LLMs Use Just 16 of 256 Exponents — So We Compressed the Rest Away
2× compression on Llama-3-8B — and perplexity went down.
Written by
Senior Software Engineer at Upstart
Responses
Markdown supported · ⌘↵ to post
No responses yet.
2× compression on Llama-3-8B — and perplexity went down.
Senior Software Engineer at Upstart
Markdown supported · ⌘↵ to post
No responses yet.