     pub tie_embedding: Option<bool>,
 }
 
-
 impl Lfm2Config {
     pub fn full_attn_idx2layer_type(&mut self) {
         if self.layer_types.is_none()
@@ -75,7 +74,9 @@ impl Lfm2Config {
             }
             Ok(layer_types)
         } else {
-            Err(anyhow!("layer_types full_attn_idxs cannot be none at the same time"))
+            Err(anyhow!(
+                "layer_types full_attn_idxs cannot be none at the same time"
+            ))
         }
     }
 }
@@ -84,5 +85,5 @@ impl Lfm2Config {
 pub struct Lfm2GenerateConfig {
     pub bos_token_id: u32,
     pub eos_token_id: u32,
-    pub pad_token_id: u32
+    pub pad_token_id: u32,
-}
+}

src/models/lfm2/generate.rs

@@ -1,15 +1,18 @@
+use crate::utils::build_completion_chunk_response;
 use crate::{
     chat_template::ChatTemplate,
-    models::{GenerateModel, lfm2::{
+    models::{
-        config::{Lfm2Config, Lfm2GenerateConfig},
+        GenerateModel,
-        model::Lfm2Model,
+        lfm2::{
-    }},
+            config::{Lfm2Config, Lfm2GenerateConfig},
+            model::Lfm2Model,
+        },
+    },
     tokenizer::TokenizerModel,
     utils::{
         build_completion_response, find_type_files, get_device, get_dtype, get_logit_processor,
     },
 };
-use crate::utils::build_completion_chunk_response;
 use aha_openai_dive::v1::resources::chat::{ChatCompletionParameters, ChatCompletionResponse};
 use anyhow::Result;
 use candle_core::{DType, Device, Tensor};
@@ -59,8 +62,6 @@ impl<'a> Lfm2GenerateModel<'a> {
             model_name,
         })
     }
-
-    
 }
 
 impl<'a> GenerateModel for Lfm2GenerateModel<'a> {
@@ -103,16 +104,20 @@ impl<'a> GenerateModel for Lfm2GenerateModel<'a> {
     }
 
     fn generate_stream(
-            &mut self,
+        &mut self,
-            mes: ChatCompletionParameters,
+        mes: ChatCompletionParameters,
-        ) -> Result<
+    ) -> Result<
-            Box<
+        Box<
-                dyn rocket::futures::Stream<Item = Result<aha_openai_dive::v1::resources::chat::ChatCompletionChunkResponse, anyhow::Error>>
+            dyn rocket::futures::Stream<
-                    + Send
+                    Item = Result<
-                    + Unpin
+                        aha_openai_dive::v1::resources::chat::ChatCompletionChunkResponse,
-                    + '_,
+                        anyhow::Error,
-            >,
+                    >,
-        > {
+                > + Send
+                + Unpin
+                + '_,
+        >,
+    > {
         let mes_render = self.chat_template.apply_chat_template(&mes)?;
         let mut logits = get_logit_processor(
             mes.temperature,
@@ -160,4 +165,4 @@ impl<'a> GenerateModel for Lfm2GenerateModel<'a> {
         };
         Ok(Box::new(Box::pin(stream)))
     }
-}
+}

src/models/lfm2/mod.rs

@@ -1,3 +1,3 @@
 pub mod config;
+pub mod generate;
 pub mod model;
-pub mod generate;

src/models/lfm2/model.rs

@@ -70,7 +70,7 @@ impl Lfm2ShortConv {
                 bx.narrow(D::Minus1, pad_num.unsigned_abs(), self.l_cache)?
             };
             self.cache = Some(conv_state);
-            let bx = bx.pad_with_zeros(D::Minus1, self.l_cache-1, self.l_cache-1)?;
+            let bx = bx.pad_with_zeros(D::Minus1, self.l_cache - 1, self.l_cache - 1)?;
             let bx = conv1d_depthwise(&bx, self.conv.weight(), self.conv.bias())?;
             bx.narrow(D::Minus1, 0, seq_len)?
         } else {
@@ -145,9 +145,8 @@ impl Lfm2DecoderLayer {
         let intermediate_size = if config.block_auto_adjust_ff_dim {
             let inter_size = 2 * config.block_ff_dim / 3;
             let inter_size = (config.block_ffn_dim_multiplier * inter_size as f64) as usize;
-            let inter_size = config.block_multiple_of
+            config.block_multiple_of
-                * ((inter_size + config.block_multiple_of - 1) / config.block_multiple_of);
+                * ((inter_size + config.block_multiple_of - 1) / config.block_multiple_of)
-            inter_size
         } else {
             config.block_ff_dim
         };
@@ -291,9 +290,7 @@ impl Lfm2Model {
             );
             match linear {
                 Ok(linear) => linear,
-                Err(_) => {
+                Err(_) => Linear::new(model.embed_tokens.embeddings().clone(), None),
-                    Linear::new(model.embed_tokens.embeddings().clone(), None)
-                }
             }
         };
         Ok(Self { model, lm_head })

src/models/mod.rs

@@ -7,6 +7,7 @@ pub mod fun_asr_nano;
 pub mod glm_asr_nano;
 pub mod glm_ocr;
 pub mod hunyuan_ocr;
+pub mod lfm2;
 pub mod mask_gct;
 pub mod minicpm4;
 pub mod paddleocr_vl;
@@ -19,7 +20,6 @@ pub mod qwen3vl;
 pub mod rmbg2_0;
 pub mod voxcpm;
 pub mod w2v_bert_2_0;
-pub mod lfm2;
 
