https://Tin10g.github.io/categories/%E5%B7%A5%E4%BD%9C%E6%97%A5%E5%BF%97/ This is Ting's BLOG ... Hugo 0.137.0 & FixIt v0.3.13 en ting10win@gmail.com (Ting) ting10win@gmail.com (Ting) Fri, 17 Apr 2026 11:25:40 +0800 https://Tin10g.github.io/posts/todo-list/ Fri, 17 Apr 2026 11:25:40 +0800ting10win@gmail.com (Ting) https://Tin10g.github.io/posts/todo-list/ 工作日志 <h2 id="part-1-python进阶" class="heading-element"><span>Part 1 Python进阶</span> <a href="#part-1-python%e8%bf%9b%e9%98%b6" class="heading-mark"> <svg class="octicon octicon-link" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="m7.775 3.275 1.25-1.25a3.5 3.5 0 1 1 4.95 4.95l-2.5 2.5a3.5 3.5 0 0 1-4.95 0 .751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018 1.998 1.998 0 0 0 2.83 0l2.5-2.5a2.002 2.002 0 0 0-2.83-2.83l-1.25 1.25a.751.751 0 0 1-1.042-.018.751.751 0 0 1-.018-1.042Zm-4.69 9.64a1.998 1.998 0 0 0 2.83 0l1.25-1.25a.751.751 0 0 1 1.042.018.751.751 0 0 1 .018 1.042l-1.25 1.25a3.5 3.5 0 1 1-4.95-4.95l2.5-2.5a3.5 3.5 0 0 1 4.95 0 .751.751 0 0 1-.018 1.042.751.751 0 0 1-1.042.018 1.998 1.998 0 0 0-2.83 0l-2.5 2.5a1.998 1.998 0 0 0 0 2.83Z"></path></svg> </a> </h2><ul> <li>Numpy, Pandas, Matplotlib掌握</li> </ul> <h2 id="part-2-深度学习框架" class="heading-element"><span>Part 2 深度学习框架</span> <a href="#part-2-%e6%b7%b1%e5%ba%a6%e5%ad%a6%e4%b9%a0%e6%a1%86%e6%9e%b6" class="heading-mark"> <svg class="octicon octicon-link" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="m7.775 3.275 1.25-1.25a3.5 3.5 0 1 1 4.95 4.95l-2.5 2.5a3.5 3.5 0 0 1-4.95 0 .751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018 1.998 1.998 0 0 0 2.83 0l2.5-2.5a2.002 2.002 0 0 0-2.83-2.83l-1.25 1.25a.751.751 0 0 1-1.042-.018.751.751 0 0 1-.018-1.042Zm-4.69 9.64a1.998 1.998 0 0 0 2.83 0l1.25-1.25a.751.751 0 0 1 1.042.018.751.751 0 0 1 .018 1.042l-1.25 1.25a3.5 3.5 0 1 1-4.95-4.95l2.5-2.5a3.5 3.5 0 0 1 4.95 0 .751.751 0 0 1-.018 1.042.751.751 0 0 1-1.042.018 1.998 1.998 0 0 0-2.83 0l-2.5 2.5a1.998 1.998 0 0 0 0 2.83Z"></path></svg> </a> </h2><ul> <li>跑通几个标准训练循环（前向传播、计算损失、反向传播、优化器更新）</li> <li>理解张量（Tensor）操作、自动求导机制、Dataset/DataLoader构建</li> </ul> <h2 id="part-3-传统深度学习网络" class="heading-element"><span>Part 3 传统深度学习网络</span> <a href="#part-3-%e4%bc%a0%e7%bb%9f%e6%b7%b1%e5%ba%a6%e5%ad%a6%e4%b9%a0%e7%bd%91%e7%bb%9c" class="heading-mark"> <svg class="octicon octicon-link" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="m7.775 3.275 1.25-1.25a3.5 3.5 0 1 1 4.95 4.95l-2.5 2.5a3.5 3.5 0 0 1-4.95 0 .751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018 1.998 1.998 0 0 0 2.83 0l2.5-2.5a2.002 2.002 0 0 0-2.83-2.83l-1.25 1.25a.751.751 0 0 1-1.042-.018.751.751 0 0 1-.018-1.042Zm-4.69 9.64a1.998 1.998 0 0 0 2.83 0l1.25-1.25a.751.751 0 0 1 1.042.018.751.751 0 0 1 .018 1.042l-1.25 1.25a3.5 3.5 0 1 1-4.95-4.95l2.5-2.5a3.5 3.5 0 0 1 4.95 0 .751.751 0 0 1-.018 1.042.751.751 0 0 1-1.042.018 1.998 1.998 0 0 0-2.83 0l-2.5 2.5a1.998 1.998 0 0 0 0 2.83Z"></path></svg> </a> </h2><ul> <li>多层感知机（MLP）：理解前馈神经网络。</li> <li>循环神经网络（RNN/LSTM/GRU）：处理序列数据的鼻祖。