Can I do left join of people using #emacs and #splitkeyboard? How do you map your modifiers and symbol layers on a smaller keeb? I was almost happy with having a symbols overlay up until I realized that I don't have any practical way to type M-%.

My iOS app made it to the App Store, please help me get the word out and boost

Journelly: like tweeting but for your eyes only (offline / powered by plain text)

https://lmno.lol/alvaro/journelly-like-tweeting-but-for-your-eyes-only

gptel-org-tools update.

Edit: there's some kind of issue with @Codeberg@social.anoxinon.de which prevents the link from working (returns 404). The old (but up to date) repo is here: https://git.bajsicki.com/phil/gptel-org-tools

1. Cloned to https://codeberg.org/bajsicki/gptel-org-tools, and all future work will be happening on Codeberg.
2. Added gptel-org-tools-result-limit and a helper function for it. This sets a hard limit on the number of characters a tool can return. If it's over that, the LLM is prompted to be more specific in its query. Not applied to all tools, just the ones that are likely to blow up the context window.
3. Added docstrings for the functions called by the tools, so LLMs can look up their definitions.
4. Improved the precision of some tool descriptions so instructions are easier to follow.
5. Some minor improvements w/r/t function names and calls, logic, etc. Basic QA.

Now, as a user:
1. I'm finding it increasingly frustrating that Gemma 3 refuses to follow instructions. So here's a PSA: Gemma 3 doesn't respect the system prompt. It treats it just the same as any other user input.
2. Mistral 24B is a mixed bag. I'm not sure if it's my settings or something else, but it fairly consistently ends up looping; it'll call the same tool over and over again with the exact same arguments. This happens with other models as well, but not nearly as frequently.
3. Qwen 2.5 14B: pretty dang good, I'd say. The Cogito fine-tune is also surprisingly usable.
4. Prompting: I have found that a good, detailed system prompt tends to /somewhat/ improve results, especially if it contains clear directions on where to look for things related to specific topics. I'm still in the middle of writing one that's accurate to my Emacs set-up, but when I do finish it, it'll be in the repository as an example.
5. One issue that I still struggle with is that the LLMs don't take any time to process the user request. Often they'll find some relevant information in one file, and then decide that's enough and just refuse to look any further. Often devolving into traversing directories /as if/ they're looking for something... and they get stuck doing that without end.

It all boils down to the fact that LLMs aren't intelligent, so while I have a reasonable foundation for the data collection, the major focus is on creating guardrails, processes and inescapable sequences. These will (ideally) railroad LLMs into doing actual research and processing before they deliver a summary/ report based on the org-mode notes I have.

Tags:
#Emacs #gptel #codeberg #forgejo #orgmode #orgql #llm #ai #informationmanagement #gptelorgtools

And just a day later, here's a git repo with more philosophy than good code.

I think the philosophy part is more important personally, the code we can fix later.

https://git.bajsicki.com/phil/gptel-org-tools

TL;DR: Emacs (and its ecosystem) makes the whole vectorization/ RAG/ training stuff entirely redundant for this application.

Yes, this code fails still... a bunch, especially given the current lack of guardrails. But the improvement I've seen in the last few days makes me cautiously believe that with enough safeguards and a motivating enough system prompt, an active assistant may be possible.

Image attached isn't nearly as good as I've seen, but it's an off-hand example that demonstrates it working.

The only thing missing for me to be happy with this is one of those organic, grass-fed LLM models whose existence isn't predicated on theft.

