每一个程序员心中的私有Her来了——Codestral-22B，一个会代码的Agent

Mistralai最新发布了22B的Codestral。

Codestrall-22B-v0.1 在一个包含 80 多种编程语言的不同数据集上进行了训练，其中包括最流行的语言，如 Python、Java、C、C++、JavaScript 和 Bash（更多详情请参见博文）。可以对模型进行查询：

• 作为指示，例如回答有关代码片段的任何问题（编写文档、解释、因式分解）或根据特定指示生成代码
• 作为 “中间填充”（FIM），预测前缀和后缀之间的中间标记（对软件开发插件（如 VS Code）非常有用。

安装

pip install mistral_inference

下载

from huggingface_hub import snapshot_download
from pathlib import Path

mistral_models_path = Path.home().joinpath('mistral_models', 'Codestral-22B-v0.1')
mistral_models_path.mkdir(parents=True, exist_ok=True)

snapshot_download(repo_id="mistralai/Codestral-22B-v0.1", allow_patterns=["params.json", "consolidated.safetensors", "tokenizer.model.v3"], local_dir=mistral_models_path)

聊天

mistral-chat $HOME/mistral_models/Codestral-22B-v0.1 --instruct --max_tokens 256

Prompt：“Write me a function that computes fibonacci in Rust”

Sure, here's a simple implementation of a function that computes the Fibonacci sequence in Rust. This function takes an integer `n` as an argument and returns the `n`th Fibonacci number.

fn fibonacci(n: u32) -> u32 {
    match n {
        0 => 0,
        1 => 1,
        _ => fibonacci(n - 1) + fibonacci(n - 2),
    }
}

fn main() {
    let n = 10;
    println!("The {}th Fibonacci number is: {}", n, fibonacci(n));
}

This function uses recursion to calculate the Fibonacci number. However, it's not the most efficient solution because it performs a lot of redundant calculations. A more efficient solution would use a loop to iteratively calculate the Fibonacci numbers.

Fill-in-the-middle (FIM)

安装 mistral_inference 并运行 pip install --upgrade mistral_common，确保已安装 mistral_common>=1.2：

from mistral_inference.model import Transformer
from mistral_inference.generate import generate
from mistral_common.tokens.tokenizers.mistral import MistralTokenizer
from mistral_common.tokens.instruct.request import FIMRequest

tokenizer = MistralTokenizer.v3()
model = Transformer.from_folder("~/codestral-22B-240529")

prefix = """def add("""
suffix = """    return sum"""

request = FIMRequest(prompt=prefix, suffix=suffix)

tokens = tokenizer.encode_fim(request).tokens

out_tokens, _ = generate([tokens], model, max_tokens=256, temperature=0.0, eos_id=tokenizer.instruct_tokenizer.tokenizer.eos_id)
result = tokenizer.decode(out_tokens[0])

middle = result.split(suffix)[0].strip()
print(middle)

结果

num1, num2):

    # Add two numbers
    sum = num1 + num2

    # return the sum