prompt length getting silly

[quiettoken]

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

Just sharing a setup that ended up working reasonably well: Mistral, TensorRT-LLM, 7900 XTX, with a lot of attention on NVFP4.

If someone wants the exact settings I can clean them up and post them.

[northgarden37]

What worked for me was pretty simple. If you're on LM Studio, I'd confirm the version first and then reduce variables until the behavior is obvious. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[quiettoken]

Good point. I ended up in about the same place using TensorRT-LLM. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.

[mlxgarden]

Appreciate the detail. I ended up in about the same place using MLX.

[mintstation99]

Interesting, I hadn't tried that. The pattern around workstation thermals has been pretty similar for me. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[mlxgarden]

My first guess would be this: Before rebuilding anything, I'd test one clean baseline with Phi and compare from there.

[cozywire]

What worked for me was pretty simple. If you're on MLX, I'd confirm the version first and then reduce variables until the behavior is obvious.

[rocmplane]

A couple things I'd check first. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at reranking. Once you know the plain path works, then add the faster path back in piece by piece.

[batchember]

What worked for me was pretty simple. On B200, the easiest win is often to back off the aggressive settings and see whether the problem survives.

[mlxgarden]

I think the bottleneck is somewhere else. I wouldn't optimize around that first. On my side the bigger swing came from the runtime and not the model.