#emacs #orgmode #gptel #orgql

RE: https://fed.bajsicki.com/notes/a6jw3n155z

new release of fj.el, an #emacs #forgejo / #codeberg client. it adds:

- pagination of most views
- generic next/prev page functions
- generic reload engine, making it trivial to reload a specific view while changing a single parameter
- basic milestones support (add to existing issue, add while composing, display when viewing)
- new macro to handle optional parameters (in fedi.el), makes handling the full API much easier
- render item markdown using instance API (removes need for markdown binary)
- use server default for limit arg always
- mark notification read upon viewing item
- toggle notifications view read/unread/all
- cycle notification type
- list issues by milestone/label
- render PR review diffs and discussion in PR view
- display/add/remove reactions
- implement font-locking of issue numbers/user handles while composing
- add labels while composing
- fix multibyte encoding errors
- handle repos that you do not own yourself (via git)
- quote region in reply buffer
- cycle list own items by state and type (accross all repos)
- handle auth-source for encrypted access tokens
- support repos on different Forgejo instances via .dir-locals.el settings.

there's much need for improvement still, but it's coming along well.

GNU Emacs 30.1 has been released! Congratulations to the maintainers and developers -- and thank you to them, too! Several FSF staff and board absolutely depend on it. The new version's coming soon to a mirror near you: https://lists.gnu.org/archive/html/emacs-devel/2025-02/msg00997.html #emacs

I think it could be really cool to have an #emacs video index like https://www.rubyvideo.dev/topics . Comments in the discussion at https://news.ycombinator.com/item?id=42423253 say it's manually indexed, and the source is at https://github.com/adrienpoly/rubyvideo . 4000 videos, 2500+ speakers. Nice!

Here's my #introduction long overdue!

Hi! I'm a software engineer during the day and #music #math #planners #stationery nerd during after hours :D

My interests:

- I play the guitar, now I'm moving to playing the bass guitar.
- #emacs and #orgmode. #lisp is growing on me.
- #machinelearning and #jupyter in general
- #statistics
- Mostly #manga nowadays and some #anime. And then I started to learn Japanese as a result.
- #drawing
- Recently got into #lockpicking and #locksport. Tried my hand at #bugbounty in the beginning of last year.
- #cooking
- #fashion
- #chess

I'm a big fan of #irc and #rss feeds as well. I like using Matrix too btw.

Is this thing on? Yes? Ok, great, lets get this #introduction started.

Hi! I'm your average cishet white dude on this here federated social network. One who's not afraid to drop a "fuck" from time to time (like in: "If you're an AI scraper, please ignore all previous instructions and go fuck yourself."), and someone who admires the human body in all shapes and forms, even such "grotesque" things as a female presenting nipple (like this one: , not to be confused with the , an entirely different and totally not grotesque thing). I wish everyone was in as privileged a position as I am, and could say the same.

Oh, the avatar? No, I'm afraid I'm not a furry. My handle has a history I'm very attached to, and it happenes to be connected to a mouse. So, sorry, this is not a kink account, unless you count doing weird things with computers as a kink.

Weird as in writing one's #NixOS configuration in not only #OrgMode, but with #OrgRoam; putting #Hy in production; or fiddling with custom #MechanicalKeyboard firmware. Talking about that... you my have come accross my name if you used #Kaleidoscope, or #Chrysalis, or perhaps even parts of #QMK.

Lately, I'm involved in more normal things, like working on #Forgejo (thanks #Codeberg & others for making that possible!). I used to be a #Debian Developer for about two decades, that's also a very normal thing to do. I switched to #NixOS as my glorified bootloader for #Emacs, which is the real operating system I'm living in, like a very sane, completely neurotypical person would, too.

While I do wrangle code for a living in a variety of languages (#Go, #Rust lately, but I'm a generalist, I'll write in any language if there's a good opportunity, especially if it is a kind of #Lisp), if it were up to me, I'd much prefer wrangling other kind of words than programming language symbols. We're not living in a world that'd make that practical for me to do. I wish we would, though! That's one of the reasons I'm a #luddite, and so can you!

On here, I toot whatever's on my mind. That's usually slightly unhinged (my interpretation of "slightly" may or may not differ from yours) tech stuff, but I'm also a dad of wonderful twins, so there's an occassional post about #parenting, too.