A Practical JSONL Schema for Multi-Turn Instruction Tuning

Define a small, explicit schema that is easy to validate and transform. The useful boundary is a repeatable workflow with named inputs, explicit checks, and a release decision that another contributor can inspect. The examples below are illustrative procedure, not a claim about a particular customer or production result.

Set the operating boundary

Compare prompt/completion and message-array representations. Treat that requirement as a recorded engineering decision rather than an informal cleanup step. Start with a small, versioned input set and keep the original source identifier beside every transformed record. That lets a reviewer trace an unexpected output back to the data, configuration, or evaluation rule that produced it.

The implementation should separate deterministic checks from judgment calls. Deterministic checks can reject missing fields, malformed files, invalid ranges, and inconsistent identifiers. Judgment calls need a documented review rule, a sample of borderline cases, and a reason for keeping or excluding them. Mixing the two creates silent changes that are hard to reproduce later.

Required keys, allowed roles, stable IDs, and optional metadata

Define required keys, allowed roles, stable IDs, and optional metadata. Treat that requirement as a recorded engineering decision rather than an informal cleanup step. Start with a small, versioned input set and keep the original source identifier beside every transformed record. That lets a reviewer trace an unexpected output back to the data, configuration, or evaluation rule that produced it.

Record the result in a small manifest: source version, transformation version, input digest, output digest, reviewer, and timestamp. A compact manifest is more useful than a long narrative when a later run needs to be compared with this one.

Three valid records and common invalid records

Show three valid records and common invalid records. Treat that requirement as a recorded engineering decision rather than an informal cleanup step. Start with a small, versioned input set and keep the original source identifier beside every transformed record. That lets a reviewer trace an unexpected output back to the data, configuration, or evaluation rule that produced it.

Validate ordering, non-empty content, and duplicate IDs

Validate ordering, non-empty content, and duplicate IDs. Treat that requirement as a recorded engineering decision rather than an informal cleanup step. Start with a small, versioned input set and keep the original source identifier beside every transformed record. That lets a reviewer trace an unexpected output back to the data, configuration, or evaluation rule that produced it.

Convert the schema into a model-specific chat template at training time

Convert the schema into a model-specific chat template at training time. Treat that requirement as a recorded engineering decision rather than an informal cleanup step. Start with a small, versioned input set and keep the original source identifier beside every transformed record. That lets a reviewer trace an unexpected output back to the data, configuration, or evaluation rule that produced it.

Schema versioning without rewriting archived datasets

Explain schema versioning without rewriting archived datasets. Treat that requirement as a recorded engineering decision rather than an informal cleanup step. Start with a small, versioned input set and keep the original source identifier beside every transformed record. That lets a reviewer trace an unexpected output back to the data, configuration, or evaluation rule that produced it.

Practical check

JSONL examples and a standard-library Python validator. The following minimal check is intentionally small enough to run before a longer workflow. Expand it only after its output identifies the missing condition or failure category.

import json
import sys
from pathlib import Path

path = Path(sys.argv[1])
valid = 0
with path.open(encoding="utf-8") as handle:
    for line_number, line in enumerate(handle, 1):
        if not line.strip():
            raise SystemExit(f"line {line_number}: blank record")
        json.loads(line)
        valid += 1
print(f"validated {valid} records")

Review and release criteria

A result is ready for the next stage only when the required fields, split boundaries, and evaluation evidence agree. Keep a rejected sample set as well as accepted output; otherwise the same edge case will return as an unexplained regression. Treating one tokenizer chat template as portable to every base model. The correct response to uncertainty is to label the limitation and add a focused validation case, not to hide it behind a blended score.

Before publishing, verify the data digest, configuration, expected output shape, and the most important rare-case slice. The final registry note should state what changed, what was checked, and what remains outside the adapter's intended scope. That makes the next iteration a comparison instead of a guess.

A small review sample should include ordinary records, rare records, and inputs designed to fail cleanly. Each slice needs a stable identifier so that changes can be compared across releases without rebuilding the entire test set.