Planning vs Reasoning: Ablations to Test Capabilities of LoRA layers
Fuente:
arXiv
Guardado en:
| Autor principal: | |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917913106055168 |
|---|---|
| author | Redkar, Neel |
| author_facet | Redkar, Neel |
| contents | Low-Rank Adaptation (LoRA) layers have emerged as a promising approach for efficient model fine-tuning, but their capabilities and limitations have not been fully explored. This paper: 1) Investigates the fundamental question of whether LoRA layers are effective at increasing reasoning + planning abilities 2) We introduce HashChain Reasoning, a novel evaluation dataset that deterministically tests reasoning capabilities.
Through systematic ablation studies on GPT-2, we demonstrate that reasoning capabilities appear to exist primarily in low-rank spaces and can be effectively enhanced using LoRA layers. The effective rank analysis of trained LoRA matrices reveals a 2-3x lower rank requirement for reasoning tasks compared to planning tasks, giving context on where LoRA layers would be effective. This also provides evidence for reasoning fundamentally preferring low-parameter spaces for generalization. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_00029 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Planning vs Reasoning: Ablations to Test Capabilities of LoRA layers Redkar, Neel Artificial Intelligence I.2.7, I.2.6 Low-Rank Adaptation (LoRA) layers have emerged as a promising approach for efficient model fine-tuning, but their capabilities and limitations have not been fully explored. This paper: 1) Investigates the fundamental question of whether LoRA layers are effective at increasing reasoning + planning abilities 2) We introduce HashChain Reasoning, a novel evaluation dataset that deterministically tests reasoning capabilities. Through systematic ablation studies on GPT-2, we demonstrate that reasoning capabilities appear to exist primarily in low-rank spaces and can be effectively enhanced using LoRA layers. The effective rank analysis of trained LoRA matrices reveals a 2-3x lower rank requirement for reasoning tasks compared to planning tasks, giving context on where LoRA layers would be effective. This also provides evidence for reasoning fundamentally preferring low-parameter spaces for generalization. |
| title | Planning vs Reasoning: Ablations to Test Capabilities of LoRA layers |
| topic | Artificial Intelligence I.2.7, I.2.6 |
| url | https://arxiv.org/abs/2412.00029 |