Parameter-efficient fine-tuning
Parameter-efficient fine-tuning adapts a pretrained model by training a limited subset of its parameters or added components rather than all its parameters.
LLMs in Production
Parameter-efficient fine-tuning is a fine-tuning approach that adapts a model by training only a small subset of parameters or added parameters.
- How it is used
- Positions PEFT as a way to adapt large models when fine-tuning memory and compute are constrained.
- How deep it goes
- Works through LoRA’s low-rank weight update and rank trade-off, but abstracts away a complete backpropagation tutorial.
- Examples
- Uses a 100 × 100 weight-update illustration and Python examples for biomedical token classification and BLOOMZ prompt tuning.
- What you can do after
- Apply it via the peft library using LoRA or prompt-tuning configurations for task-specific adaptation.
Read how LLMs in Production explains parameter-efficient fine-tuning →
The Art of AI Product Development
Parameter-efficient fine-tuning is a fine-tuning method that adapts a model by updating a small subset or additional parameterization rather than all model parameters.
- How it is used
- Supports repeated language-model customization when updating every parameter would make client-specific training costly or hard to scale.
- How deep it goes
- Defines PEFT by its limited parameter updates and compares its costs with full fine-tuning, without specifying an implementation variant.
- Examples
- Uses a topic-classification pilot with 92% accuracy and the need to maintain models for multiple client taxonomies.
- What you can do after
- Evaluate/understand it as a lower-cost fine-tuning option when selecting models and planning repeated customization.
Read how The Art of AI Product Development explains parameter-efficient fine-tuning →
Domain-Specific Small Language Models
A family of fine-tuning methods that adapt a pretrained model using only a small subset of parameters or added modules to reduce compute and cost.
- How it is used
- Adapts pretrained language models to specialized tasks or domains under tighter compute, hardware, and cost constraints.
- How deep it goes
- Explains the PEFT family, compares major methods, and works through LoRA training, saving, reloading, and inference.
- Examples
- Uses FLAN-T5-small, the SAMSum dataset, 8-bit loading, selected q and v modules, adapter checkpoints, and generated summaries.
- What you can do after
- Apply via Hugging Face Transformers and PEFT to train, save, reload, and evaluate a LoRA adapter.
Read how Domain-Specific Small Language Models explains parameter-efficient fine-tuning →
Financial AI in Practice
Parameter-efficient fine-tuning adapts a pretrained model by training a small subset of its parameters or adding lightweight trainable components.
- How it is used
- Adapts pretrained models to a domain-specific style or behavior while leaving factual retrieval to RAG.
- How deep it goes
- Defines PEFT, compares LoRA, adapters, and prefix tuning, and sketches a LoRA workflow without detailing every intricacy.
- Examples
- Uses 200 paired community posts and analyst-written summaries to teach the AlphaStream tone.
- What you can do after
- Apply PEFT to customize a model’s style at lower cost than full fine-tuning, with validation and version-transition planning.
Read how Financial AI in Practice explains parameter-efficient fine-tuning →
Building Reliable AI Systems
Fine-tuning methods that adapt a small subset or representation of a model’s parameters to reduce training resource requirements.
- How it is used
- Uses parameter-efficient updates when prompting and retrieval alone do not meet specialized-task needs.
- How deep it goes
- Explains LoRA’s frozen base and trainable adapters, then compares LoRA and QLoRA by parameter efficiency, performance, and memory.
- Examples
- Compares full fine-tuning, LoRA, and QLoRA across parameter share, performance, memory, and hardware choice.
- What you can do after
- Apply LoRA or QLoRA to fine-tune large models with reduced training-resource demands.
Read how Building Reliable AI Systems explains parameter-efficient fine-tuning →
Related concepts
- low-rank adaptationincludes · Low-rank adaptation is a parameter-efficient fine-tuning method that adapts a model by learning low-rank updates to its weights, typically while keeping most original weights fixed.
- fine-tuningkind of · Fine-tuning is adapting a pretrained machine-learning model by training it further on targeted data or objectives for a task, domain, or behavior.
- QLoRAincludes · QLoRA is a parameter-efficient fine-tuning method that quantizes a pretrained model’s weights while training low-rank adapters to reduce memory requirements.
- catastrophic forgettingaddresses · Catastrophic forgetting is the substantial loss of previously learned information or capabilities when a model learns new data, tasks, or objectives.
- DoRAincludes · DoRA is a parameter-efficient fine-tuning method that decomposes model weights into separately tunable magnitude and directional components, typically applying low-rank adaptation to the directional component.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.