Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | https://arxiv.org/abs/2406.06623 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866909220894408704 |
|---|---|
| author | Hartford, Eric Atkins, Lucas Neto, Fernando Fernandes Golchinfar, David |
| author_facet | Hartford, Eric Atkins, Lucas Neto, Fernando Fernandes Golchinfar, David |
| contents | Efficiently post-training large language models remains a challenging task due to the vast computational resources required. We present Spectrum, a method that accelerates LLM training by selectively targeting layer modules based on their signal-to-noise ratio (SNR), and freezing the remaining modules. Our approach, which utilizes an algorithm to compute module SNRs prior to training, has shown to effectively match the performance of full fine-tuning while reducing GPU memory usage. Experiments comparing Spectrum to existing methods such as QLoRA demonstrate its effectiveness in terms of model quality and VRAM efficiency in distributed environments. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_06623 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Spectrum: Targeted Training on Signal to Noise Ratio Hartford, Eric Atkins, Lucas Neto, Fernando Fernandes Golchinfar, David Machine Learning Efficiently post-training large language models remains a challenging task due to the vast computational resources required. We present Spectrum, a method that accelerates LLM training by selectively targeting layer modules based on their signal-to-noise ratio (SNR), and freezing the remaining modules. Our approach, which utilizes an algorithm to compute module SNRs prior to training, has shown to effectively match the performance of full fine-tuning while reducing GPU memory usage. Experiments comparing Spectrum to existing methods such as QLoRA demonstrate its effectiveness in terms of model quality and VRAM efficiency in distributed environments. |
| title | Spectrum: Targeted Training on Signal to Noise Ratio |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2406.06623 |