KV cache compression worth it for Mistral

[ggufvault]

That matches what I was seeing. That matches what I saw on B200. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[ggufwire22]

My first guess would be this: On GB200, the easiest win is often to back off the aggressive settings and see whether the problem survives. Not glamorous advice, but repeatable baselines save a lot of time.

[edgequill]

Appreciate the detail. The pattern around token cost has been pretty similar for me. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[promptpoint]

A couple things I'd check first. If you're on MLX, I'd confirm the version first and then reduce variables until the behavior is obvious. Not glamorous advice, but repeatable baselines save a lot of time.

[bytelab]

What worked for me was pretty simple. If you're on PyTorch, I'd confirm the version first and then reduce variables until the behavior is obvious. Not glamorous advice, but repeatable baselines save a lot of time.

[kernelghost]

Ran a quick check on my box. The issue went away once I changed one setting and restarted the stack cleanly. I expected the model choice to matter more than it did.