Patch-Level Training for Large Language Models
MP3•Episode home
Manage episode 430520574 series 2954468
Content provided by Rob. All podcast content including episodes, graphics, and podcast descriptions are uploaded and provided directly by Rob or their podcast platform partner. If you believe someone is using your copyrighted work without your permission, you can follow the process outlined here https://player.fm/legal.
As Large Language Models (LLMs) achieve remarkable progress in language understanding and generation, their training efficiency has become a critical concern. Traditionally, LLMs are trained to predict the next token in a sequence. Despite the success of token-level training, it suffers from considerable computational costs due to the need to process an extensive number of tokens. To mitigate this issue, this paper introduces patch-level training for LLMs, which reduces the sequence length by compressing multiple tokens into a single patch. During patch-level training, we feed the language model shorter sequences of patches and train it to predict the next patch, thereby processing the majority of the training data at a significantly reduced computational cost. Following this, the model continues token-level training on the remaining training data to align with the inference mode. Experiments on a diverse range of models (370M-2.7B parameters) demonstrate that patch-level training can reduce overall computational costs to 0.5$\times$, without compromising the model performance compared to token-level training. Source code: \url{https://github.com/shaochenze/PatchTrain}.
2024: Chenze Shao, Fandong Meng, Jie Zhou
https://arxiv.org/pdf/2407.12665v1
…
continue reading
2024: Chenze Shao, Fandong Meng, Jie Zhou
https://arxiv.org/pdf/2407.12665v1
298 episodes