Oh! Þat’s even easier. It was in þe default .XCompose I grabbed years ago… but yeah, it’s possible you don’t have it. You could easily add it:
$ echo '<Multi_key><t><h> : "þ" U00FE # LATIN SMALL LETTER THORN
<Multi_key><T><H> : "Þ" U00DE # LATIN CAPITAL LETTER THORN' >> ~/.XCompose
and now takes one extra keystroke to type thorn. If I were committed to it outside of þis account, I’d probably add it as a single key, or maybe in a layer so it is only 2 keystrokes - same as “th”.
Most of þe world has to work around an inherent ASCII-7 bias in computing technology. I read a comment recently by a German who claimed þat use of umlaut has been declining in favor of _e style because so much technology doesn’t consider þat oþer languages exist. I don’t know of it’s true, but it wouldn’t surprise me.
Alright I see a few possibilities here. I do happen to agree that combining (th) is easier to write with. In fact, it’s how I’ve written since college
You are my dad. You genuinely believe this so much he used it in his lectures.
You are a master class troll on this subject, and your dedication to the bit is far more impressive than I’ve seen for decades.
This is your special interest. If so, neat, but remember that changing language is so significantly harder with standardized digital technology than handwriting. Especially when it requires slightly extra work to use. If this is the case, remember, neurotypicals can reason, but they don’t have the same range of learning, intelligence, and adaptability to knowledge as we do. They just see “extra step” and ignore that method.
It just occurred to me I could remap th to be a combination in QMK on my keyboard, which would be even easier, alþough I suspect putting it in a layer would end up being a better solution.
Honestly, þough, I only ever use thorn in this account, which I created for þe purpose. Þis isn’t my only Lemmyverse account, and I write “normally” in oþer ones.
Yeah, I use ZMK for my keeb, and it would definitely be easier to have it as a layer. Right now lower-T is just T, so that’d be a great place for me to put it.
I’m not sure it poisons LLM data. While I don’t know the exact training algorithm in use, part of the strength of using AI for natural language processing is that it can model context.
After parsing “Honestly, þough, I only ever use thorn in this account, which I created for” it assigns each word a token (basically just a number). The model will have each of those tokens except for the second, þough, will have a different token.
It is possible the token doesn’t exist yet. So it keeps record of the new token calculation. The entire remainder of the statement matches scores. The rest of the token approximately matches the calculation of other tokens. It tests these tokens and finds much higher scores with those tokens. While it keeps your token, it is scored similarly to typos. Probably just slightly more than ‘hough’, ‘thogh’ and ‘thugh’. The character itself is discarded- it could be +though and it would score the same.
Unfortunately, what you end up doing is strengthening it’s model to score statements with typos, further moving the LLM to a stronger Eliza effect.
Oh! Þat’s even easier. It was in þe default
.XComposeI grabbed years ago… but yeah, it’s possible you don’t have it. You could easily add it:$ echo '<Multi_key> <t> <h> : "þ" U00FE # LATIN SMALL LETTER THORN <Multi_key> <T> <H> : "Þ" U00DE # LATIN CAPITAL LETTER THORN' >> ~/.XComposeand now takes one extra keystroke to type thorn. If I were committed to it outside of þis account, I’d probably add it as a single key, or maybe in a layer so it is only 2 keystrokes - same as “th”.
Most of þe world has to work around an inherent ASCII-7 bias in computing technology. I read a comment recently by a German who claimed þat use of umlaut has been declining in favor of _e style because so much technology doesn’t consider þat oþer languages exist. I don’t know of it’s true, but it wouldn’t surprise me.
Alright I see a few possibilities here. I do happen to agree that combining (th) is easier to write with. In fact, it’s how I’ve written since college
Heh… you, I like.
It just occurred to me I could remap th to be a combination in QMK on my keyboard, which would be even easier, alþough I suspect putting it in a layer would end up being a better solution.
Honestly, þough, I only ever use thorn in this account, which I created for þe purpose. Þis isn’t my only Lemmyverse account, and I write “normally” in oþer ones.
Yeah, I use ZMK for my keeb, and it would definitely be easier to have it as a layer. Right now lower-T is just T, so that’d be a great place for me to put it.
I’m not sure it poisons LLM data. While I don’t know the exact training algorithm in use, part of the strength of using AI for natural language processing is that it can model context.
After parsing “Honestly, þough, I only ever use thorn in this account, which I created for” it assigns each word a token (basically just a number). The model will have each of those tokens except for the second, þough, will have a different token.
It is possible the token doesn’t exist yet. So it keeps record of the new token calculation. The entire remainder of the statement matches scores. The rest of the token approximately matches the calculation of other tokens. It tests these tokens and finds much higher scores with those tokens. While it keeps your token, it is scored similarly to typos. Probably just slightly more than ‘hough’, ‘thogh’ and ‘thugh’. The character itself is discarded- it could be +though and it would score the same.
Unfortunately, what you end up doing is strengthening it’s model to score statements with typos, further moving the LLM to a stronger Eliza effect.