I see, it's more of a comfortability level thing. I can understand why that would make someone a little apprehensive.
I haven't played around with local models before, but I should check them out – Thanks for the suggestion. I've got everything virtualized on a HP ProLiant DL360p Gen8, so no good GPU to speak of at the moment, but I may pick something up soon. On Thursday, November 28, 2024 at 4:09:17 PM UTC-5 [email protected] wrote: > On Thu, Nov 28, 2024 at 4:00 PM Gary Roach <[email protected]> wrote: > >> I was actually thinking about making an importer that sends transaction >> statements to chatgpt and extracts the information in beancount format. >> It's amazing at parsing pdfs and csv files, and unlike institution specific >> importers you'd never have to worry about the institution making format >> changes which break your importers. >> > > If you have a semi-decent GPU (I have an old RTX 3060) you can run free > models on your own computer and do some extraction. > I kicked the tires on "Llama 3.2 Vision" a few days ago this way and could > run some OCR tasks. > No need to send things up to an API if the free models are good enough for > your particular task. > >> Convenience typically sacrifices some amount of security, but why are we >> concerned about our banking transactions being made accessible to other >> companies? Aren't they already public in the sense that every transaction >> involves multiple parties other than yourself (banking institution/broker, >> employer, merchant, credit card processor, etc...). >> >> Is there something I'm missing that could be exploited if an organization >> or even an individual accessed my entire ledger? >> > > Every person has a different threshold, but personally I'm not comfortable > with my personal data going out to an API. > Got nothing to hide, but I don't walk around naked either. > > > > > >> On Saturday, November 23, 2024 at 3:04:47 PM UTC-5 Martin Blais wrote: >> >>> On Sat, Nov 23, 2024 at 3:02 PM Martin Blais <[email protected]> wrote: >>> >>>> On Sat, Nov 23, 2024 at 3:01 PM Martin Blais <[email protected]> wrote: >>>> >>>>> Hey Marvin, >>>>> Do you know if there's a Google service for code completion similar to >>>>> Copilot? >>>>> Do you know if people are realistically running CodeGemma locally? >>>>> I see it on HF: https://huggingface.co/blog/codegemma >>>>> >>>> >>>> Hmm, I see it's supported by Ollama: >>>> https://ollama.com/library/codegemma >>>> I wonder if it's easy to setup in Emacs >>>> >>> >>> Oh my... Ellama. >>> https://www.rahuljuliato.com/posts/ellama >>> (Sorry, I'm still catching up with the universe.) >>> >>> >>> >>>> On Sat, Nov 23, 2024 at 2:47 PM Marvin Ritter <[email protected]> >>>>> wrote: >>>>> >>>>>> If you have Copliot enabled I would recommend enabling it for >>>>>> specific file types/languages and disable it by default. I think it's >>>>>> easy >>>>>> to forget a file type with sensitive content. And you can always enable >>>>>> it >>>>>> for a language if you forgot it. >>>>>> >>>>>> On Sun, Nov 17, 2024 at 6:19 AM Red S <[email protected]> wrote: >>>>>> >>>>>>> If you installed Github Copilot in your personal code >>>>>>> editor/computer, be aware that it uploads "snippets" of your input >>>>>>> files to >>>>>>> it and possibly to third-party APIs (e.g., OpenAI). I think people are >>>>>>> just >>>>>>> beginning to become aware of the implications of this due to their >>>>>>> employers crafting policies around what LLMs they can use and what-not, >>>>>>> but >>>>>>> it's still early days and it's easy to accidentally screw up, so here >>>>>>> are some thoughts about this. >>>>>>> >>>>>>> I think it's really easy to install Github Copilot to get code >>>>>>> completions in say, Emacs, and then to open up your ledger and it's in >>>>>>> Copilot minor-mode everywhere (for example if you enabled it via >>>>>>> `(add-hook >>>>>>> 'prog-mode-hook 