File size: 2,810 Bytes
950121e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 | """tutor/model_loader.py — generates child-facing feedback (templates + optional LLM)."""
from __future__ import annotations
import os, random
from typing import Optional
_CORRECT = {
"en": ["Amazing work! 🎉 You got it!","Superstar! ⭐ That's exactly right!",
"Brilliant! 🌟 You're on fire!","Perfect! Keep it up! 🚀","You nailed it! 🎊"],
"fr": ["Bravo ! 🎉 C'est exactement ça !","Super ! ⭐ Tu es fantastique !",
"Excellent ! 🌟 Continue comme ça !","Parfait ! 🥳 Tu es une étoile !"],
"kin": ["Byiza cyane! 🎉 Wabitsinze!","Ntangaza! ⭐ Uravuga ukuri!",
"Komeza! 🌟 Uri inzobere!","Superbe! 🥳 Urashoboye!"],
"sw": ["Vizuri sana! 🎉 Umeshinda!","Hongera! ⭐ Jibu sahihi!",
"Endelea hivyo! 🌟 Umefaulu!","Bora sana! 🥳 Ushindi!"],
}
_WRONG = {
"en": ["Almost! Give it another try 💪","Not quite — you can do it! 🌈",
"Good try! Think again 🤔","Keep going — try once more! 💫"],
"fr": ["Presque ! Essaie encore 💪","Pas tout à fait — tu peux le faire ! 🌈",
"Bon essai ! Réfléchis encore 🤔"],
"kin": ["Hafi! Gerageza nanone 💪","Ntabwo ari byo — gerageza! 🌈","Gerageza nanone! 🤔"],
"sw": ["Karibu! Jaribu tena 💪","Si sahihi — unaweza! 🌈","Jaribu tena! 🤔"],
}
_sdk_checked = False
_sdk_ok = False
def _check_sdk():
global _sdk_checked, _sdk_ok
if _sdk_checked:
return _sdk_ok
_sdk_checked = True
try:
import anthropic # type: ignore # noqa
_sdk_ok = bool(os.environ.get("ANTHROPIC_API_KEY"))
except ImportError:
pass
return _sdk_ok
def _llm_feedback(is_correct, answer, lang, child_text) -> Optional[str]:
if not _check_sdk():
return None
try:
import anthropic # type: ignore
client = anthropic.Anthropic()
outcome = "correct" if is_correct else "incorrect"
msg = client.messages.create(
model="claude-haiku-4-5-20251001", max_tokens=60,
messages=[{"role":"user","content":
f"You are a warm math tutor for a child aged 5-9. "
f"Child answered '{child_text}', correct answer is {answer}, "
f"their answer was {outcome}. Reply in language '{lang}' "
f"with ONE short encouraging sentence (max 12 words)."}])
return msg.content[0].text.strip()
except Exception:
return None
def generate_feedback(is_correct: bool, answer: int,
lang: str = "en", child_text: str = "") -> str:
llm = _llm_feedback(is_correct, answer, lang, child_text)
if llm:
return llm
bank = _CORRECT if is_correct else _WRONG
return random.choice(bank.get(lang, bank["en"]))
|