Update README.md

README.md

@@ -11,13 +11,17 @@ library_name: transformers
 
 [![License](https://img.shields.io/badge/License-Apache-f5de53?&color=f5de53)](LICENSE)
 
-PCMind-2.1-Kaiyuan-2B is a cutting-edge, **fully open-source language model** trained using Ascend 910A clusters. 
+PCMind-2.1-Kaiyuan-2B is a cutting-edge, **fully open-source language model** (i.e., open dataset) trained on a Ascend 910A cluster. 
 With 1.4B non-embedding parameters and training on 2.2 trillion tokens, 
 it achieves performance competitive with current state-of-the-art fully open models and even rivals some leading open-weight models of similar scale.
 
-## Introduction
+<center>
+![Model Performance Comparison](model_performance_comparison.svg)
+</center>
+
+We will publish the datasets used to train Kaiyuan-2B soon.
 
-![](model_performance_comparison.png)
+## Introduction
 
 Our data preprocessing and pre-training pipeline is designed for enhanced training efficiency and model quality, 
 achieved through several key innovations:
@@ -43,7 +47,7 @@ achieved through several key innovations:
 The model architecture is similar to [Qwen/Qwen3-1.7B](https://huggingface.co/Qwen/Qwen3-1.7B),
 and can be easily loaded by libraries like `transformers`.
 
-Please use [`demo.py`](demo.py) as an example of use.
+Please use [`demo.py`](demo.py) as an example.
 
 *Note: This is a pretrained base model only and has not undergone fine-tuning,
 reinforcement learning (RL), or any other post-training procedures.