TToyo2511/ttoyo_advance_2c18

Qwen3-4B Agent SFT for ALFWorld & DBBench TrainingData+ 20260228-2c18

This repository provides a LoRA adapter fine-tuned from Qwen/Qwen3-4B-Instruct-2507 using LoRA + Unsloth.

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

Training Objective

This adapter is trained to improve multi-turn agent task performance on ALFWorld (household tasks) and DBBench (database operations).

Loss is applied to all assistant turns in the multi-turn trajectory, enabling the model to learn environment observation, action selection, tool use, and recovery from errors.

Training Configuration

• Base model: Qwen/Qwen3-4B-Instruct-2507
• Method: LoRA (full precision base)
• Max sequence length: 2048
• Epochs: 2
• Learning rate: 5e-06
• LoRA: r=128, alpha=128

Key Changes in 2c18 (Reverse Direction from 2c17)

2c17で悪化した変更の逆方向を試行:

1. LR: 5.5e-6 → 5.2e-6（2c17で5.7e-6に上げて悪化したので下げる）

2. Differentiated upsampling (reverse direction):

   • ALF samples: 85x → 82x（2c17で88xに上げたので下げる）
   • DB samples: 55x → 58x（2c17で52xに下げたので上げる）

3. Sample composition: 2c14と完全同一を維持

   • ALF: 8 samples
   • DB: 5 samples
   • DBBench data: 5%

Usage

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

base = "Qwen/Qwen3-4B-Instruct-2507"
adapter = "TToyo2511/ttoyo_advance_2c18" #★TTT20260228 2c18版

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/sft_alfworld_trajectory_dataset_v5
• u-10bei/dbbench_sft_dataset_react_v4 (5% sampled)

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.