Is this a common/reasonable recipe for full finetuning Qwen3.5-4B?

If other than AI… * Qwen3.5 Fine-tuning Guide | Unsloth Documentation * Qwen3.5 Best Practices — swift 4.1.0.dev0 documentation