config.json embedding size of "vocab_size": 100352 does not match 100277

#6
by treehugg3 - opened

Did you ever find a fix for this? I cannot use it with llama.cpp because of this mistake.

Relevant: https://huggingface.co/databricks/dbrx-base/discussions/18

I fixed it here by extending the vocabulary: https://huggingface.co/treehugg3/dbrx-base-tokenizer-llamacpp

Sign up or log in to comment