You can deploy this model and start earning money today!

TK0801dezi/
Qwen3-4B-Instruct-SFT-ver3

Name: Qwen3-4B-Instruct-SFT-ver3
Availability: InStock

u-10bei/structured_data_with_cot_dataset_512_v2

Qwen/Qwen3-4B-Instruct-2507

adapter

apache-2.0

Views

No views yet

Qwen3-4B-Instruct-SFT-ver3

This repository provides a LoRA adapter fine-tuned from Qwen/Qwen3-4B-Instruct-2507 using QLoRA (4-bit, Unsloth).

This repository contains LoRA adapter weights only. The base model must be loaded separately.

Training Objective

This adapter is trained to improve structured output accuracy (JSON / YAML / XML / TOML / CSV).

Loss is applied only to the final assistant output, while intermediate reasoning (Chain-of-Thought) is masked.

Training Configuration

Base model: Qwen/Qwen3-4B-Instruct-2507
Max sequence length: 512
Epochs: 2
Learning rate: 2e-6
LoRA: r=128, alpha=256
Optimization Strategy:
- Upsampling: Enabled (CSV: 1.97x, TOML: 1.76x, YAML: 1.72x)
- Learn Mode: from_marker

Usage

python

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
import torch

base = "Qwen/Qwen3-4B-Instruct-2507"
adapter = "TK0801dezi/Qwen3-4B-Instruct-SFT-ver3"

tokenizer = AutoTokenizer.from_pretrained(base)
model = AutoModelForCausalLM.from_pretrained(
    base,
    torch_dtype=torch.float16,
    device_map="auto",
)
model = PeftModel.from_pretrained(model, adapter)

Sources & Terms (IMPORTANT)

Training data: u-10bei/structured_data_with_cot_dataset_512_v2

Dataset License: MIT License. This dataset is used and distributed under the terms of the MIT License. Compliance: Users must comply with the MIT license (including copyright notice) and the base model's original terms of use.