467d9d1d7c
Three unit-1.1 lessons with gpt-oss-20b through LM Studio and the server —
the first real model on the reworked turn. The letter-level check went out
right, and the +25 clamp held: a reported 80 on the first answer was stored
as 25. What failed was how the model wrote its blocks, a different way each
session. All three transcripts are in test/fixtures/, verbatim, and each
failure below is a test against them.
Marks lost. The prompt asks for `여덟 | wrong | 여덜`. The first session wrote
`we | wrong | 우라 → 우리`, English prompt first; the second wrote no ::result
at all and marked only in prose, `✗ 나 | I (humble) → 저`. evidence.ts keys on
the first field of a ::result row, so nothing was ever recorded — no
evidence, no schedule, no confusions, and a 다지기 review that could never
close. The artifact would have lost them the same way. domain/marking.ts
attaches each mark to its word only where that is unambiguous: one Korean
word first, or through a prompt of the exercise he answered, read via that
exercise's ::words as the letter check reads it. With no ::result block the
✓/✗ lines are read on the same terms, so a mark can never name a word the
exercise did not ask for; a mark on a whole sentence is still dropped. What
he mistook a word for is taken from what he actually wrote whenever the mark
itself gives no other word — the third session put the right answer there.
Its third session, marked through all of this: 20 evidence rows, 20 cards.
Progress on requests. The prompt allows marks, ::confirmed and ::progress
only in reply to an answer. The model wrote ::progress on every message, and
three requests for a new exercise took the unit from 50% to 80% with nothing
answered. A reply to anything but an answer now changes none of them.
Feedback swallowed. The model closed no blocks, so lib read what followed
each one as rows: "your score is about 5%" became a result row the student
never saw, and a "---" became a recall item he was asked to write in 한글.
Another session fenced every block in ```. gloss.ts now decides every
block's extent from the raw text — at "::", the next block, a rule or fence
line, a blank line with no row after it, or for the piped blocks the first
line without a "|" — and hands lib the blocks properly closed. The gate
audit, now run through the parser the lesson uses, still flags 7 and 2.
Answers given away. Translate rows came with their meanings ("나 | I") and
recall hints were the answers ("two | 이"). A translate row keeps only its
Korean line, and a recall hint that is the expected word, or any word the
message declares, is dropped.
Also: the spelling a recall prompt expects now keeps its qualifiers. With 나
"I, me (casual)" and 저 "I, me (humble)" in one list, "I (humble)" matched
나: no letter check was sent, and the mark for 저 was filed under 나.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
9 lines
3.9 KiB
JSON
9 lines
3.9 KiB
JSON
{
|
||
"source": "gpt-oss-20b (LM Studio) through server/src/backends/openai.ts, unit 1.1, 16 Sep 2026. Verbatim.",
|
||
"answered": "**Unit 1.1 자음과 모음 – Consonants & vowels**\n\nThe first thing you’ll learn is how Korean letters are put together into *blocks* (or syllables). \nEvery block has two parts: an initial consonant (the *기초 자음*) and a vowel. \nWe will only use the plain, single‑letter consonants that can start a block:\n\nㄱ ㄴ ㄷ ㄹ ㅁ ㅂ ㅅ ㅇ ㅈㅎ\n\nand the ten basic vowels:\n\nㅏ ㅑ ㅓ ㅕ ㅗ ㅛ ㅜ ㅠ ㅡ ㅣ\n\nNo final consonant will appear yet, so each block looks like “ㄱ + ㅏ” → **가**, “ㅇ + ㅔ” → **여**.\n\nBecause the syllable is just a pair of letters, you can read it by looking at the shape: \nthe initial consonant on the left and the vowel on the right. \nIf you see ㄴ and ㅓ together, you know it’s **너**; if you see ㅇ and ㅜ together, it’s **우리**.\n\nThe goal of this unit is to recognize these blocks and write them out. \nYou’ll practice by reading Korean words that contain only one block (no final consonant) and writing the English meaning or the Korean word itself.\n\n---\n\n### First exercise – *Recall* \n\nWrite the Korean syllables for each English prompt. \n(Use the new words listed in this unit.)\n\n::task recall\nI | 나 \nyou | 너 \nwe | 우리 \nthere | 거기 \none | 하나\n\n---\n\n::words\n나 | I | (plain consonant ㄴ + vowel ㅏ) \n너 | you | (ㄴ + ㅓ) \n우리 | we | (ㅇ + ㅜ) \n거기 | there | (여(ㅇ+ㅕ), 기(ㄱ+ㅣ)) \n하나 | one | (ㅎ + ㅏ)\n\n---\n\n*You have just started. Keep practicing the shapes of these syllables, and you’ll be ready to read more Korean words soon.*",
|
||
"answer": "My written answers:\nI → 나\nyou → 너\nwe → 우라\nthere → 거기\none → 하나\n--- → 나\n\n════ LETTER-LEVEL CHECK — computed by the app ════\nThis is the actual jamo comparison. Use it exactly. Do NOT work out for yourself which letter was wrong, and never call a letter wrong that is listed as correct.\n• we\n wanted 우리 [우=ㅇ+ㅜ · 리=ㄹ+ㅣ]\n wrote 우라 [우=ㅇ+ㅜ · 라=ㄹ+ㅏ]\n syllable 2 (라 for 리) — WRONG: vowel: wrote ㅏ, should be ㅣ — CORRECT, do not call these mistakes: first consonant ㄹ, batchim none\n\n(No lookups.)",
|
||
"marking": "You did most of them right, just a couple of small mistakes:\n\n::result\nI | ok | 나 \nyou | ok | 너 \nwe | wrong | 우라 → 우리 \nthere | ok | 거기 \none | ok | 하나\n\nYou wrote an extra line \"--- → 나\" that didn’t match any prompt – you can ignore it.\n\nThe first time you’ve answered, your score for this unit is about **5 %**. Keep practicing the shapes of these syllables; we’ll add a few more words next.\n\n---\n\n### Next exercise – *Recall* \n\nWrite the Korean syllables for each English prompt. \n(Use the new words listed in this unit.)\n\n::task recall\ntwo | 이 \nthat | 그 \nI (humble) | 저\n\n---\n\n::words\n나 | I | (ㄴ + ㅏ) \n너 | you | (ㄴ + ㅓ) \n우리 | we | (ㅇ + ㅜ) \n거기 | there | (여(ㅇ+ㅕ), 기(ㄱ+ㅣ)) \n하나 | one | (ㅎ + ㅏ) \n우라 | wrong answer for 우리 | (ㅇ + ㅜ, 라(ㄹ+ㅏ)) \n이 | two | (ㅇ + ㅣ) \n그 | that | (ㄱ + ㅡ) \n저 | I (humble) | (ㅈ + ㅓ)\n\n::progress 5 %",
|
||
"asked": "Give me a translate exercise that uses the words 도서관 and 병원.",
|
||
"reply": "**Translate – read the Korean, type the English meaning**\n\nHere are simple words that contain only one block each (no final consonant). \n\n::task translate\n나 | I \n너 | you \n우리 | we \n거기 | there \n하나 | one\n\n---\n\n::words\n나 | I | (ㄴ + ㅏ) \n너 | you | (ㄴ + ㅓ) \n우리 | we | (ㅇ + ㅜ) \n거기 | there | (여(ㅇ+ㅕ), 기(ㄱ+ㅣ)) \n하나 | one | (ㅎ + ㅏ)\n\n---\n\n*Try to read each Korean word and write the English translation. Good luck!*"
|
||
}
|