指導 and the kanji 導 are both level 23, and なければならない is N4…
I’m pretty sure 〜ておる isn’t N5 either.
In fact, a quick google search says it’s N3.
Using the 連用形 instead of て-form (e.g. 望み in the first paragraph) is also N3 from a quick search.
I asked it to add furigana… I’m sure you could teach it the right syntax tho:)
ある日()、山田()さんと田中()さんは、リラックスした釣()り旅行()に行()くことにしました。天気()が晴()れで、風()も穏()やかだったので、最高()の釣()り日和()でした。
山田さんは、川()のほとりで木()の下()に座()りました。その間()、田中さんは竿()を持()って、静()かな池()に向()かいました。二人()は、それぞれの場所()で静()かに時間()を過()ごしていました。
釣()りを始()めると、山田さんはたくさんの魚()が泳()いでいるのを見()つけました。彼()は、魚(
The more I see ChatGPT get used as a resource the more I see no reason to use it as it’s so confidently wrong about so much.
Yeah initially I thought it was a great idea, but it clearly struggles in many areas. Will try to carefully change what I ask - but I can’t really ‘verify’ the results that well ^^.
However, I know this community is extremely nice and positive, the exact opposite of stack overflow. So if you guys are willing to help me out by checking some of the results, we might be able to build this thing)
I don’t want to talk too much and avoid lots of forths and backs tho, so see ya again after enough experimentation. Please leave any helpful tips here if you know any.
The goal is to have Chad GPT properly adhere to the limits. Imagine we could design it in a way where we can plug in any values! Like Duolingo level X, Genki Y, up to page 123 from Heising etc.
I agree, it’s not flawless, but the potential is fascinating
Tested it lots, it just really can’t properly read the items. Always returns the wrong items. Even when I just ask to return all radicals from level 1. But I can feed it a spreadsheet with a complete list and work from there.
Do we have one somewhere? Need something in XLS format
Well I think it has only data since a certain date, I don’t remember if it’s 2021? And Wanikani has changed some items since so that’s also a possible explanation
I thought that too, initially. But some generations had plain guesses including some radicals I couldn’t find at all. It refered to the striaght line called stick as dash for example. Never saw anyone/thing call the horizontal | radical a dash lol
Not sure if one already exists somewhere, but you could probably take one of the WaniKani Anki decks and feed it through an online anki to xls/xlsx converter.
Couldn’t find a way to make it fit the 2048 character limitation for Chat GPT, but for classic GPT3 or GPT4 I can pre-feed docs or use Plugins - already seeing some success with it
I used Wanikani Item Inspector to download all the kanji and vocabulary that I’ve learned (level 16). I also created another list of words/phrases that I want to know. Uploaded both into Chat GPT. It is able to create stories based on what I know so I can understand most of the stories. I also have it provide a list of words I don’t know that are used in the stories.
I’ve learned that I have forgotten some words, but I’m able to read more of these stories than other basic stories I was able to find.
Thought I should update my take on this since I am using it now. It is definitely stealing something from somewhere so let your moral compass guide you on whether or not it is worth using but it has been working for me.