Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2410.05661 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914967171629056 |
|---|---|
| author | Wang, Siqi Chen, Zhengyu Li, Bei He, Keqing Zhang, Min Wang, Jingang |
| author_facet | Wang, Siqi Chen, Zhengyu Li, Bei He, Keqing Zhang, Min Wang, Jingang |
| contents | The scaling of large language models (LLMs) is a critical research area for the efficiency and effectiveness of model training and deployment. Our work investigates the transferability and discrepancies of scaling laws between Dense Models and Mixture of Experts (MoE) models. Through a combination of theoretical analysis and extensive experiments, including consistent loss scaling, optimal batch size and learning rate scaling, and resource allocation strategies scaling, our findings reveal that the power-law scaling framework also applies to MoE Models, indicating that the fundamental principles governing the scaling behavior of these models are preserved, even though the architecture differs. Additionally, MoE Models demonstrate superior generalization, resulting in lower testing losses with the same training compute budget compared to Dense Models. These findings indicate the scaling consistency and transfer generalization capabilities of MoE Models, providing new insights for optimizing MoE Model training and deployment strategies. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_05661 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Scaling Laws Across Model Architectures: A Comparative Analysis of Dense and MoE Models in Large Language Models Wang, Siqi Chen, Zhengyu Li, Bei He, Keqing Zhang, Min Wang, Jingang Machine Learning Artificial Intelligence The scaling of large language models (LLMs) is a critical research area for the efficiency and effectiveness of model training and deployment. Our work investigates the transferability and discrepancies of scaling laws between Dense Models and Mixture of Experts (MoE) models. Through a combination of theoretical analysis and extensive experiments, including consistent loss scaling, optimal batch size and learning rate scaling, and resource allocation strategies scaling, our findings reveal that the power-law scaling framework also applies to MoE Models, indicating that the fundamental principles governing the scaling behavior of these models are preserved, even though the architecture differs. Additionally, MoE Models demonstrate superior generalization, resulting in lower testing losses with the same training compute budget compared to Dense Models. These findings indicate the scaling consistency and transfer generalization capabilities of MoE Models, providing new insights for optimizing MoE Model training and deployment strategies. |
| title | Scaling Laws Across Model Architectures: A Comparative Analysis of Dense and MoE Models in Large Language Models |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2410.05661 |