gensai.aiLoRA is a lightweight fine-tuning method. Instead of changing the billions of weights in a model, it trains small additional matrices, adapters of a few megabytes, that sit on top of the original model without altering it. The same model can then be specialised for several uses at a much lower computing cost.
In practice at Gensai
LoRA is Gensai's preferred option when an open-weight model needs to adopt the vocabulary or tone of a profession without full retraining.
Same theme
LLM (Large Language Model)SLM (Small Language Model)DLLM (Diffusion Large Language Model)Hallucination (AI)Prompt chainingPre-trained modelFoundation modelGenerative AITransformerTokenContext windowPromptPrompt engineeringContext (AI)Memory (AI)Fine-tuningOpen-weightQuantizationDistillationMixture of Experts (MoE)Reasoning modelTemperature (AI)