[2025.06.19] pretoken corpus and add cold start
This commit is contained in:
@@ -4,12 +4,15 @@ LLM-Dolly is a custom LLM built from scratch
|
||||
|
||||
🗓️ 更新日志
|
||||
|
||||
#### 2025-06-05
|
||||
- 📝 上传tokenizer的训练权重(1.56字符/token),调整tokenizer的格式和训练方式,和qwen2的tokenizer风格保持一致.
|
||||
#### 2025-06-19
|
||||
- 📝 pretoken corpus and update cold start code.
|
||||
|
||||
<details>
|
||||
<summary>查看更多</summary>
|
||||
|
||||
#### 2025-06-05
|
||||
- 📝 上传tokenizer的训练权重(1.56字符/token),调整tokenizer的格式和训练方式,和qwen2的tokenizer风格保持一致.
|
||||
|
||||
#### 2025-05-26
|
||||
- 📝 增加LLM文档,涵盖LLM全周期技术点。
|
||||
|
||||
|
||||
Reference in New Issue
Block a user