#aboutlastnight
I went to a party, met a nice Belgian guy and spent talking French to him all night. Fabrice’s English is rudimentary so it was impossible for us to switch.
Hallelujah! I keep lamenting my French getting stale and lost but it turns out it doesn’t. Mb it’s true they say it is like with the bycicle, impossible to unlearn.
I thought I was unable to speak French with Ju but Ju is bilingual, the slightest hiccup in my French, I explain it in English, and there we go Englishing away. English is cheating. Also I don’t use Ju as a language exchange ahaha.
I need more French-speaking people who do not speak English. Funny enough you have lots of them in Barcelona. I prefer them not be French though ahaha. Everybody hates French. Northern Africans would be nice, and Belgians ofc.
I went to a party, met a nice Belgian guy and spent talking French to him all night. Fabrice’s English is rudimentary so it was impossible for us to switch.
Hallelujah! I keep lamenting my French getting stale and lost but it turns out it doesn’t. Mb it’s true they say it is like with the bycicle, impossible to unlearn.
I thought I was unable to speak French with Ju but Ju is bilingual, the slightest hiccup in my French, I explain it in English, and there we go Englishing away. English is cheating. Also I don’t use Ju as a language exchange ahaha.
I need more French-speaking people who do not speak English. Funny enough you have lots of them in Barcelona. I prefer them not be French though ahaha. Everybody hates French. Northern Africans would be nice, and Belgians ofc.
Endless hype unhype cycle.
https://twitter.com/jordiae/status/1670502752877682688?s=46&t=tJA3XyJ6UXEb-9Si1PfeJQ
https://twitter.com/jordiae/status/1670502752877682688?s=46&t=tJA3XyJ6UXEb-9Si1PfeJQ
Twitter
I’m sorry but the whole thing is hilarious:
- ML influencer hypes article, presumably without having read it.
- Adds fancy picture reading it to make the tweet viral.
- ML Twitter people criticize the article.
- ML influencer retracts the tweet.
- ML influencer hypes article, presumably without having read it.
- Adds fancy picture reading it to make the tweet viral.
- ML Twitter people criticize the article.
- ML influencer retracts the tweet.
Ksenia's Public Channel on #whatever
Endless hype unhype cycle. https://twitter.com/jordiae/status/1670502752877682688?s=46&t=tJA3XyJ6UXEb-9Si1PfeJQ
Turns out he read it. He just didn’t think about the conclusions that well because lo and behold its MIT quality approved. (MIT quality not anymore.)
Now finally a good question*:
Who first thought that using GPT4 to evaluate GPT4 is a good idea?
*from twi
Who first thought that using GPT4 to evaluate GPT4 is a good idea?
*from twi
The perks of working in Spanish are that you start speaking Spanish faster than you speak English.
I think I learned 4 languages to fluency by myself just because I really love to read. I don’t know any other way of upping your vocabulary fast and in an expansive and sophisticated way.
(My sister is upset because her German isn’t progressing as she’d like. She says: “as soon as the class is done it drains out of me as if it were water”.
Well, if you learn it in vacuum without applying it to anything it does. As soon as you open the book you will see how much you already know.)
(My sister is upset because her German isn’t progressing as she’d like. She says: “as soon as the class is done it drains out of me as if it were water”.
Well, if you learn it in vacuum without applying it to anything it does. As soon as you open the book you will see how much you already know.)
My ex linguistics channel was basically only about this.
(Which is a must-have part. Not must-have in a sense “everybody wearing it this spring”. But in a sense that without it obviously there is no paper.)
Ksenia's Public Channel on #whatever
I don’t like writing “Related work” part 😔
Wrote it.
ChtGPT use of qualitative adverbs and adjectives (the correct word for them is “modifiers” but I know you probably won’t understand what I mean, I didn’t know this term either) is completely off the wall.
I mean, big in big impact, considerably in considerably effective etc.
When you ask it to rewrite a simple not very idiomatically correct sentence / paragraph, it goes all in. It invents all those stupid pompous modifiers, as formidable as in formidable competitor, unequivocally positioning, substantially decline etc. Also nouns. They may seem ok on their own but they are out of place in my text.
Also it writes ENORMOUS sentences. They are convoluted, tangled, intricate etc. You break your brain trying to read this non sense.
If I see all that I will holler: ChatGPT written! It will smell ChatGPT.
ChtGPT use of qualitative adverbs and adjectives (the correct word for them is “modifiers” but I know you probably won’t understand what I mean, I didn’t know this term either) is completely off the wall.
I mean, big in big impact, considerably in considerably effective etc.
When you ask it to rewrite a simple not very idiomatically correct sentence / paragraph, it goes all in. It invents all those stupid pompous modifiers, as formidable as in formidable competitor, unequivocally positioning, substantially decline etc. Also nouns. They may seem ok on their own but they are out of place in my text.
Also it writes ENORMOUS sentences. They are convoluted, tangled, intricate etc. You break your brain trying to read this non sense.
If I see all that I will holler: ChatGPT written! It will smell ChatGPT.
Ok. My algorithm:
1. Ask it to rewrite the text. Phrase by phrase, never the entire paragraph because it starts mixing everything all over the place.
2. Put everything together, check that the logic holds.
3. Ask it to rewrite everything in a concise scientific way. Phrase by phrase. It compresses the sentences but you are sure that every detail that was reflected in the pompous (1) version is on its place.
4. Take the Oxford Collocations Dictionary, go all over the text and replace every crazy modifier with a normal human appropriate word.
5. Profit.
1. Ask it to rewrite the text. Phrase by phrase, never the entire paragraph because it starts mixing everything all over the place.
2. Put everything together, check that the logic holds.
3. Ask it to rewrite everything in a concise scientific way. Phrase by phrase. It compresses the sentences but you are sure that every detail that was reflected in the pompous (1) version is on its place.
4. Take the Oxford Collocations Dictionary, go all over the text and replace every crazy modifier with a normal human appropriate word.
5. Profit.
I am sure that I can put (1) and (3) together but I had to check first.
Also about the above mentioned shortcomings of GPT4 stylistics wrt academical writing. I think when I ask it to rewrite the phrase / paragraph again and again and again the temperature of the model starts to rise and it goes crazy with vocabulary amplifying the semantical circle around concepts so much that the majority of collocations start to sound off. I have an acute sense of language and I begin to notice it.
I am not sure what to do with it but in those cases I started rewriting the phrases piece by piece with the help of GPT4, the collocation dictionary and my own brain.
I am not sure what to do with it but in those cases I started rewriting the phrases piece by piece with the help of GPT4, the collocation dictionary and my own brain.
(Semantic circle about concepts, what the hell does it even mean.)
I think in simpler words I meant that synonyms move away from the original meaning plus the whole construction loses intended nuances all over.
(At professional orientation tests in school I got the highest scores at maths and linguistics, all the other fields trailed far behind.
I never studied linguistics academically but when I was learning 3 languages in 2014-2019 I self-taught myself applied linguistics in order to learn them in a grounded in factual evidence way.
3k people really liked my telegram channel about it. I think those of you who still read my ramblings here come from there. It is indeed a treasure of language learning knowledge. And some practical linguistic observations.)
I never studied linguistics academically but when I was learning 3 languages in 2014-2019 I self-taught myself applied linguistics in order to learn them in a grounded in factual evidence way.
3k people really liked my telegram channel about it. I think those of you who still read my ramblings here come from there. It is indeed a treasure of language learning knowledge. And some practical linguistic observations.)