laroxinia.bsky.social
I like thinking about things
7 posts
0 followers
2 following
comment in response to
post
Please do!
comment in response to
post
I contend that projects such as OLMo, which are fully open source and not trained on stolen data, are proof that there is absolutely a way LLMs can be built and used in an ethical way. Though such projects are definitely exceptions and currently rare.
comment in response to
post
If you're aware of the limits of LLMs, then it's possible to use them for some things. It's a bit like an intern. I don't want an intern to run a country, but the intern can definitely collate given information for me.
Wouldn't trust the intern's memory, but if they can give sources I can check...
comment in response to
post
Furthermore, using a search engine with some keywords and "inurl:reddit.com" as an operator should give you an alright index of experiences too.
comment in response to
post
unique design choices, leading to a product so different from other laptops that reading user experiences becomes needed to make a choice. Perplexity summarizes them OK, but it also indexes Reddit itself well and provides source links, giving you a framework of info to start and then deeper detail.)
comment in response to
post
(I don't know what your opinion is on using this kind of tool for this purpose, but Perplexity gives the option of searching through Reddit, and I find that real user experiences are an essential complement to raw benchmarking. Especially as Apple maintains a separate ecosystem, and has some very...
comment in response to
post
I used my tech knowledge to prompt an LLM into crawling Reddit and the internet to help you with info. Protip: read the Reddit threads in the sources - it contains a lot of real end user experiences. Happy to answer any questions you might have! www.perplexity.ai/search/for-e...