Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

LISA: Layerwise Importance Sampling for Memory-Efficient Large Language Model Fine-Tuning

About

The machine learning community has witnessed impressive advancements since large language models (LLMs) first appeared. Yet, their massive memory consumption has become a significant roadblock to large-scale training. For instance, a 7B model typically requires at least 60 GB of GPU memory with full parameter training, which presents challenges for researchers without access to high-resource environments. Parameter Efficient Fine-Tuning techniques such as Low-Rank Adaptation (LoRA) have been proposed to alleviate this problem. However, in most large-scale fine-tuning settings, their performance does not reach the level of full parameter training because they confine the parameter search to a low-rank subspace. Attempting to complement this deficiency, we investigate the layerwise properties of LoRA on fine-tuning tasks and observe an unexpected but consistent skewness of weight norms across different layers. Utilizing this key observation, a surprisingly simple training strategy is discovered, which outperforms both LoRA and full parameter training in a wide range of settings with memory costs as low as LoRA. We name it Layerwise Importance Sampled AdamW (LISA), a promising alternative for LoRA, which applies the idea of importance sampling to different layers in LLMs and randomly freezes most middle layers during optimization. Experimental results show that with similar or less GPU memory consumption, LISA surpasses LoRA or even full parameter tuning in downstream fine-tuning tasks, where LISA consistently outperforms LoRA by over 10%-35% in terms of MT-Bench score while achieving on-par or better performance in MMLU, AGIEval and WinoGrande. On large models, specifically LLaMA-2-70B, LISA surpasses LoRA on MT-Bench, GSM8K, and PubMedQA, demonstrating its effectiveness across different domains.

Rui Pan, Xiang Liu, Shizhe Diao, Renjie Pi, Jipeng Zhang, Chi Han, Tong Zhang• 2024

Related benchmarks

TaskDatasetResultRank
Language ModelingWikiText-2 (test)--
1541
Mathematical ReasoningGSM8K (test)
Accuracy79.7
797
Multi-turn Dialogue EvaluationMT-Bench
Overall Score5.92
331
Commonsense ReasoningCommon Sense Reasoning Tasks
Avg Score80
241
Language UnderstandingMMLU (test)--
136
Language UnderstandingMMLU
Humanities Avg44.9
33
Instruction FollowingMT-Bench (test)
Overall Score5.92
27
Revision GenerationITERATER sent
SARI0.3982
23
Revision GenerationArgRevision
SARI34.38
23
Revision GenerationITERATER doc
SARI38.66
23
Showing 10 of 11 rows

Other info

Follow for update