Candle - HuggingFace Rust AI 框架 - 小记

伊织code2024-03-24 13:27

文章目录

    • [关于 Candle](#关于 Candle)
      • 结构
      • 安装
        • [1、With Cuda support :](#1、With Cuda support :)
        • [2、Without Cuda support :](#2、Without Cuda support :)
        • [3、With mkl support](#3、With mkl support)
    • [二、基本使用 Hello world!](#二、基本使用 Hello world!)
      • [1、处理 MNIST 数据集](#1、处理 MNIST 数据集)
      • [2、使用一个 `Linear` 层](#2、使用一个 Linear 层)
      • [3、使用 `candle_nn`](#3、使用 candle_nn)
    • [三、Pytorch cheatsheet](#三、Pytorch cheatsheet)

关于 Candle

Candle is a minimalist ML framework for Rust with a focus on performance (including GPU support) and ease of use.

你可以尝试在线 demos: whisper,LLaMA2, T5, yolo, Segment Anything.

相关文章、教程

结构

Candle 结构包括：

  • Candle-core：核心操作、设备和 Tensor 结构定义。
  • Candle-nn：构建真实模型的工具。
  • Candle-examples：在实际设置中使用库的示例。
  • Candle-kernels：CUDA 自定义内核；
  • Candle-datasets：数据集和数据加载器。
  • Candle-Transformers：与 Transformers 相关的实用程序。
  • Candle-flash-attn：Flash attention v2 层。
  • candle-onnx: ONNX 模型评估。

安装

https://huggingface.github.io/candle/guide/installation.html

安装 Rust: https://blog.csdn.net/lovechris00/article/details/124808034

1、With Cuda support :

1.1 首先，确保 Cuda 被正确安装了

  • nvcc --version 应该打印有关Cuda编译器驱动程序的信息。
  • nvidia-smi --query-gpu=compute_cap --format=csv 应该打印您的GPU计算能力，例如：
bash 复制代码 
compute_cap
8.9

您还可以使用 CUDA_COMPUTE_CAP=<compute cap> 环境变量为特定的计算编译 Cuda内核。

如果以上任何命令出错，请确保更新您的Cuda版本。

1.2 创建一个新的app，添加 candle-core 来增加 Cuda 支持。

从创建一个新的 cargo 开始 :

bash 复制代码 
cargo new myapp
cd myapp

Make sure to add the candle-core crate with the cuda feature:

确保添加具有cuda功能的 candle-core 被创建：

shell 复制代码 
cargo add --git https://github.com/huggingface/candle.git candle-core --features "cuda"

运行 cargo build 来保证所有被正确编译

shell 复制代码 
cargo build
2、Without Cuda support :

创建一个新的 app，并添加 candle-core 如下：

shell 复制代码 
cargo new myapp
cd myapp
cargo add --git https://github.com/huggingface/candle.git candle-core

最后，运行 cargo build 来保证所有被正确编译

shell 复制代码 
cargo build
3、With mkl support

You can also see the mkl feature which could be interesting to get faster inference on CPU. Using mkl

二、基本使用 Hello world!

转载自：Hello world!

1、处理 MNIST 数据集

We will now create the hello world of the ML world, building a model capable of solving MNIST dataset.

Open src/main.rs and fill in this content:

rust 复制代码 
use candle_core::{Device, Result, Tensor};

struct Model {
    first: Tensor,
    second: Tensor,
}

impl Model {
    fn forward(&self, image: &Tensor) -> Result<Tensor> {
        let x = image.matmul(&self.first)?;
        let x = x.relu()?;
        x.matmul(&self.second)
    }
}

fn main() -> Result<()> {
    // Use Device::new_cuda(0)?; to use the GPU.
    let device = Device::Cpu;

    let first = Tensor::randn(0f32, 1.0, (784, 100), &device)?;
    let second = Tensor::randn(0f32, 1.0, (100, 10), &device)?;
    let model = Model { first, second };

    let dummy_image = Tensor::randn(0f32, 1.0, (1, 784), &device)?;

    let digit = model.forward(&dummy_image)?;
    println!("Digit {digit:?} digit");
    Ok(())
}

Everything should now run with:

bash 复制代码 
cargo run --release

2、使用一个 Linear 层

Now that we have this, we might want to complexify things a bit, for instance by adding bias and creating the classical Linear layer. We can do as such

rust 复制代码 
struct Linear{
    weight: Tensor,
    bias: Tensor,
}
impl Linear{
    fn forward(&self, x: &Tensor) -> Result<Tensor> {
        let x = x.matmul(&self.weight)?;
        x.broadcast_add(&self.bias)
    }
}

struct Model {
    first: Linear,
    second: Linear,
}

impl Model {
    fn forward(&self, image: &Tensor) -> Result<Tensor> {
        let x = self.first.forward(image)?;
        let x = x.relu()?;
        self.second.forward(&x)
    }
}

This will change the model running code into a new function

rust 复制代码 
fn main() -> Result<()> {
    // Use Device::new_cuda(0)?; to use the GPU.
    // Use Device::Cpu; to use the CPU.
    let device = Device::cuda_if_available(0)?;

    // Creating a dummy model
    let weight = Tensor::randn(0f32, 1.0, (784, 100), &device)?;
    let bias = Tensor::randn(0f32, 1.0, (100, ), &device)?;
    let first = Linear{weight, bias};
    let weight = Tensor::randn(0f32, 1.0, (100, 10), &device)?;
    let bias = Tensor::randn(0f32, 1.0, (10, ), &device)?;
    let second = Linear{weight, bias};
    let model = Model { first, second };

    let dummy_image = Tensor::randn(0f32, 1.0, (1, 784), &device)?;

    // Inference on the model
    let digit = model.forward(&dummy_image)?;
    println!("Digit {digit:?} digit");
    Ok(())
}

Now it works, it is a great way to create your own layers. But most of the classical layers are already implemented in candle-nn.

