qwen2.5-7b-alpaca

Fine-tuned from Qwen/Qwen2.5-7B-Instruct on the alpaca (51390 rows) dataset.

Intended use

General-purpose instruction-following assistant fine-tuned on the Alpaca dataset. Suitable for text generation, question answering, and instruction-following tasks.

Training details

• Method: QLoRA (LoRA rank 32, NF4 base weights, bfloat16 compute)
• Hardware: NVIDIA GeForce RTX 5090 32GB
• Duration: Pipeline validation run (200 steps)
• Dataset: alpaca (51390 rows)
• Base model: Qwen/Qwen2.5-7B-Instruct
• Final val loss: 2.322
• Trained by: Hika — MKD Co., Ltd.

Evaluation

Full report: eval_results/run_2026-05-27_0039/summary.md

Eval	Base	This model	Δ
MMLU	pending	pending	pending
MT-Bench	pending	pending	pending

Bias, risks, limitations

This model may produce incorrect, biased, or harmful content. Not intended for medical, legal, or safety-critical applications. Users should verify outputs independently.

License

Inherits license of Qwen/Qwen2.5-7B-Instruct.

Citation

MKD Co., Ltd. — Fine-Tuning Platform (2026)