Args: images (torch.Tensor): [b, 3, H, W] Returns: image_features (torch.Tensor): [b, n_patch, d]

Args: images (torch.Tensor): [bs, 3, H, W] Returns: res (torch.Tensor): [bs, t, c]

Args: x_or_tuple (Union[Tuple[torch.Tensor, torch.Tensor], torch.Tensor]: if it is a tuple of torch.Tensor, the

Intermediate layer accessor (NOTE: This is a WIP experiment). Inspired by DINO / DINOv2 interface

Get current round formatted question prompt during sft training

ViT weight initialization, original timm impl (for reproducibility)

Parameters: y: List of tuples representing the message and response pairs. Each message and response should be a string, which may be in

Reset a new message.

Convert the conversation to gradio chatbot format.

Convert the conversation to OpenAI chat completion format.