Method __init__

architecture/embeddings.py:1544–1555

(
        self,
        text_embed_dim: int = 1024,
        image_embed_dim: int = 768,
        cross_attention_dim: int = 768,
        num_image_text_embeds: int = 10,
    )

class TextImageProjection(nn.Module):
    def __init__(
        self,
        text_embed_dim: int = 1024,
        image_embed_dim: int = 768,
        cross_attention_dim: int = 768,
        num_image_text_embeds: int = 10,
    ):
        super().__init__()

        self.num_image_text_embeds = num_image_text_embeds
        self.image_embeds = nn.Linear(image_embed_dim, self.num_image_text_embeds * cross_attention_dim)
        self.text_proj = nn.Linear(text_embed_dim, cross_attention_dim)

    def forward(self, text_embeds: torch.Tensor, image_embeds: torch.Tensor):
        batch_size = text_embeds.shape[0]