Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2405.09857 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929345470136320 |
|---|---|
| author | Feng, Dawei Zhang, Yihai Xu, Zhixuan |
| author_facet | Feng, Dawei Zhang, Yihai Xu, Zhixuan |
| contents | Pretrained Large Language Models (LLM) such as ChatGPT, Claude, etc. have demonstrated strong capabilities in various fields of natural language generation. However, there are still many problems when using LLM in specialized domain-specific fields. When using generative AI to process downstream tasks, a common approach is to add new knowledge (e.g., private domain knowledge, cutting-edge information) to a pretrained model through continued training or fine-tuning. However, whether there is a universal paradigm for domain adaptation training is still an open question. In this article, we proposed Information Gain Optimized Tokenizer (IGOT), which analyzes the special token set of downstream tasks, constructs a new subset using heuristic function $ϕ$ with the special token and its information gain, to build new domain-specific tokenizer, and continues pretraining on the downstream task data. We explored the many positive effects of this method's customized tokenizer on domain-adaptive pretraining and verified this method can perform better than the ordinary method of just collecting data and fine-tuning. Based on our experiment, the continued pretraining process of IGOT with LLaMA-7B achieved 11.9\% token saving, 12.2\% training time saving, and 5.8\% maximum GPU VRAM usage saving, combined with the T5 model, we can even reach a 31.5\% of training time saving, making porting general generative AI to specific domains more effective than before. In domain-specific tasks, supervised $IGOT_τ$ shows great performance on reducing both the convergence radius and convergence point during keep pretraining. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_09857 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | IGOT: Information Gain Optimized Tokenizer on Domain Adaptive Pretraining Feng, Dawei Zhang, Yihai Xu, Zhixuan Computation and Language Artificial Intelligence Pretrained Large Language Models (LLM) such as ChatGPT, Claude, etc. have demonstrated strong capabilities in various fields of natural language generation. However, there are still many problems when using LLM in specialized domain-specific fields. When using generative AI to process downstream tasks, a common approach is to add new knowledge (e.g., private domain knowledge, cutting-edge information) to a pretrained model through continued training or fine-tuning. However, whether there is a universal paradigm for domain adaptation training is still an open question. In this article, we proposed Information Gain Optimized Tokenizer (IGOT), which analyzes the special token set of downstream tasks, constructs a new subset using heuristic function $ϕ$ with the special token and its information gain, to build new domain-specific tokenizer, and continues pretraining on the downstream task data. We explored the many positive effects of this method's customized tokenizer on domain-adaptive pretraining and verified this method can perform better than the ordinary method of just collecting data and fine-tuning. Based on our experiment, the continued pretraining process of IGOT with LLaMA-7B achieved 11.9\% token saving, 12.2\% training time saving, and 5.8\% maximum GPU VRAM usage saving, combined with the T5 model, we can even reach a 31.5\% of training time saving, making porting general generative AI to specific domains more effective than before. In domain-specific tasks, supervised $IGOT_τ$ shows great performance on reducing both the convergence radius and convergence point during keep pretraining. |
| title | IGOT: Information Gain Optimized Tokenizer on Domain Adaptive Pretraining |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2405.09857 |