Create data from remote api.

[PoTaTo-Mika]

This pull request built a python script which uses DeepSeek R1 API to solve math problems, then it will build a json file for further training and so on.
As inferring the full R1 is unaffordable for most developers (at least 8*H200), I made this part to reduce the cost, so more and more people can make contribution to this landmark project.
I know my code is very rudimentary, so I hope more people can join in, add more and better features to it, and make it more stable at the same time!

[vins2win]

it is better to avoid hard coding of API keys :

instead of
client = OpenAI(api_key="Your Deepseek API", base_url="https://api.deepseek.com")

use

api_key = os.getenv("DEEPOSEEK_API_KEY")
client = OpenAI(api_key=api_key, base_url="https://api.deepseek.com")

add exception handling :

example :

with open(math_problems, "r") as file: math_problems = file.readlines()

to

try: with open(math_problems, "r") as file: math_problems = file.readlines() except FileNotFoundError: print(f"File {math_problems} not found.") exit(1)

[PoTaTo-Mika]

it is better to avoid hard coding of API keys :

instead of client = OpenAI(api_key="Your Deepseek API", base_url="https://api.deepseek.com")

use

api_key = os.getenv("DEEPOSEEK_API_KEY")
client = OpenAI(api_key=api_key, base_url="https://api.deepseek.com")

add exception handling :

example :

with open(math_problems, "r") as file: math_problems = file.readlines()

to

try: with open(math_problems, "r") as file: math_problems = file.readlines() except FileNotFoundError: print(f"File {math_problems} not found.") exit(1)

Thanks, I've changed my code to make it eazier to manage the api key.