randomtravellerai/mobile-game-test-plans-qwen2.5-coder-7b-lora

Mobile Game End-to-End Test Plan Generator

Fine-tuned Qwen2.5-Coder-7B-Instruct model specialized in writing comprehensive end-to-end test plans and test cases for mobile games.

What it does

Given a mobile game feature description, the model generates professional QA test plans including:

• Preconditions
• Step-by-step test steps
• Expected results
• Edge cases and boundary conditions

Training approach

Parameter	Value
Base model	Qwen/Qwen2.5-Coder-7B-Instruct
Method	Supervised Fine-Tuning (SFT) with LoRA
LoRA rank	16
LoRA alpha	32
LoRA dropout	0.1
Target modules	q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
Learning rate	3e-4
Epochs	3
Batch size	1 (effective = 8 via gradient accumulation)
Max sequence length	4096
Packing	True
Loss on assistant only	True

Dataset

• Size: ~720 training samples, ~80 test samples
• Source: Synthetic dataset covering 20 game genres and 15 feature categories
• Format: Conversational (messages with user / assistant roles)
• Coverage: RPG, Puzzle, Strategy, FPS, MOBA, Battle Royale, Runner, Simulation, Sports, Card Game, Tower Defense, Racing, Fighting, Adventure, Survival, MMORPG, Idle, Tycoon, Rhythm, Sandbox
• Feature categories: shop, combat, quest, guild, gacha, accessibility, notification, monetization, customization, level, account, settings, progression, social, bug, localization, privacy, competitive, story

How to use

Training

pip install transformers trl peft accelerate datasets
python train.py

Inference

from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

model_id = "randomtravellerai/mobile-game-test-plans-qwen2.5-coder-7b-lora"
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(model_id)

messages = [
    {"role": "user", "content": "Write an end-to-end test plan for this mobile RPG game feature:\nFeature: Player opens the shop from the main hub, browses weapon skins, previews them on their character, and purchases using in-game currency."}
]

pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
result = pipe(messages, max_new_tokens=1024)
print(result[0]["generated_text"])

Research basis

Training recipe adapted from:

• Parameter-Efficient Fine-Tuning of Large Language Models for Unit Test Generation (arXiv:2411.02462) — LoRA hyperparameters
• Enhancing Large Language Models for Text-to-Testcase Generation (arXiv:2402.11910) — prompt design and task framing

License

Apache-2.0 (matching base model)

Generated by ML Intern

This model repository was generated by ML Intern, an agent for machine learning research and development on the Hugging Face Hub.

• Try ML Intern: https://smolagents-ml-intern.hf.space
• Source code: https://github.com/huggingface/ml-intern