prompt length getting silly

[mlxgarden]

Thought I'd post a small recipe because a lot of threads ask similar things.

This isn't revolutionary, but the combination of LM Studio and 5090 felt surprisingly usable once I tuned around tool calling.

Maybe useful as a starting point for newer folks.

[tensorworks]

Interesting, I hadn't tried that. I ended up in about the same place using llama.cpp. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.

[batchember]

Tried to reproduce it just now. The issue went away once I changed one setting and restarted the stack cleanly.

[paperport64]

I think the bottleneck is somewhere else. I get why that seems right, but on GB200 the bottleneck felt elsewhere. Not saying you're wrong, just that I wouldn't lock that in as the lesson yet.

[mlxgarden]

A couple things I'd check first. If you're on PyTorch, I'd confirm the version first and then reduce variables until the behavior is obvious.

[cacheline]

I tested this tonight. My result was mixed: first run looked bad, second run looked normal. The main thing that moved was latency, not quality.

[mintstation99]

Interesting, I hadn't tried that. The pattern around prefill/decode split has been pretty similar for me. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[mlxgarden]

My first guess would be this: On A6000, the easiest win is often to back off the aggressive settings and see whether the problem survives.

[quiettoken]

same here on my side.

[batchember]

I think the bottleneck is somewhere else. I've seen people blame prefill/decode split when the real issue was much more mundane.