Add link to GitHub repository and fix usage example

This PR adds the official GitHub repository link for the `VideoChat2-TPO` model to the model card. The paper states that the code will be released, and providing this link makes it easier for users to find and utilize the underlying code. Additionally, a minor correction is applied to the sample usage code to fix a variable name typo, improving its direct usability.

README.md

@@ -10,6 +10,8 @@ pipeline_tag: video-text-to-text
 
 This model is based on the paper [Task Preference Optimization: Improving Multimodal Large Language Models with Vision Task Alignment](https://huggingface.co/papers/2412.19326).
 
+Code: [https://github.com/OpenGVLab/VideoChat-TPO](https://github.com/OpenGVLab/VideoChat-TPO)
+
 ## 🏃 Installation
 
 ```
@@ -27,5 +29,5 @@ model_path = "OpenGVLab/VideoChat-TPO"
 tokenizer =  AutoTokenizer.from_pretrained(model_path,
 trust_remote_code=True,
 use_fast=False,)
-model = AutoModel.from_pretrained(model_path,  trust_remote_code=True, _tokenizer=self.tokenizer).eval()
+model = AutoModel.from_pretrained(model_path,  trust_remote_code=True, _tokenizer=tokenizer).eval()
 ```