Bump version to v1.5.0

Bump version to v1.2.3

1. Add guidance for fine-tuning the pre-trained InternVL-Chat-V1.2 and V1.2-Plus
2. Fix retrieval fine-tune bug & support COCO fine-tune
3. Fix AI2D training & testing & results
4. Fix OOM error during training when using transformers==4.36.2 in LLaVA codebase
5. Provide url for downloading GeoQA+ images

Bump version to v1.2.2 (#48)

* Remove repetition_penalty

* Update grounding evaluation

* Improve model code

* Don't use repetition_penalty

* Update BLOG.md

* Update README.md

* Update zero configs

* Update training code

Update README.md

Support MathVista

Rename InternVL shell scripts