dapt-qwen3-4b-thai-law-lora

Thai Legal Domain-Adapted Qwen3-4B-Thinking

This model is a domain-adapted version of Qwen/Qwen3-4B-Thinking-2507 fine-tuned on Thai legal documents including:

• ประมวลกฎหมายแพ่งและพาณิชย์ (Civil and Commercial Code)
• ประมวลรัษฎากร (Tax Revenue Code)
• Court case judgments

Training Details

• Base Model: Qwen/Qwen3-4B-Thinking-2507
• Training Method: LoRA (Low-Rank Adaptation) with Unsloth
• Sequence Length: 2048
• Learning Rate: 1e-05
• Epochs: 1

Usage

from unsloth import FastLanguageModel

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name="kamonwit/qwen3-4b-thai-law-dapt",
    max_seq_length=2048,
    dtype=None,
    load_in_4bit=True,
)

# Enable inference mode
FastLanguageModel.for_inference(model)

# Generate text
inputs = tokenizer("ตามประมวลกฎหมายแพ่งและพาณิชย์", return_tensors="pt")
outputs = model.generate(**inputs, max_length=200, do_sample=True, temperature=0.7)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Note

This is a LoRA adapter that needs to be loaded with the base model. For standalone usage, use the merged version if available.

Framework versions

• PEFT 0.17.1