Thank you for your contribution to KV cache compression. I would like to ask 1. Does the llama-3 model, e.g., llama-3-8b, support this? 2. Does CHAI still work well with the GQA model?
Thank you for your contribution to KV cache compression. I would like to ask