llama-batched-bench help

[Master-Pr0grammer]

I just switched from ollama and I am new to llama.cpp so I'm sorry if this isa dumb question, but I wanted to set up an embedding model and after my slow experience with ollama, I wanted to set it up to be as fast as possible to handle RAG with long documents.

Anyways, I wanted to use llama-batched-bench to test different batch sizes, or parallel requests that would produce the most throughput, but I cant seem to figure out how this works. I found an example for llama-bench with different prompt input lengths, but llama-bench and llama-batched-bench are very different.

Also, while I am here, how does one set varying context lengths in llama-bench?