291–300 of 752 posts
I would like to use GPT for engineering problems at work, but it’s just not practical for me since I work on a large internal codebase. It is very rare for me to need self-contained code solutions on the order of <100 lines.
Earlier quoted context omitted.
This is one of the causes there's a push to run your own engines for large language models: if you run your own service you can control the environment, data and reproducibility.
This is exactly what my employer is doing, they pay so that our internal data (from employee queries) does not become part of the model. They've blocked the public chat gpt etc.
Earlier quoted context omitted.
How does this work exactly? What is ChatGPT doing for you?
For each cheesy training video, playwright opens the page, runs a timer, clicks an element to load the quiz page, copies the question and possible answers, and sends them to the GPT-4 API for the best possible solution, clicks on the element that was answered as the most likely solution, and repeats. Since these video/quizzes are recurring and terrible, it was worth automating
I tried, not actually at work but for something open source, to get it to help me rework a terraform provider using the old SDK to use the new 'framework' - I've only ever written Go for terraform providers (and that not much) so I often have dumb confusions about how to do something that end up more about Go than the API itself, it's just not necessarily obvious initially, or what to search for etc., so I thought it…
I've had it do some things in brilliant ways. Asking it to create typescript types for given data (and multiple examples to do union types). I've had it help me create migrations for Prisma schema that doesn't lose data. I've asked it to convert one format (html to jsx) and it can get that right.
Often you still have to correct it, but it gives amazing starting points.
But if it's wrong arguing it with it is a mistake. It's not aware, and it can't learn on the fly (currently). If you close the chat and then repeat the prompts exactly it will make the same mistakes again.
I’ve been actively using it and it’s become my go-to in a lot of cases - Google is more for verification when I smell something off or if it doesn’t have up to date information. Here are some examples: • reviewing contract changes, explaining hard to parse legalese • advice on accounting/tax when billing international clients • visa application • boilerplate django code • learnt all about smtp relays, and requirement…
You mentioned 50/50 correctness in domain questions. I can't be sure that other hard to verify questions do not follow these percentage..
I use it a lot for Linux administration, troubleshooting, and scripting as well as some programming. I've only recently started using GPT4 and the API so I've only been using the chat interface so far. Examples of some stuff I've asked it just today: - Asked which config files handled sleep mode when lid is closed and kept fixing it and asking for more possible locations until it fixed my issue (going into sleep duri…
>Google is fucked The last year or two have proved to me that they deserve it. Their search engine is utter dogshit now. Ultra commercialised, full of ads, so hard to find a result that is actually what I searched for. Guess that's what Google level greed will do.
Excite, Yahoo, and Alta Vista are welcoming Google to search giants retirement home.
We dread when the SEO game ruins ChatGPT
I pasted some headers (some of them very long, in multiple chunks) and asked the bot to write me a summary of what the API does, and what could be used to make it better. Answer were spot on . Free documentation! If you also paste the source, it can give you a documentation for a particular function as well. Same for rather complex Makefiles. Otherwise I do like others, I use it as a quick stack overflow for uncommon…
Earlier quoted context omitted.
So on my coding problems I haven't had much luck. It doesn't seem to know Bazel, the Rust code I asked about was completely hallucinated, but it did solve a problem with Azure DevOps I had. I think if the training set did not contain enough of something it can't really think of a solution. What is really nice though it's as you say the refinement of questions. Sometimes it's hard to think of the right query, maybe yo…
I've found that it's much worse for languages like rust than it is for things like typescript and python. The thing AI seems to be really great at is writing boilerplate code like argument parsing for CLI tools.
Earlier quoted context omitted.
So on my coding problems I haven't had much luck. It doesn't seem to know Bazel, the Rust code I asked about was completely hallucinated, but it did solve a problem with Azure DevOps I had. I think if the training set did not contain enough of something it can't really think of a solution. What is really nice though it's as you say the refinement of questions. Sometimes it's hard to think of the right query, maybe yo…
Yeah, I gave it a simple task of encoding a secret message in a sentence by using the first letter of every word. Hello = "Hey everyone, lick less onions". I worked with the prompts for over an hour to try to get it to complete the task, and while I did have some success, it really struggled to reason about the task or provide a valid response. If it can't even reason about a child's game, I can imagine it struggles…
Bing in precise mode had no trouble at all.
Me - Let's generate an acrostic sentence for "jack and jill went up the hill"
Bing - Sure! Here’s an acrostic sentence for the phrase “jack and jill went up the hill”:
Just as children know and never doubt, joyfully in love, laughing while enjoying nature together under pure trees, having endless happy interactions, laughing loudly.