add pt model, tokenizer, feat extractor

Best model so far on v4 valid (epoch=4, freeze backbones)

Update config.json

Update config.json

Update config.json

Update config.json

update from clip_vision_model to clip

Update text_config to text_config_dict and vision_config to vision_config_dict

Add test model (CC+WIT+COCO) 7 epochs