'copilot-mode)` or similar, to be turned on everywhere >>>>>>> ("it's amazing, right?")), which means you get completions on its >>>>>>> contents. >>>>>>> AFAICT it's impossible to know how much context is sent up to the >>>>>>> models >>>>>>> for queries. GH claims general "context" is sent: >>>>>>> >>>>>>> >>>>>>> Glad you brought this up. The first thing I did before installing >>>>>>> Copilot long ago was to solve for this. I use both Copilot and Codeium >>>>>>> with >>>>>>> Neovim personally. In short, here are some options I found. These work >>>>>>> well >>>>>>> for folks who use terminal based editors (vim/emacs, mostly): >>>>>>> >>>>>>> 1. configure Copilot/Codeium/AI in your editor to be disabled >>>>>>> for certain file types >>>>>>> 2. configure your editor to disable the Copilot/Codeium/AI >>>>>>> plugin for certain file types >>>>>>> 3. entirely disable network access from your editor >>>>>>> >>>>>>> (1) involves trusting the plugin under question, which isn’t a great >>>>>>> idea. >>>>>>> >>>>>>> (2) is better, but I found how easy it was to mess this up and get >>>>>>> it wrong. Editor configurations for power users span many files and >>>>>>> directories, and it’s easy to overlook something when updating your >>>>>>> config >>>>>>> >>>>>>> (3) is best (most secure), and I use it for things I need most >>>>>>> security for (files with account numbers, passwords, cloud API keys, >>>>>>> and >>>>>>> other sensitive data). My setup is to run a separate instance of neovim >>>>>>> via >>>>>>> flatpak. Under the hood, it’s essentially containerized execution of >>>>>>> neovim, which means all one has to do is to disable the network >>>>>>> interface >>>>>>> on that container like so: >>>>>>> my_editor_secure () { # my editor uses a gpg plugin for which it >>>>>>> needs to access the gpg-agent flatpak run --user --unshare=network >>>>>>> --socket=gpg-agent io.neovim.nvim $* + } >>>>>>> >>>>>>> Which guarantees nothing will leave your computer. You could simply >>>>>>> make this your default editor command, and occasionally run it with >>>>>>> network >>>>>>> access enabled if you need to update plugins and such. >>>>>>> >>>>>>> -- >>>>>>> You received this message because you are subscribed to the Google >>>>>>> Groups "Beancount" group. >>>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>>> send an email to [email protected]. >>>>>>> To view this discussion visit >>>>>>> https://groups.google.com/d/msgid/beancount/e955bcd7-6ab1-4e2f-bf35-e9d755858a02n%40googlegroups.com >>>>>>> >>>>>>> <https://groups.google.com/d/msgid/beancount/e955bcd7-6ab1-4e2f-bf35-e9d755858a02n%40googlegroups.com?utm_medium=email&utm_source=footer> >>>>>>> . >>>>>>> >>>>>> -- >>>>>> You received this message because you are subscribed to the Google >>>>>> Groups "Beancount" group. >>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>> send an email to [email protected]. >>>>>> To view this discussion visit >>>>>> https://groups.google.com/d/msgid/beancount/CAPytOJG4LUocPDv3HEaRmBk3u%2BzFijE5a72g6xhMe1asjaC-GQ%40mail.gmail.com >>>>>> >>>>>> <https://groups.google.com/d/msgid/beancount/CAPytOJG4LUocPDv3HEaRmBk3u%2BzFijE5a72g6xhMe1asjaC-GQ%40mail.gmail.com?utm_medium=email&utm_source=footer> >>>>>> . >>>>>> >>>>> -- >> You received this message because you are subscribed to the Google Groups >> "Beancount" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> > To view this discussion visit >> https://groups.google.com/d/msgid/beancount/e7df6e60-cf2d-4914-83ff-ac8779432e4en%40googlegroups.com >> >> <https://groups.google.com/d/msgid/beancount/e7df6e60-cf2d-4914-83ff-ac8779432e4en%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> > -- You received this message because you are subscribed to the Google Groups "Beancount" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/d/msgid/beancount/5bd7b98a-f458-4665-9521-c0f0b075d888n%40googlegroups.com.
