The error message suggests that you are using the legacy behavior of the LlamaTokenizer class from the Transformers library. This behavior does not properly handle tokens that come after special tokens.

To resolve this issue, it is recommended to read the related pull request at the provided GitHub link (https://github.com/huggingface/transformers/pull/24565) for more information on how to update your code to the latest behavior.

Additionally, the error also mentions a problem with the device_map and weights being offloaded to the disk. To fix this, you need to provide an offload_folder for the weights or ensure that you have the safetensors library installed if the model you are using offers the weights in that format

You are using the legacy behaviour of the class transformersmodelsllamatokenization_llamaLlamaTokenizer This means that tokens that come after special tokens will not be properly handled We recommend

原文地址: https://www.cveoy.top/t/topic/iAQh 著作权归作者所有。请勿转载和采集!

免费AI点我,无需注册和登录