+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: Optimization & Performance (https://www.datasculpting.ai/inference/forumdisplay.php?fid=10)
+--- Forum: Benchmarks (https://www.datasculpting.ai/inference/forumdisplay.php?fid=43)
+--- Thread: first LoRA that actually helped: early impressions (/showthread.php?tid=1488)
first LoRA that actually helped: early impressions - northgrid33 - 10-18-2026
Curious what people here are seeing.
I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling prefill/decode split.
Not looking for the perfect answer, mostly what worked in real use.
Re: first LoRA that actually helped: early impressions - routearch - 10-18-2026
Good point. 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.
Re: first LoRA that actually helped: early impressions - batchsize - 10-18-2026
Good point. I ended up in about the same place using MLX. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.
Re: first LoRA that actually helped: early impressions - batchsize - 10-19-2026
Good point. The pattern around MoE serving has been pretty similar for me. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.
Re: first LoRA that actually helped: early impressions - localtinker - 10-19-2026
What exact setup are you running? Is this local-only or are you routing anything externally?
Re: first LoRA that actually helped: early impressions - nanobrook - 10-19-2026
Quote:Good point. The pattern around MoE serving has been pretty similar for me. I got better results once I stopped chasing one metric and just looked at w
Thanks, that helps. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimization layer.