Honestly, if I can use an AI agent that is not coming from an evil corporation that’s destroying the environment, I would use it to help me solve problems like figuring out why I have a build error, why I’m getting certain error messages, why a piece of code isn’t working as I expected, find potential bugs, etc.
It can be a great tool to help you grow, like having a tutor explain where and why you made a mistake so you can learn from it.
NOT if you ask it to do the job for you and generate your code.
I used a local model today to write a long form regex for me today. it would have taken me 30 minutes. instead it took 3 seconds and worked out of the box.
I think this is a perfect example of what LLMs should be used for.
absolutely. I know regex, better than most devs I know personally.
I also don’t use it all the time. I happened to have it up and just said, “hey fucker, give me the pcre regex to find all instances of these keywords in a string like this”
reviewed it, threw what I had into my code, ran my tests and everything was golden.
If I had to pick one, it’d be Olmo 3 from Ai2:
Non-profit. Ai2 is a research institute, not a company that needs to sell you a subscription or harvest your prompts.
Fully open, not just “open weights”
Ai2 publishes the training data, the training code, every checkpoint and the logs. Anyone can check exactly what went in, and in principle rebuild it from scratch. You don’t have to take their word for anything.
Runs locally. The 7B version runs on an ordinary gaming laptop or a recent MacBook. Your code never leaves your machine, there’s no data centre involved when you use it, and the energy cost is your laptop being on.
Apache 2.0 licence. No usage restrictions and no strings attached.
Good enough for the tutor use case. Explaining errors, walking through why code misbehaves, and pointing out likely bugs are all well within what it can do.
To be fair about it: the training data is largely scraped from the web, like nearly every model. The difference is that Olmo is honest and auditable about it.
If strict consent matters most to you, StarCoder2 is cleaner (only permissively licensed code, with an opt-out)
Honestly, if I can use an AI agent that is not coming from an evil corporation that’s destroying the environment, I would use it to help me solve problems like figuring out why I have a build error, why I’m getting certain error messages, why a piece of code isn’t working as I expected, find potential bugs, etc.
It can be a great tool to help you grow, like having a tutor explain where and why you made a mistake so you can learn from it.
NOT if you ask it to do the job for you and generate your code.
I used a local model today to write a long form regex for me today. it would have taken me 30 minutes. instead it took 3 seconds and worked out of the box.
I think this is a perfect example of what LLMs should be used for.
Ok. But did you understand the regex? That’s what people are missing out on. The learning aspect.
absolutely. I know regex, better than most devs I know personally.
I also don’t use it all the time. I happened to have it up and just said, “hey fucker, give me the pcre regex to find all instances of these keywords in a string like this”
reviewed it, threw what I had into my code, ran my tests and everything was golden.
That’s the neat part, you can.
If I had to pick one, it’d be Olmo 3 from Ai2: Non-profit. Ai2 is a research institute, not a company that needs to sell you a subscription or harvest your prompts. Fully open, not just “open weights” Ai2 publishes the training data, the training code, every checkpoint and the logs. Anyone can check exactly what went in, and in principle rebuild it from scratch. You don’t have to take their word for anything.
Runs locally. The 7B version runs on an ordinary gaming laptop or a recent MacBook. Your code never leaves your machine, there’s no data centre involved when you use it, and the energy cost is your laptop being on. Apache 2.0 licence. No usage restrictions and no strings attached.
Good enough for the tutor use case. Explaining errors, walking through why code misbehaves, and pointing out likely bugs are all well within what it can do.
To be fair about it: the training data is largely scraped from the web, like nearly every model. The difference is that Olmo is honest and auditable about it.
If strict consent matters most to you, StarCoder2 is cleaner (only permissively licensed code, with an opt-out)
Devstral is very good