</li> <li>卷积神经网络（CNN）：虽然主要用于图像，但在文本分类（如TextCNN）和图像文本检测中仍有应用。</li> </ul> <h2 id="part-4-词向量与特征表示" class="heading-element"><span>Part 4 词向量与特征表示</span> <a href="#part-4-%e8%af%8d%e5%90%91%e9%87%8f%e4%b8%8e%e7%89%b9%e5%be%81%e8%a1%a8%e7%a4%ba" class="heading-mark"> <svg class="octicon octicon-link" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="m7.775 3.275 1.25-1.25a3.5 3.5 0 1 1 4.95 4.95l-2.5 2.5a3.5 3.5 0 0 1-4.95 0 .751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018 1.998 1.998 0 0 0 2.83 0l2.5-2.5a2.002 2.002 0 0 0-2.83-2.83l-1.25 1.25a.751.751 0 0 1-1.042-.018.751.751 0 0 1-.018-1.042Zm-4.69 9.64a1.998 1.998 0 0 0 2.83 0l1.25-1.25a.751.751 0 0 1 1.042.018.751.751 0 0 1 .018 1.042l-1.25 1.25a3.5 3.5 0 1 1-4.95-4.95l2.5-2.5a3.5 3.5 0 0 1 4.95 0 .751.751 0 0 1-.018 1.042.751.751 0 0 1-1.042.018 1.998 1.998 0 0 0-2.83 0l-2.5 2.5a1.998 1.998 0 0 0 0 2.83Z"></path></svg> </a> </h2><ul> <li>了解 Word2Vec, GloVe 等静态词向量</li> <li>理解文本的分布式表示（Semantic Representation）</li> </ul> <h2 id="part-5-序列标注与文本分类" class="heading-element"><span>Part 5 序列标注与文本分类</span> <a href="#part-5-%e5%ba%8f%e5%88%97%e6%a0%87%e6%b3%a8%e4%b8%8e%e6%96%87%e6%9c%ac%e5%88%86%e7%b1%bb" class="heading-mark"> <svg class="octicon octicon-link" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="m7.775 3.275 1.25-1.25a3.5 3.5 0 1 1 4.95 4.95l-2.5 2.5a3.5 3.5 0 0 1-4.95 0 .751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018 1.998 1.998 0 0 0 2.83 0l2.5-2.5a2.002 2.002 0 0 0-2.83-2.83l-1.25 1.25a.751.751 0 0 1-1.042-.018.751.751 0 0 1-.018-1.042Zm-4.69 9.64a1.998 1.998 0 0 0 2.83 0l1.25-1.25a.751.751 0 0 1 1.042.018.751.751 0 0 1 .018 1.042l-1.25 1.25a3.5 3.5 0 1 1-4.95-4.95l2.5-2.5a3.5 3.5 0 0 1 4.95 0 .751.751 0 0 1-.018 1.042.751.751 0 0 1-1.042.018 1.998 1.998 0 0 0-2.83 0l-2.5 2.5a1.998 1.998 0 0 0 0 2.83Z"></path></svg> </a> </h2><ul> <li>序列标注（Sequence Labeling）： CRF（条件随机场）、BiLSTM-CRF模型。这是命名实体识别（NER）、分词的基础。</li> <li>文本分类（Text Classification）：掌握基础的分类模型，了解导师论文中提到的原型网络（Prototypical Networks）。</li> </ul> <h2 id="part-6-注意力机制与大模型" class="heading-element"><span>Part 6 注意力机制与大模型</span> <a href="#part-6-%e6%b3%a8%e6%84%8f%e5%8a%9b%e6%9c%ba%e5%88%b6%e4%b8%8e%e5%a4%a7%e6%a8%a1%e5%9e%8b" class="heading-mark"> <svg class="octicon octicon-link" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="m7.775 3.275 1.25-1.25a3.5 3.5 0 1 1 4.95 4.95l-2.5 2.5a3.5 3.5 0 0 1-4.95 0 .751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018 1.998 1.998 0 0 0 2.83 0l2.5-2.5a2.002 2.002 0 0 0-2.83-2.83l-1.25 1.25a.751.751 0 0 1-1.042-.018.751.751 0 0 1-.018-1.042Zm-4.69 9.64a1.998 1.998 0 0 0 2.83 0l1.25-1.25a.751.751 0 0 1 1.042.018.751.751 0 0 1 .018 1.042l-1.25 1.25a3.5 3.5 0 1 1-4.95-4.95l2.5-2.5a3.5 3.5 0 0 1 4.95 0 .751.751 0 0 1-.018 1.042.751.751 0 0 1-1.042.018 1.998 1.998 0 0 0-2.83 0l-2.5 2.5a1.998 1.998 0 0 0 0 2.83Z"></path></svg> </a> </h2><ul> <li>Transformer 架构：深入啃透 Self-Attention 机制，这是目前NLP的基石</li> <li>预训练语言模型（PLMs）：熟练使用 HuggingFace transformers 库，掌握 BERT, RoBERTa 的微调（Fine-tuning）</li> <li>提示学习（Prompt-learning）：近两年的绝对热点，需重点学习</li> </ul>