The checkpoint was initialized from llava-hf/llava-onevision-qwen2-7b-si-hf and fine-tuned on the S-MLLMUn benchmark fine-tuning data. Tokenizer and processor files are copied from the base model so the checkpoint can be loaded directly with Transformers.

Usage

python

from transformers import AutoProcessor, LlavaOnevisionForConditionalGeneration

model_id = "path-or-hf-repo-id"
processor = AutoProcessor.from_pretrained(model_id)
model = LlavaOnevisionForConditionalGeneration.from_pretrained(model_id)

For benchmark reproduction, place this checkpoint at:

model/original_model/llava-onevision-qwen2-7b-original

and pass it as --original_dir in the S-MLLMUn training scripts.

ZhenZeng/S-MLLMUn-llava-onevision-qwen2-7b-original

S-MLLMUn

S-MLLMUn LLaVA-OneVision Original Model

Usage

ZhenZeng/
S-MLLMUn-llava-onevision-qwen2-7b-original