Is It a Free Lunch for Removing Outliers during Pretraining?
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_ | 1866910335970050048 |
|---|---|
| author | Liao, Baohao Monz, Christof |
| author_facet | Liao, Baohao Monz, Christof |
| contents | With the growing size of large language models, the role of quantization becomes increasingly significant. However, outliers present in weights or activations notably influence the performance of quantized models. Recently, \citet{qtransformer} introduced a novel softmax function aimed at pretraining models in an outlier-free manner, thereby enhancing their suitability for quantization. Interestingly, we observed that such an approach leads to performance degradation in full precision. Building on this insight, we enhance the method by ensuring its normalization is invariant to sequence length, a crucial factor for bridging the gap between pretraining and fine-tuning. Moreover, this improved method also facilitates successful pretraining of causal language models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_12102 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Is It a Free Lunch for Removing Outliers during Pretraining? Liao, Baohao Monz, Christof Computation and Language Artificial Intelligence With the growing size of large language models, the role of quantization becomes increasingly significant. However, outliers present in weights or activations notably influence the performance of quantized models. Recently, \citet{qtransformer} introduced a novel softmax function aimed at pretraining models in an outlier-free manner, thereby enhancing their suitability for quantization. Interestingly, we observed that such an approach leads to performance degradation in full precision. Building on this insight, we enhance the method by ensuring its normalization is invariant to sequence length, a crucial factor for bridging the gap between pretraining and fine-tuning. Moreover, this improved method also facilitates successful pretraining of causal language models. |
| title | Is It a Free Lunch for Removing Outliers during Pretraining? |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2402.12102 |