[Usage] llava:apply_delta is killed

[k1e3v1i4n]

When did you clone our code?

I cloned the code base before 5/1/23, but have pulled the latest code base

Describe the issue

Issue: After download done, the script of apply_delta is killed

Command: (llava) [root@VM-87-49-centos /data/LLaVA]# python3 -m llava.model.apply_delta --base /data/LLaMA/13B_hf --target /data/LLaVA/LLaVA-13B-v1 --delta liuhaotian/LLaVA-13b-delta-v1-1

Log:

• This IS expected if you are initializing CLIPVisionModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).
• This IS NOT expected if you are initializing CLIPVisionModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model). Loading checkpoint shards: 33% | 1/3 [01:02<02:05, 62.94s/it]Killed

Screenshots:

[k1e3v1i4n]

OOM