refactor generate code

This commit is contained in:
jhqxxx
2026-04-02 22:22:52 +08:00
parent b254d21efc
commit bd8eee6520
59 changed files with 1745 additions and 1800 deletions
+1 -7
View File
@@ -39,17 +39,11 @@ fn glm_ocr_generate() -> Result<()> {
let mut model = GlmOcrGenerateModel::init(&model_path, None, None)?;
let i_duration = i_start.elapsed();
println!("Time elapsed in load model is: {:?}", i_duration);
let i_start = Instant::now();
let res = model.generate(mes)?;
let i_duration = i_start.elapsed();
println!("generate: \n {:?}", res);
if let Some(usage) = &res.usage {
let num_token = usage.total_tokens;
let duration_secs = i_duration.as_secs_f64();
let tps = num_token as f64 / duration_secs;
println!("Tokens per second (TPS): {:.2}", tps);
println!("usage: \n {:?}", usage);
}
println!("Time elapsed in generate is: {:?}", i_duration);
Ok(())
}