Comment on Making MCP cheaper via CLIparentComments−thellimistOP6moWhat do you mean?−hiccuphippo6moThe article says the LLM has to load 15540 tokens every time, I wonder if that can be reduced while retaining the context maybe with deduplications, removing superfluous words, using shorter expressions with the same meaning or things like that.
Comments
What do you mean?
The article says the LLM has to load 15540 tokens every time, I wonder if that can be reduced while retaining the context maybe with deduplications, removing superfluous words, using shorter expressions with the same meaning or things like that.