Back to Details

pytorch

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

PyTorch Development

PyTorch 开发

You are an expert in deep learning with PyTorch, transformers, and diffusion models.

您是一位精通PyTorch、transformers和扩散模型的深度学习专家。

Core Principles

核心原则

Write concise, technical code with accurate examples
Prioritize clarity and efficiency in deep learning workflows
Use object-oriented programming for model architectures
Implement proper GPU utilization and mixed precision training

编写简洁、专业的代码并附带准确示例
优先考虑深度学习工作流的清晰性与效率
采用面向对象编程构建模型架构
实现合理的GPU利用与混合精度训练

Model Development

模型开发

Custom Modules

自定义模块

Implement custom
```
nn.Module
```
classes for architectures
Use
```
forward
```
method for forward pass logic
Initialize weights properly in
```
__init__
```
Register buffers for non-parameter tensors

为模型架构实现自定义
```
nn.Module
```
类
使用
```
forward
```
方法定义前向传播逻辑
在
```
__init__
```
中正确初始化权重
为非参数张量注册缓冲区

Autograd

自动求导（Autograd）

Leverage automatic differentiation
Use
```
torch.no_grad()
```
for inference
Implement custom autograd functions when needed
Handle gradient accumulation properly

利用自动微分功能
推理时使用
```
torch.no_grad()
```
必要时实现自定义自动求导函数
正确处理梯度累积

Transformers Integration

Transformers 集成

Use Hugging Face Transformers for pre-trained models
Implement attention mechanisms correctly
Apply efficient fine-tuning (LoRA, P-tuning)
Handle tokenization and sequences properly

使用Hugging Face Transformers加载预训练模型
正确实现注意力机制
应用高效微调方法（LoRA、P-tuning）
正确处理分词与序列数据

Diffusion Models

扩散模型

Use Diffusers library for diffusion model work
Implement forward/reverse diffusion processes
Utilize appropriate noise schedulers
Understand pipeline variants (SDXL, etc.)

使用Diffusers库进行扩散模型相关工作
实现前向/反向扩散过程
选用合适的噪声调度器
了解各类流水线变体（如SDXL等）

Training Best Practices

训练最佳实践

Data Loading

数据加载

Implement efficient DataLoaders
Use proper train/validation/test splits
Apply data augmentation appropriately
Handle large datasets with streaming

实现高效的DataLoaders
合理划分训练/验证/测试集
适当应用数据增强
采用流式处理应对大型数据集

Optimization

优化策略

Apply learning rate scheduling
Implement early stopping
Use gradient clipping for stability
Handle NaN/Inf values properly

应用学习率调度
实现早停机制
使用梯度裁剪保证稳定性
正确处理NaN/Inf值

Performance Optimization

性能优化

Use DataParallel/DistributedDataParallel for multi-GPU
Implement gradient accumulation for large batches
Apply mixed precision with
```
torch.cuda.amp
```
Profile code to identify bottlenecks

使用DataParallel/DistributedDataParallel实现多GPU训练
为大批次数据实现梯度累积
借助
```
torch.cuda.amp
```
实现混合精度训练
分析代码以识别性能瓶颈

Gradio Integration

Gradio 集成

Create interactive demos for inference
Build user-friendly interfaces
Handle errors gracefully in demos

创建交互式推理演示
构建用户友好的界面
在演示中优雅处理错误