unify cargo version and add some ci rules
This commit is contained in:
+105
-1
@@ -1 +1,105 @@
|
||||
pub mod chat_template;
|
||||
use anyhow::{Result, anyhow};
|
||||
use minijinja::{Environment, Value as MiniJinjaValue, context};
|
||||
use openai_dive::v1::resources::chat::ChatCompletionParameters;
|
||||
|
||||
use crate::utils::string_to_static_str;
|
||||
|
||||
pub fn get_template(path: String) -> Result<String> {
|
||||
let tokenizer_config_file = path.clone() + "/tokenizer_config.json";
|
||||
assert!(
|
||||
std::path::Path::new(&tokenizer_config_file).exists(),
|
||||
"tokenizer_config.json not exists in model path"
|
||||
);
|
||||
let tokenizer_config: serde_json::Value =
|
||||
serde_json::from_slice(&std::fs::read(tokenizer_config_file)?)
|
||||
.map_err(|e| anyhow!(format!("load tokenizer_config file error:{}", e)))?;
|
||||
let chat_template = tokenizer_config["chat_template"]
|
||||
.as_str()
|
||||
.ok_or(anyhow!(format!("chat_template to str error")))?;
|
||||
// 修复模板中的问题行
|
||||
let fixed_template = chat_template
|
||||
.replace(
|
||||
"message.content.startswith('<tool_response>')",
|
||||
"message.content is startingwith('<tool_response>')", // 使用minijinja中的 is startingwith 替换
|
||||
)
|
||||
.replace(
|
||||
"message.content.endswith('</tool_response>')",
|
||||
"message.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
|
||||
);
|
||||
Ok(fixed_template)
|
||||
}
|
||||
|
||||
pub struct ChatTemplate<'a> {
|
||||
env: Environment<'a>,
|
||||
}
|
||||
|
||||
impl<'a> ChatTemplate<'a> {
|
||||
pub fn init(path: &str) -> Result<Self> {
|
||||
let path = path.to_string();
|
||||
assert!(
|
||||
std::path::Path::new(&path).exists(),
|
||||
"model path file not exists"
|
||||
);
|
||||
let template = get_template(path)?;
|
||||
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
|
||||
env.add_filter("tojson", |v: MiniJinjaValue| {
|
||||
serde_json::to_string(&v).unwrap()
|
||||
});
|
||||
|
||||
env.add_filter("split", |s: String, delimiter: String| {
|
||||
s.split(&delimiter)
|
||||
.map(|s| s.to_string())
|
||||
.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())?;
|
||||
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,
|
||||
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)
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user