Category Archives: Uncategorized

How ExecuTorch handles cross attention KV cache?

Posted on December 23, 2025 by Mengwei Liu

Context In encoder-decoder transformer models, the decoder layer normally consists of a cross attention which performs key and value projections for encoder hidden states and calculate attention score between that and the query projection. Notice that in common Seq2seq models … Continue reading →

Posted in Uncategorized | Tagged , , , , | Leave a comment

Hello world!

Posted on November 26, 2012 by Mengwei Liu

Welcome to WordPress.com! This is your very first post. Click the Edit link to modify or delete it, or start a new post. If you like, use this post to tell readers why you started this blog and what you … Continue reading →

Posted in Uncategorized | 1 Comment