config.py

MODEL_ID = "Qwen/Qwen2.5-Coder-0.5B-Instruct"
DATASET_ID = "gretelai/synthetic_text_to_sql"
OUTPUT_DIR = "./qwen-text-to-sql-lora"

MAX_TRAIN_SAMPLES = 20_000
MAX_VALIDATION_SAMPLES = 1_000
GENERATION_EVAL_SAMPLES = 50
SEED = 42

SYSTEM_PROMPT = """
You are an expert SQL generator.

Given a database schema/context and a natural-language request,
generate the SQL query that correctly answers the request.

Rules:
1. Use only tables and columns available in the provided database context.
2. Return SQL only.
3. Do not include Markdown code fences.
4. Do not explain the query.
""".strip()