takumi123xxx/qwen3-0.6b-japanese-lora

Qwen3-0.6B Japanese LoRA Adapter

🤖 This model was fine-tuned using Claude Code - Anthropic's official CLI for Claude.

Model Description

This is a LoRA adapter for Qwen/Qwen3-0.6B, fine-tuned on Japanese instruction data.

Training Details

Parameter	Value
Base Model	Qwen/Qwen3-0.6B
Dataset	fujiki/japanese_alpaca_data
Method	LoRA (PEFT)
LoRA Rank (r)	8
LoRA Alpha	16
Target Modules	q_proj, v_proj
Trainable Parameters	1,146,880 (0.19%)
Training Steps	30 (test run)
Final Loss	2.27

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-0.6B")
model = PeftModel.from_pretrained(base_model, "takumi123xxx/qwen3-0.6b-japanese-lora")
tokenizer = AutoTokenizer.from_pretrained("takumi123xxx/qwen3-0.6b-japanese-lora")

Created With

🤖 Generated with Claude Code

This model was created as a demonstration of fine-tuning LLMs using Claude Code's NVIDIA GPU training MCP server integration.