Scholar Feed

Parameter-efficient fine-tuning (LoRA family)

FFA-LoRA

Superseded baseline#7 of 1,113 most-superseded

Superseded — cited as a baseline and beaten by newer methods

9 papers critique it · 17 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites FFA-LoRA as a baseline.

FFA-LoRA~sunimproving freezes one LoRA matrix to suppress cross-client interference. While effective, these methods trade off scalability, flexibility, or representational capacity.
— ADF-LoRA: Alternating Low-Rank Aggregation for Decentralized Federated Fine-Tuning
“However, using a fixed random matrix for $A$ limits the learning capability of LoRA, and we observe that optimizing only $B$ leads to significantly slower convergence.”
— FedSVD: Adaptive Orthogonalization for Private Federated Learning with LoRA
“Notably, combining direct weight aggregation with GaLore as an optimizer for local training steps significantly outperforms leading state-of-the-art LoRA methods like FlexLoRA and FFA-LoRA.”
— Exploring Gradient Subspaces: Addressing and Overcoming LoRA's Limitations in Federated Fine-Tuning of Large Language Models
“FFA-LoRA sun2024improving manages exact updates by freezing the $A$ parameter at initialization but reduces the model expressivity relative to vanilla LoRA.”
— Ravan: Multi-Head Low-Rank Adaptation for Federated Fine-Tuning
“However, we observe that with fewer finetuning parameters, FFA-LoRA is less robust than FedAVG of LoRA, primarily due to its reduced expressiveness from freezing down-projections.”
— Robust Federated Finetuning of LLMs via Alternating Optimization of LoRA
“Federated Freeze A LoRA (FFA-LoRA) sun2024improving mitigates this by keeping one set of adapters trainable, improving aggregation stability but limiting the training flexibility of other adapters.”
— FedEx-LoRA: Exact Aggregation for Federated and Efficient Fine-Tuning of Foundation Models
“However, freezing non-zero matrices will hinder the model from converging to a good local minimum, since random initialization is nearly impossible to produce optimal parameters for downstream tasks.”
— DEeR: Deviation Eliminating and Noise Regulating for Privacy-preserving Federated Low-rank Adaptation
“because certain matrices remain fixed, LoRA's capacity to adapt is limited, which often leading to suboptimal performance”
— Communication-Efficient and Accurate Approach for Aggregation in Federated Low-Rank Adaptation
“However, we observe that with fewer fine-tuning parameters, FFA-LoRA is less robust than FedAVG for LoRA modules, primarily due to its limited expressiveness stemming from the restricted number of trainable parameters.”
— Robust Federated Finetuning of Foundation Models via Alternating Minimization of LoRA

Beaten on benchmarks

Head-to-head results where a newer method reports beating FFA-LoRA. Values are copied from the source paper's tables — verify against the cited paper.

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.