update the text

Author: ZDandsomSP

index.html

@@ -116,11 +116,11 @@ <h2>Analyse <span>Universal</span> Time Series <br> Through <span>Large Model</s
           </div>
           <div class="col-lg-6 content">
             <h2>Welcome to Timer - Universal Time Series Analysis</h2>
-            <h3>Timer is the brainchild of a dedicated team of researchers and developers from the <span>School of Software, at Tsinghua University.</span> Our work is inspired by the transformative impact of large language models on natural language processing. We've taken the same approach to time series data, creating a model <span>pre-trained on massive datasets and fine-tuning for specific tasks</span> with few-shot ability.</h3>
+            <h3>Timer originated from the <span>School of Software at Tsinghua University</span>, and was developed for the field of time series analysis by constructing large-scale time series datasets and pre training formats. By training on a <span>large dataset</span>, Timer exhibits <span>few sample generalization and multi task adaptation capabilities</span>, with considerable temporal analysis and data generation capabilities for real-world scenarios. It has the following characteristics:</h3>
             <ul>
-              <li><i class="bi bi-check-circle"></i> Knowledge base based on large-scale pre-training.</li>
-              <li><i class="bi bi-check-circle"></i> Diverse downstream tasks to adapt to complex industrial scenarios.</li>
-              <li><i class="bi bi-check-circle"></i> Analytical ability and generalization ability for real-world data.</li>
+              <li><i class="bi bi-check-circle"></i> Generalization: Achieving cutting-edge deep model prediction performance based on small sample fine-tuning.</li>
+              <li><i class="bi bi-check-circle"></i> Universality: Suitable for multiple tasks, supports variable input-output length.</li>
+              <li><i class="bi bi-check-circle"></i> Scalability: The model achieves improved performance as the number of parameters or pre training scale increases.</li>
             </ul>
           </div>
         </div>

index_zh.html

@@ -100,11 +100,11 @@ <h2>基于<span>大模型</span>的通用<span>时间序列分析</span></h2>
           </div>
           <div class="col-lg-6 content">
             <h2>欢迎使用Timer-深度时序分析通用基础模型</h2>
-            <h3>Timer是<span>清华大学软件学院</span>在时序分析领域深耕的结晶。我们的工作受到大型语言模型对自然语言处理变革性影响的启发，在时间序列领域使用大模型技术创建了该领域首个真正意义上的大模型，并且<span>在大量数据集上进行预训练，并依靠少样本能力对特定任务进行微调</span>，从而赋予深度模型通用能力。我们的模型拥有以下特点：</h3>
+            <h3>Timer 模型发源于<span>清华大学软件学院</span>，针对时序分析领域，构建了大规模时序数据集和预训练格式。通过<span>在大量数据集上进行训练</span>，Timer 展现出<span>少样本泛化以及多任务适配能力</span>，具备可观的时序分析能力和对真实场景的数据生成能力，拥有以下特点：</h3>
             <ul>
-              <li><i class="bi bi-check-circle"></i> 基于大规模预训练的知识库</li>
-              <li><i class="bi bi-check-circle"></i> 多样化的下游任务以适应复杂的行业场景</li>
-              <li><i class="bi bi-check-circle"></i> 可靠分析能力和对真实世界数据的理解能力</li>
+              <li><i class="bi bi-check-circle"></i> 泛化性：基于少样本微调取得领域前沿深度模型预测效果</li>
+              <li><i class="bi bi-check-circle"></i> 通用性：适配多种任务，支持可变输入输出长度</li>
+              <li><i class="bi bi-check-circle"></i> 可扩展性：模型随着参数量或预训练规模扩大取得效果提升</li>
             </ul>
           </div>
         </div>