Dataset Card for GPT-5.4-Reasoning-1500-Ultra-Logic
Dataset Details
Dataset Description
Suggestion: I would use this to fine-tune qwen3.5 35b a3b moe, or 27b variant. However, for maximum efficiency, 2bb-20b LLMs like qwen3.5 9b and 4b, gpt-oss 20b work perfectly. Fine-tuning the newest versions (specialized reasoning variants) will yield the most significant logic jumps.
