QuAILoRA: Quantization-Aware Initialization for LoRA
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866929550387052544 |
|---|---|
| author | Lawton, Neal Padmakumar, Aishwarya Gaspers, Judith FitzGerald, Jack Kumar, Anoop Steeg, Greg Ver Galstyan, Aram |
| author_facet | Lawton, Neal Padmakumar, Aishwarya Gaspers, Judith FitzGerald, Jack Kumar, Anoop Steeg, Greg Ver Galstyan, Aram |
| contents | QLoRA reduces the memory-cost of fine-tuning a large language model (LLM) with LoRA by quantizing the base LLM. However, quantization introduces quantization errors that negatively impact model performance after fine-tuning. In this paper we introduce QuAILoRA, a quantization-aware initialization for LoRA that mitigates this negative impact by decreasing quantization errors at initialization. Our method spends a small amount of computational overhead to compute this quantization-aware initialization, without increasing the memory-cost of fine-tuning. We evaluate our method on several causal language modeling and downstream evaluation tasks using several different model sizes and families. We observe that almost all LLMs fined-tuned with QuAILoRA achieve better validation perplexity. When evaluated on downstream tasks, we find that QuAILoRA yields improvements proportional to the negative effect of quantization error. On average, applying QuAILoRA to 4-bit QLoRA models yields 75% of the validation perplexity decrease and 86% of the downstream task accuracy increase as doubling the quantization precision to 8-bit, without increasing GPU memory utilization during fine-tuning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_14713 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | QuAILoRA: Quantization-Aware Initialization for LoRA Lawton, Neal Padmakumar, Aishwarya Gaspers, Judith FitzGerald, Jack Kumar, Anoop Steeg, Greg Ver Galstyan, Aram Machine Learning Computation and Language 68T50 QLoRA reduces the memory-cost of fine-tuning a large language model (LLM) with LoRA by quantizing the base LLM. However, quantization introduces quantization errors that negatively impact model performance after fine-tuning. In this paper we introduce QuAILoRA, a quantization-aware initialization for LoRA that mitigates this negative impact by decreasing quantization errors at initialization. Our method spends a small amount of computational overhead to compute this quantization-aware initialization, without increasing the memory-cost of fine-tuning. We evaluate our method on several causal language modeling and downstream evaluation tasks using several different model sizes and families. We observe that almost all LLMs fined-tuned with QuAILoRA achieve better validation perplexity. When evaluated on downstream tasks, we find that QuAILoRA yields improvements proportional to the negative effect of quantization error. On average, applying QuAILoRA to 4-bit QLoRA models yields 75% of the validation perplexity decrease and 86% of the downstream task accuracy increase as doubling the quantization precision to 8-bit, without increasing GPU memory utilization during fine-tuning. |
| title | QuAILoRA: Quantization-Aware Initialization for LoRA |
| topic | Machine Learning Computation and Language 68T50 |
| url | https://arxiv.org/abs/2410.14713 |