 use aha_openai_dive::v1::resources::chat::{
     ChatCompletionChunkResponse, ChatCompletionParameters, ChatCompletionResponse,
@@ -28,7 +28,15 @@ use anyhow::{Result, anyhow};
 use rocket::futures::Stream;
 
 use crate::models::{
-    deepseek_ocr::generate::DeepseekOCRGenerateModel, fun_asr_nano::generate::FunAsrNanoGenerateModel, glm_asr_nano::generate::GlmAsrNanoGenerateModel, glm_ocr::generate::GlmOcrGenerateModel, hunyuan_ocr::generate::HunyuanOCRGenerateModel, lfm2::generate::Lfm2GenerateModel, minicpm4::generate::MiniCPMGenerateModel, paddleocr_vl::generate::PaddleOCRVLGenerateModel, qwen2_5vl::generate::Qwen2_5VLGenerateModel, qwen3::generate::Qwen3GenerateModel, qwen3_5::generate::Qwen3_5GenerateModel, qwen3_asr::generate::Qwen3AsrGenerateModel, qwen3vl::generate::Qwen3VLGenerateModel, rmbg2_0::generate::RMBG2_0Model, voxcpm::generate::VoxCPMGenerate
+    deepseek_ocr::generate::DeepseekOCRGenerateModel,
+    fun_asr_nano::generate::FunAsrNanoGenerateModel,
+    glm_asr_nano::generate::GlmAsrNanoGenerateModel, glm_ocr::generate::GlmOcrGenerateModel,
+    hunyuan_ocr::generate::HunyuanOCRGenerateModel, lfm2::generate::Lfm2GenerateModel,
+    minicpm4::generate::MiniCPMGenerateModel, paddleocr_vl::generate::PaddleOCRVLGenerateModel,
+    qwen2_5vl::generate::Qwen2_5VLGenerateModel, qwen3::generate::Qwen3GenerateModel,
+    qwen3_5::generate::Qwen3_5GenerateModel, qwen3_asr::generate::Qwen3AsrGenerateModel,
+    qwen3vl::generate::Qwen3VLGenerateModel, rmbg2_0::generate::RMBG2_0Model,
+    voxcpm::generate::VoxCPMGenerate,
 };
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq, clap::ValueEnum)]
@@ -130,7 +138,8 @@ impl WhichModel {
     pub fn model_type(self) -> &'static str {
         match self {
             // LLM models
-            WhichModel::MiniCPM4_0_5B | WhichModel::Qwen3_0_6B
+            WhichModel::MiniCPM4_0_5B
+            | WhichModel::Qwen3_0_6B
             | WhichModel::LFM2_1_2B
             | WhichModel::LFM2_5_1_2BInstruct => "llm",
             WhichModel::Qwen2_5vl3B

tests/config_tests.rs

@@ -1,5 +1,8 @@
 use aha::models::{
-    deepseek_ocr::config::DeepseekOCRConfig, hunyuan_ocr::config::HunYuanVLConfig, lfm2::config::Lfm2Config, minicpm4::config::MiniCPM4Config, paddleocr_vl::config::PaddleOCRVLConfig, qwen2_5vl::config::Qwen2_5VLConfig, qwen3vl::config::Qwen3VLConfig, voxcpm::config::VoxCPMConfig
+    deepseek_ocr::config::DeepseekOCRConfig, hunyuan_ocr::config::HunYuanVLConfig,
+    lfm2::config::Lfm2Config, minicpm4::config::MiniCPM4Config,
+    paddleocr_vl::config::PaddleOCRVLConfig, qwen2_5vl::config::Qwen2_5VLConfig,
+    qwen3vl::config::Qwen3VLConfig, voxcpm::config::VoxCPMConfig,
 };
 use anyhow::Result;
 

tests/messy_test.rs

@@ -43,7 +43,7 @@ fn messy_test() -> Result<()> {
     //     aha::utils::get_default_save_dir().ok_or(anyhow::anyhow!("Failed to get save dir"))?;
     // let model_path = format!("{}/deepseek-ai/DeepSeek-OCR-2/", save_dir);
     // let stem = std::path::Path::new(&model_path)
-    //     .file_name() 
+    //     .file_name()
     //     .and_then(|s| s.to_str())
     //     .unwrap_or("qwen3.5");
     // println!("stem: {:?}", stem);

tests/test_lfm2.rs

@@ -1,7 +1,10 @@
-use std::{pin::pin, time::Instant};
+use aha::{
+    chat::ChatCompletionParameters,
+    models::{GenerateModel, lfm2::generate::Lfm2GenerateModel},
+};
 use anyhow::Result;
-use aha::{chat::ChatCompletionParameters, models::{GenerateModel, lfm2::generate::Lfm2GenerateModel}};
 use rocket::futures::StreamExt;
+use std::{pin::pin, time::Instant};
 
 #[test]
 fn lfm2_generate() -> Result<()> {
@@ -44,7 +47,6 @@ fn lfm2_generate() -> Result<()> {
     Ok(())
 }
 
-
 #[tokio::test]
 async fn lfm2_stream() -> Result<()> {
     // test with cuda: RUST_BACKTRACE=1 cargo test -F cuda --test test_lfm2 lfm2_stream -r -- --nocapture
@@ -88,4 +90,4 @@ async fn lfm2_stream() -> Result<()> {
     println!("Time elapsed in generate is: {:?}", i_duration);
 
     Ok(())
-}
+}