The trap with fine-tuning has always been that the learning lives inside one model's weights. With PorTAL, we pulled the task representation out of the base entirely. Learn it once, and porting to a new model is just refitting a thin converter, no retraining from scratch. It turns your fine-tunes into a memory bank that plugs into whatever model is best. Saving time and $$$
もっと見る
Introducing PorTAL: Portable Task Adapters for LLMs. A novel recipe to cheaply port fine-tuning between models. It matches per task LoRA accuracy at half the cost, lowering the switching overhead of adapting tasks across LLMs.
At Ramp, every new model release used to mean retraining our fine-tunes from scratch. PorTAL learns the task once, then efficiently refits it onto any new base model, even across model families.
もっと見る
