tckm.yan/aha
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

from gguf build chat_template and tokenizer

Browse Source
This commit is contained in:
jhqxxx
2026-03-11 17:30:28 +08:00
parent 86141322bb
commit 4d544ab68f
14 changed files with 374 additions and 118 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/chat_template/mod.rs
+55 -72
View File
@@ -2,7 +2,35 @@ use aha_openai_dive::v1::resources::chat::ChatCompletionParameters;
use anyhow::{Result, anyhow};
use minijinja::{Environment, Value as MiniJinjaValue, context};
use crate::utils::string_to_static_str;
use crate::utils::{extract_metadata_value, string_to_static_str};
pub fn fix_template(chat_template: &str) -> String {
chat_template
.replace(
"content.startswith('<tool_response>')",
"content is startingwith('<tool_response>')", // 使用minijinja中的 is startingwith 替换
)
.replace(
"content.endswith('</tool_response>')",
"content is endingwith('</tool_response>')", // 使用minijinja中的 is endingwith 替换
)
.replace(
"content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n')",
"((content | split('</think>'))[0] | rstrip('\\n') | split('<think>'))[-1] | lstrip('\\n')", // 使用自定义的split, rstrip, lstrip过滤器替换
)
.replace(
"content.split('</think>')[-1].lstrip('\\n')",
"(content | split('</think>'))[-1] | lstrip('\\n')", // 使用自定义的过滤器替换
)
.replace(
"reasoning_content.strip('\\n')",
"reasoning_content | strip('\\n')", // 使用自定义的过滤器替换
)
.replace(
"content.lstrip('\\n')",
"content | lstrip('\\n')", // 使用自定义的过滤器替换
)
}
pub fn get_template(path: String) -> Result<String> {
let tokenizer_config_file = path.clone() + "/tokenizer_config.json";
@@ -47,31 +75,7 @@ pub fn get_template(path: String) -> Result<String> {
};
let chat_template = chat_template.ok_or(anyhow!(format!("chat_template is none")))?;
// 修复模板中的问题行
let fixed_template = chat_template
.replace(
"content.startswith('<tool_response>')",
"content is startingwith('<tool_response>')", // 使用minijinja中的 is startingwith 替换
)
.replace(
"content.endswith('</tool_response>')",
"content is endingwith('</tool_response>')", // 使用minijinja中的 is endingwith 替换
)
.replace(
"content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n')",
"((content | split('</think>'))[0] | rstrip('\\n') | split('<think>'))[-1] | lstrip('\\n')", // 使用自定义的split, rstrip, lstrip过滤器替换
)
.replace(
"content.split('</think>')[-1].lstrip('\\n')",
"(content | split('</think>'))[-1] | lstrip('\\n')", // 使用自定义的过滤器替换
)
.replace(
"reasoning_content.strip('\\n')",
"reasoning_content | strip('\\n')", // 使用自定义的过滤器替换
)
.replace(
"content.lstrip('\\n')",
"content | lstrip('\\n')", // 使用自定义的过滤器替换
);
let fixed_template = fix_template(&chat_template);
Ok(fixed_template)
}
@@ -80,29 +84,7 @@ pub struct ChatTemplate<'a> {
}
impl<'a> ChatTemplate<'a> {
pub fn init(path: &str) -> Result<Self> {
let path: String = path.to_string();
if !std::path::Path::new(&path).exists() {
return Err(anyhow!("model path not found"));
}
let template = get_template(path.clone())?;
// let template = match get_template(path.clone()) {
// Ok(template) => template,
// Err(e) => {
// let jinja_path = path + "/chat_template.jinja";
// if !std::path::Path::new(&jinja_path).exists() {
// return Err(anyhow!(
// "get_template err {e} and chat_template.jinja not found"
// ));
// }
// std::fs::read_to_string(&jinja_path)
// .map_err(|e| anyhow!("Failed to read chat_template.jinja: {}", e))?
// }
// };
let template = string_to_static_str(template);
// 加载jinjaenv处理chat_template
let mut env = Environment::new();
// 添加自定义过滤器
fn setup_environment(env: &mut Environment<'a>) {
env.add_filter("tojson", |v: MiniJinjaValue| {
serde_json::to_string(&v).unwrap()
});
@@ -113,44 +95,44 @@ impl<'a> ChatTemplate<'a> {
.collect::<Vec<String>>()
});
// 添加 lstrip 过滤器
env.add_filter("lstrip", |s: String, chars: Option<String>| match chars {
Some(chars_str) => s.trim_start_matches(chars_str.as_str()).to_string(),
None => s.trim_start().to_string(),
});
// 添加 rstrip 过滤器
env.add_filter("rstrip", |s: String, chars: Option<String>| match chars {
Some(chars_str) => s.trim_end_matches(chars_str.as_str()).to_string(),
None => s.trim_end().to_string(),
});
// let template = get_template(path.to_string())?;
}
pub fn init(path: &str) -> Result<Self> {
let path: String = path.to_string();
if !std::path::Path::new(&path).exists() {
return Err(anyhow!("model path not found"));
}
let template = get_template(path.clone())?;
let template = string_to_static_str(template);
// 加载jinjaenv处理chat_template
let mut env = Environment::new();
Self::setup_environment(&mut env);
let _ = env.add_template("chat", template);
Ok(Self { env })
}
pub fn str_init(chat_template: &str) -> Result<Self> {
let fixed_template = fix_template(chat_template);
let template = string_to_static_str(fixed_template);
// 加载jinjaenv处理chat_template
let mut env = Environment::new();
Self::setup_environment(&mut env);
let _ = env.add_template("chat", template);
Ok(Self { env })
}
pub fn apply_chat_template(&self, messages: &ChatCompletionParameters) -> Result<String> {
let context = context! {
messages => &messages.messages,
tools => &messages.tools.as_ref(),
add_generation_prompt => true,
};
let template = self
.env
.get_template("chat")
.map_err(|e| anyhow!(format!("render template error {}", e)))?;
let message_str = template
.render(context)
.map_err(|e| anyhow!(format!("render template error {}", e)))?;
Ok(message_str)
}
pub fn apply_chat_temp_think(
&self,
messages: &ChatCompletionParameters,
enable_thinking: Option<bool>,
) -> Result<String> {
let enable_thinking = extract_metadata_value::<bool>(&messages.metadata, "enable_thinking");
let context = context! {
messages => &messages.messages,
tools => &messages.tools.as_ref(),
@@ -166,4 +148,5 @@ impl<'a> ChatTemplate<'a> {
.map_err(|e| anyhow!(format!("render template error {}", e)))?;
Ok(message_str)
}
}