Small Models are Enough

• Small models are good enough, often for repetitive or low-intelligence tasks. (small_good_enough, 19)
• A fine-tuned small model can be better than a big generalist untuned model. (small_fine_tuned, 3)
• Some think and prefer that small models have less “alignment” than larger models (small_less_alignment, 2)
• If you break a task into chunks, small models can become very efficient. Related to Model Optimization Strategies. (small_chunk_work, 2)