Hard

Implement a synthetic-data pipeline: generate candidate examples, validate them with a rul

SFT, Instruction Tuning, Data & PEFT · Problem 4 of 6

Chapter 08SFT, Instruction Tuning, Data & PEFT

Implement a synthetic-data pipeline: generate candidate examples, validate them with a rul

HardProblem 4 / 6

Implement a synthetic-data pipeline: generate candidate examples, validate them with a rule-based checker, deduplicate, and report the resulting source mixture.

Implement the function/class skeleton in the editor. Any correct approach is accepted.

Hints

solution.pypython

local draft

import hashlib, re
from collections import Counter

def normalize(text):
    raise NotImplementedError

def prompt_hash(prompt):
    raise NotImplementedError

def rule_checker(ex):
    raise NotImplementedError

def build_dataset(generators):
    raise NotImplementedError

⌘/Ctrl + ↵ to submit

AI review

Ready when you are

Submit your solution and a structured review appears here — verdict, score, and concrete feedback. Any correct approach passes.

Chapter 08SFT, Instruction Tuning, Data & PEFT

Implement a synthetic-data pipeline: generate candidate examples, validate them with a rul

HardProblem 4 / 6

Implement a synthetic-data pipeline: generate candidate examples, validate them with a rule-based checker, deduplicate, and report the resulting source mixture.

Implement the function/class skeleton in the editor. Any correct approach is accepted.

Hints