Downloads · 30 days
9
13% of all-time downloads
yamaTK/qwen3-4b-lora-y-v42
qwen3-4b-lora-y-v42 is a text generation model from yamaTK. Use it when you need the model to write or continue text. It is set up for peft. The card lists the license as apache-2.0.
LoRA adapter fine-tuned from Qwen/Qwen3-4B-Instruct-2507 using QLoRA (4-bit, Unsloth).
Downloads · 30 days
9
13% of all-time downloads
All-time downloads
68
Public
Repo size
1.1 GB
Likes
0
Public
Click a slice to open those files.
.safetensors1.1 GB · 99%
From the Hugging Face model README
LoRA adapter fine-tuned from Qwen/Qwen3-4B-Instruct-2507 using QLoRA (4-bit, Unsloth).
This repository contains adapter weights only. Please load the base model separately.
Improve structured-output accuracy for JSON / YAML / XML / TOML / CSV.
Loss is applied only to the final assistant response. Intermediate reasoning (Chain-of-Thought) is masked during training.
Qwen/Qwen3-4B-Instruct-2507merged_dataset_final_clean_v41.jsonl51224.15e-060.0618478165770801540.00023098994964083745r=128, alpha=128, dropout=0.04q_proj,k_proj,v_proj,o_proj,gate_proj,up_proj,down_proj3248from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
import torch
base = "Qwen/Qwen3-4B-Instruct-2507"
adapter = "yamaTK/qwen3-4b-lora-y-v42"
tokenizer = AutoTokenizer.from_pretrained(base)
model = AutoModelForCausalLM.from_pretrained(
base,
torch_dtype=torch.float16,
device_map="auto",
)
model = PeftModel.from_pretrained(model, adapter)
Qwen/Qwen3-4B-Instruct-2507 をベースに、QLoRA(4-bit, Unsloth)で学習した LoRA アダプターです。
このリポジトリにはアダプター重みのみが含まれます。ベースモデルは別途読み込んでください。
JSON / YAML / XML / TOML / CSV の構造化出力精度を高めることを目的に学習しています。
学習時の損失は最終アシスタント出力のみに適用し、中間推論(Chain-of-Thought)はマスクしています。
Qwen/Qwen3-4B-Instruct-2507merged_dataset_final_clean_v41.jsonl51224.15e-060.0618478165770801540.00023098994964083745r=128, alpha=128, dropout=0.04q_proj,k_proj,v_proj,o_proj,gate_proj,up_proj,down_proj3248使い方は上記 Usage と同じです。