Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2503.22329 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909556031881216 |
|---|---|
| author | Owen, Louis Chowdhury, Nilabhra Roy Kumar, Abhay Güra, Fabian |
| author_facet | Owen, Louis Chowdhury, Nilabhra Roy Kumar, Abhay Güra, Fabian |
| contents | Motivated in part by their relevance for low-precision training and quantization, massive activations in large language models (LLMs) have recently emerged as a topic of interest. However, existing analyses are limited in scope, and generalizability across architectures is unclear. This paper helps address some of these gaps by conducting an analysis of massive activations across a broad range of LLMs, including both GLU-based and non-GLU-based architectures. Our findings challenge several prior assumptions, most importantly: (1) not all massive activations are detrimental, i.e. suppressing them does not lead to an explosion of perplexity or a collapse in downstream task performance; (2) proposed mitigation strategies such as Attention KV bias are model-specific and ineffective in certain cases. We consequently investigate novel hybrid mitigation strategies; in particular pairing Target Variance Rescaling (TVR) with Attention KV bias or Dynamic Tanh (DyT) successfully balances the mitigation of massive activations with preserved downstream model performance in the scenarios we investigated. Our code is available at: https://github.com/bluorion-com/refine_massive_activations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_22329 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | A Refined Analysis of Massive Activations in LLMs Owen, Louis Chowdhury, Nilabhra Roy Kumar, Abhay Güra, Fabian Computation and Language Motivated in part by their relevance for low-precision training and quantization, massive activations in large language models (LLMs) have recently emerged as a topic of interest. However, existing analyses are limited in scope, and generalizability across architectures is unclear. This paper helps address some of these gaps by conducting an analysis of massive activations across a broad range of LLMs, including both GLU-based and non-GLU-based architectures. Our findings challenge several prior assumptions, most importantly: (1) not all massive activations are detrimental, i.e. suppressing them does not lead to an explosion of perplexity or a collapse in downstream task performance; (2) proposed mitigation strategies such as Attention KV bias are model-specific and ineffective in certain cases. We consequently investigate novel hybrid mitigation strategies; in particular pairing Target Variance Rescaling (TVR) with Attention KV bias or Dynamic Tanh (DyT) successfully balances the mitigation of massive activations with preserved downstream model performance in the scenarios we investigated. Our code is available at: https://github.com/bluorion-com/refine_massive_activations. |
| title | A Refined Analysis of Massive Activations in LLMs |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2503.22329 |