File size: 4,123 Bytes
4d67ec7 | 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 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 | def build_answer_prompt(
question: str,
chapter: str,
section_title: str,
chunks: list[dict],
) -> list[dict]:
"""
Builds the Groq chat messages for the ANSWER generation call.
Returns a list of message dicts in OpenAI/Groq format:
[{"role": "system", ...}, {"role": "user", ...}]
"""
# Format retrieved chunks as numbered context blocks
context_blocks = ""
for i, chunk in enumerate(chunks, 1):
similarity_pct = int(chunk.get("similarity", 0) * 100)
context_blocks += (
f"[{i}] (relevance: {similarity_pct}%) "
f"{chunk.get('chapter', '')} > {chunk.get('section_title', '')}\n"
f"{chunk.get('text_content', '').strip()}\n\n"
)
system_prompt = (
"You are an expert NEET Biology teacher with deep knowledge of NCERT textbooks.\n"
"Your job is to answer the student's question using ONLY the provided NCERT context below.\n"
"Rules:\n"
" 1. Base your answer strictly on the NCERT context provided.\n"
" 2. Do NOT introduce facts not present in the context.\n"
" 3. If the context does not contain enough information, say: "
"'The provided NCERT context does not fully address this question.'\n"
" 4. Keep the answer concise, factually accurate, and at NEET exam level.\n"
f" 5. The question belongs to the chapter: '{chapter}', topic: '{section_title}'."
)
user_prompt = (
f"Question: {question}\n\n"
f"NCERT Context:\n{context_blocks}"
"Answer:"
)
return [
{"role": "system", "content": system_prompt},
{"role": "user", "content": user_prompt},
]
def build_justification_prompt(
question: str,
answer: str,
chunks: list[dict],
) -> list[dict]:
"""
Builds the Groq chat messages for the JUSTIFICATION generation call.
The LLM must explain WHY the answer is correct, citing context points.
"""
context_blocks = "\n".join(
f"[{i}] {c.get('text_content', '').strip()}"
for i, c in enumerate(chunks, 1)
)
system_prompt = (
"You are an expert NEET Biology teacher.\n"
"Given the student's question, the answer, and NCERT context passages, "
"provide a step-by-step biological justification for why the answer is correct.\n"
"Rules:\n"
" 1. Use ONLY the NCERT context to justify.\n"
" 2. Number each reasoning step.\n"
" 3. Be specific — reference biological terms and mechanisms.\n"
" 4. Do NOT add facts not found in the context."
)
user_prompt = (
f"Question: {question}\n\n"
f"Answer Given: {answer}\n\n"
f"NCERT Context:\n{context_blocks}\n\n"
"Step-by-step justification:"
)
return [
{"role": "system", "content": system_prompt},
{"role": "user", "content": user_prompt},
]
def build_verifier_prompt(
question: str,
answer: str,
chunks: list[dict],
) -> list[dict]:
"""
Builds the Groq chat messages for the VERIFIER call.
The LLM must output YES or NO on whether the answer is supported by context.
"""
context_blocks = "\n".join(
f"[{i}] {c.get('text_content', '').strip()}"
for i, c in enumerate(chunks, 1)
)
system_prompt = (
"You are a strict NEET Biology answer verifier.\n"
"Given a question, an answer, and NCERT context passages, "
"determine if the answer is fully supported by the context.\n"
"Respond with EXACTLY one of:\n"
" YES - if the answer is clearly supported by the NCERT context.\n"
" NO - if the answer contains facts not present in the context, or contradicts it.\n"
"Then on a new line, give a one-sentence explanation."
)
user_prompt = (
f"Question: {question}\n\n"
f"Answer: {answer}\n\n"
f"NCERT Context:\n{context_blocks}\n\n"
"Verdict (YES/NO):"
)
return [
{"role": "system", "content": system_prompt},
{"role": "user", "content": user_prompt},
]
|