WARNING: tokenization mismatch: 759 vs. 756. (ignored)

[460041021]

Thanks for your exallent work!

I strictly followed the provided package versions, but encountered an issue during the LoRA fine-tuning process.

Some weights of the model checkpoint at /app/checkpoints/robopoint-v1-vicuna-v1.5-13b were not used when initializing LlavaLlamaForCausalLM: ['model.vision_tower.vision_tower.vision_model.embeddings.class_embedding', 'model.vision_tower.vision_tower.vision_model.embeddings.patch_embedding.weight', 'model.vision_tower.vision_tower.vision_model.embeddings.position_embedding.weight', 
...

WARNING: tokenization mismatch: 759 vs. 756. (ignored)
WARNING: tokenization mismatch: 759 vs. 756. (ignored)
...

There are many errors like this. Is this normal?