tokenizer.json modified after tokenizer.save_pretrained of OLMO models

[zzf1130]

System Info

• transformers version: 4.45.0
• Platform: Linux-6.8.0-48-generic-x86_64-with-glibc2.39
• Python version: 3.10.15
• Huggingface_hub version: 0.26.2
• Safetensors version: 0.4.5
• Accelerate version: 1.0.1
• Accelerate config: not found
• PyTorch version (GPU?): 2.4.0+rocm6.1 (True)
• Tensorflow version (GPU?): not installed (NA)
• Flax version (CPU?/GPU?/TPU?): not installed (NA)
• Jax version: not installed
• JaxLib version: not installed
• Using distributed or parallel set-up in script?:
• Using GPU in script?:
• GPU type: AMD Instinct MI250X/MI250

Who can help?

@ArthurZucker and @itazap

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction

When I load and then save the tokenizer with OLMO models, the tokenizer.json files appear different, particularly with the merge key.

The code to reproduce that is :

from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("allenai/OLMo-1B-0724-hf")
tokenizer.save_pretrained("saved_tokenizer")

Expected behavior

The original tokenizer.json and the saved tokenizer.json should be the same.

[LysandreJik]

Hey @zzf1130, I believe this change is to make the tokenizer.json more flexible/future-proof, and is therefore a voluntary change.

I will let @ArthurZucker comment on it, thank you!