alebertoz/qwen3-0.6B-text2sql

Qwen3-0.6B Text-to-SQL (Manufacturing)

Fine-tuned Qwen3-0.6B for Text-to-SQL on manufacturing data (tissue paper plant).

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("alebertoz/qwen3-0.6B-text2sql")
tokenizer = AutoTokenizer.from_pretrained("alebertoz/qwen3-0.6B-text2sql")

messages = [
    {"role": "system", "content": "You are a SQL assistant..."},
    {"role": "user", "content": "What is the total production for REW machines?"}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Training

• Base model: Qwen/Qwen3-0.6B
• Method: SFT with QLoRA (4-bit)
• Framework: Unsloth + TRL
• Dataset: Custom manufacturing Text-to-SQL pairs

Database Schema

• newshiftstats: Shift-level metrics (machine_id, shift_start, produced_amount, oee_total, etc.)
• recipestats: Recipe-level metrics per shift
• All columns are TEXT - use CAST() for numeric operations