On the Scaling of PEFT: Towards Million Personal Models of Trillion Parameters
This paper reframes parameter-efficient fine-tuning (PEFT) not merely as a cheaper alternative to full fine-tuning, but as a substrate for persistent, instance-specific personal models layered atop shared foundation models. The authors analyze three scaling axes: Scale Up (stronger base models amplifying adapter utility), Scale Down (minimum viable adapter size), and Scale Out (managing millions of concurrent adapted instances). They introduce MinT as an infrastructure reference for adapter identity, versioning, provenance, evaluation, and serving at scale.
Related guides (4)
Related events (8)
Parameter-Efficient Fine-Tuning using 🤗 PEFT
Hugging Face introduces the PEFT library, which enables parameter-efficient fine-tuning of large language models using techniques such as LoRA, prefix tuning, and prompt tuning. The library allows practitioners to adapt large pretrained models to downstream tasks while updating only a small fraction of model parameters, dramatically reducing compute and memory requirements. This lowers the barrier to fine-tuning frontier-scale models on consumer hardware.
PEFT-Arena: Benchmarking Parameter-Efficient Finetuning via Stability-Plasticity Trade-offs
PEFT-Arena is a new benchmark that evaluates parameter-efficient finetuning methods jointly on downstream task performance and retention of pretrained general capabilities, framing the problem as a stability-plasticity dilemma. Across methods tested under comparable parameter budgets, orthogonal finetuning achieves the best Pareto frontier. The paper provides geometric analyses in both weight space (spectral/singular-value structure) and activation space (representation distortion metrics) to explain why different PEFT methods differ in forgetting behavior. A practical finding is that final SFT checkpoints often overshoot an optimal retention operating point, motivating path-wise rewinding as a post-hoc correction.
🤗 PEFT Welcomes New Merging Methods
Hugging Face's PEFT library has added new methods for merging parameter-efficient fine-tuned adapters (e.g., LoRA). The update enables combining multiple fine-tuned adapters into a single model, expanding the toolkit for practitioners working with adapter-based fine-tuning. This is a tooling update relevant to the growing ecosystem of efficient fine-tuning and model composition workflows.
Complete-muE: Optimal Hyperparameter Transfer and Scaling for MoE Models
Complete-muE is a framework for transferring hyperparameters across dense FFN and Mixture-of-Experts (MoE) transformer architectures, addressing limitations of existing tools like μP and SDE that cannot handle simultaneous architecture and token-per-expert changes. It uses a two-bridge system: Bridge I maps dense FFN to Dense MoE via active-width μP with normalized router scale, and Bridge II maps Dense MoE to sparse MoE via activated-expert scaling with a first-order SDE correction. The practical outcome is a 'tune dense once, transfer to all' recipe that enables near-optimal hyperparameter reuse across MoE configurations without costly re-tuning. Experiments on language model and diffusion model pretraining confirm stable hyperparameter optima across architectures and parameter counts.
ChunkFT: Memory-Efficient Full Fine-Tuning via Byte-Streamed Chunk Optimization
ChunkFT is a fine-tuning framework that reformulates full-parameter optimization around a dynamically activated working set of sub-tensors, enabling gradient computation without dense gradient materialization. It achieves full-parameter fine-tuning of a 7B model in 13.72GB GPU memory on a single RTX 4090, and scales Llama 3-70B fine-tuning to 2×H800 GPUs. Downstream evaluations on language understanding, math reasoning, and MT-Bench show ChunkFT matches or exceeds full-parameter fine-tuning quality while outperforming existing memory-efficient baselines such as LoRA-class methods. A theoretical convergence analysis in the deterministic setting is also provided.
MobileMoE: Scaling Mixture-of-Experts for Sub-Billion Parameter On-Device Deployment
MobileMoE introduces a family of on-device MoE language models with 0.3–0.9B active parameters and 1.3–5.3B total parameters, targeting mobile deployment under memory and compute constraints. The authors derive an on-device MoE scaling law identifying a sweet spot of moderate sparsity with fine-grained and shared experts, then train models through a four-stage recipe including quantization-aware training on open-source data. Across 14 benchmarks, MobileMoE matches or exceeds leading dense on-device LLMs with 2–4× fewer inference FLOPs, and delivers 1.8–3.8× faster prefill and 2.2–3.4× faster decode than dense baselines on commodity smartphones at comparable INT4 memory.
Fine-tuning 20B LLMs with RLHF on a 24GB consumer GPU
Hugging Face demonstrates a method for running RLHF fine-tuning on 20-billion-parameter language models using a single 24GB consumer GPU by combining TRL and PEFT (parameter-efficient fine-tuning). The approach uses techniques like LoRA and quantization to dramatically reduce memory requirements. This lowers the hardware barrier for RLHF experimentation from multi-GPU server setups to consumer-grade hardware.
Quantifying Hyperparameter Transfer: Embedding Layer Learning Rate as Key Driver of μP Benefits
This paper develops a three-metric framework to quantify hyperparameter transfer quality across model scales, targeting the problem of extrapolating optimal hyperparameters from small to large LLMs. The central empirical finding is that the well-known advantage of Maximal Update Parameterization (μP) over standard parameterization (SP) with AdamW largely reduces to a single factor: the embedding layer learning rate. In SP, the embedding layer acts as a training bottleneck causing instabilities; scaling its learning rate by model width to match μP substantially stabilizes training and improves transfer. The paper also characterizes how weight decay affects scaling law fit quality versus extrapolation robustness in opposite directions.