admin管理员组

文章数量:1332873

I am dealing with two embeddings, text and image both are last_hidden_state of transfomer models (bert and vit), so the shapes are (batch, seq, emd_dim). I want to feed text information to image using a cross attention mechanism and I was wondering whether this line of code will give me what I need:

cross_attention = nn.MultiheadAttention(embed_dim=768, num_heads=12, dropout=0.1)
attn_output, attn_output_weights = cross_attention(text_last, img_last, img_last)

I tried the provided code but I am not sure whether it is the right approach

本文标签： deep learningMultiModal Cross attentionStack Overflow

版权声明：本文标题：deep learning - MultiModal Cross attention - Stack Overflow 内容由网友自发贡献，该文观点仅代表作者本人， 转载请联系作者并注明出处：http://www.betaflare.com/web/1742316094a2451845.html， 本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌抄袭侵权/违法违规的内容，一经查实，本站将立刻删除。