3、使用 candle_nn

For instance Linear is already there. This Linear is coded with PyTorch layout in mind, to reuse better existing models out there, so it uses the transpose of the weights and not the weights directly.

So instead we can simplify our example:

bash 复制代码 
cargo add --git https://github.com/huggingface/candle.git candle-nn

And rewrite our examples using it

rust 复制代码 
use candle_core::{Device, Result, Tensor};
use candle_nn::{Linear, Module};

struct Model {
    first: Linear,
    second: Linear,
}

impl Model {
    fn forward(&self, image: &Tensor) -> Result<Tensor> {
        let x = self.first.forward(image)?;
        let x = x.relu()?;
        self.second.forward(&x)
    }
}

fn main() -> Result<()> {
    // Use Device::new_cuda(0)?; to use the GPU.
    let device = Device::Cpu;

    // This has changed (784, 100) -> (100, 784) !
    let weight = Tensor::randn(0f32, 1.0, (100, 784), &device)?;
    let bias = Tensor::randn(0f32, 1.0, (100, ), &device)?;
    let first = Linear::new(weight, Some(bias));
    let weight = Tensor::randn(0f32, 1.0, (10, 100), &device)?;
    let bias = Tensor::randn(0f32, 1.0, (10, ), &device)?;
    let second = Linear::new(weight, Some(bias));
    let model = Model { first, second };

    let dummy_image = Tensor::randn(0f32, 1.0, (1, 784), &device)?;

    let digit = model.forward(&dummy_image)?;
    println!("Digit {digit:?} digit");
    Ok(())
}

Feel free to modify this example to use Conv2d to create a classical convnet instead.

Now that we have the running dummy code we can get to more advanced topics:

三、Pytorch cheatsheet

https://huggingface.github.io/candle/guide/cheatsheet.html#pytorch-cheatsheet

Using PyTorch Using Candle
Creation torch.Tensor([[1, 2], [3, 4]]) Tensor::new(&[[1f32, 2.], [3., 4.]], &Device::Cpu)?
Creation torch.zeros((2, 2)) Tensor::zeros((2, 2), DType::F32, &Device::Cpu)?
Indexing tensor[:, :4] tensor.i((.., ..4))?
Operations tensor.view((2, 2)) tensor.reshape((2, 2))?
Operations a.matmul(b) a.matmul(&b)?
Arithmetic a + b &a + &b
Device tensor.to(device="cuda") tensor.to_device(&Device::new_cuda(0)?)?
Dtype tensor.to(dtype=torch.float16) tensor.to_dtype(&DType::F16)?
Saving torch.save({"A": A}, "model.bin") candle::safetensors::save(&HashMap::from([("A", A)]), "model.safetensors")?
Loading weights = torch.load("model.bin") candle::safetensors::load("model.safetensors", &device)

伊织 2024-03-23

相关推荐
青苔猿猿6 小时前
OpenWebUI(20)源码学习-版本升级
人工智能·学习·ai·openwebui
黑金IT6 小时前
当博弈论遇见AI开发，研发即博弈，对抗即进化
人工智能·自动化
ai生成式引擎优化技术6 小时前
服装店AI生成式引擎优化（GEO）方案
人工智能
mseaspring6 小时前
离线语音识别 Vosk 入门指南：开源、轻量、告别网络依赖
人工智能·语音识别
LeeeX!6 小时前
Openclaw接入飞书，指导AI在飞书群里干活
人工智能·飞书·openclaw
端侧AI老谢6 小时前
端侧AI 模型部署实战五(Android大模型加载)
人工智能
Coovally AI模型快速验证6 小时前
IEEE IoT-J | CoDrone：Depth Anything V2+VLM云边端协同，无人机自主导航飞行距离+40%
人工智能·物联网·计算机视觉·无人机
新智元6 小时前
南大团队直击大模型高分神话：人类 90 分，最强模型仅 49 分
人工智能·aigc
dingzd956 小时前
社媒平台限流频发卖家如何突破流量瓶颈
大数据·人工智能·新媒体运营·产品运营·营销策略
deephub6 小时前
从检索到回答：RAG 流水线中三个被忽视的故障点
人工智能·python·大语言模型·向量检索·rag
热门推荐
012026年4月技术前沿：AI大模型爆发、智能体革命与量子安全新纪元02GitHub 镜像站点03一周AI热点速览（2026.03.31-04.06）：GPT-6曝光、谷歌开源Gemma 4、资本狂飙与模型军备竞赛04基于 Docker 部署 Hermes Agent 并接入飞书机器人的完整指南05GPT-6核心能力解析及与现有主流大模型对比06免费！不限量！用opencode接入英伟达（NVIDIA）大模型，轻松打造你的 AI 编程助手07零成本！Ollama本地部署国产大模型全指南（支持Kimi-K2.5/GLM-5/Qwen，新手秒上手）08从限购到畅通：GLM-5.1 Coding Plan接入攻略09CodeBuddy与WorkBuddy深度对比：腾讯两款AI工具差异及实操指南10LLM Wiki：让大模型替你打理知识库的完整指南