+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: Optimization & Performance (https://www.datasculpting.ai/inference/forumdisplay.php?fid=10)
+--- Forum: Quantization (https://www.datasculpting.ai/inference/forumdisplay.php?fid=40)
+--- Thread: quant choices for 70B local (/showthread.php?tid=321)
Pages:
1
2
quant choices for 70B local - vrambandit - 09-16-2026
This keeps coming up in conversation so I figured I'd start a thread.
The more I work with local setups, the more it seems like long context decides everything downstream.
Curious whether that lines up with what other people are seeing. The logs were somehow calmer than I was.
Re: quant choices for 70B local - southtinker - 09-17-2026
Good point. 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.
Re: quant choices for 70B local - nanobrook - 09-17-2026
That matches what I was seeing. That matches what I saw on 5090. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.
Re: quant choices for 70B local - fusedtinker - 09-17-2026
Appreciate the detail. I ended up in about the same place using Ollama. The biggest change for me was simplifying the stack instead of adding one more optimization layer.
There are a few different conversations hiding inside this thread, so being explicit about the goal helps.
Re: quant choices for 70B local - marbletrack - 09-18-2026
A couple things I'd check first. On Threadripper box, the easiest win is often to back off the aggressive settings and see whether the problem survives. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.
There are a few different conversations hiding inside this thread, so being explicit about the goal helps.
Re: quant choices for 70B local - vrambandit - 09-18-2026
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. Once you know the plain path works, then add the faster path back in piece by piece. At this point the real bottleneck is my patience.
Re: quant choices for 70B local - relaynest82 - 09-20-2026
Quote: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. O
Maybe, but I had the opposite result. I get why that seems right, but on 5090 the bottleneck felt elsewhere. Not saying you're wrong, just that I wouldn't lock that in as the lesson yet.
Re: quant choices for 70B local - cudaline - 09-20-2026
that fixed it for me too.
Re: quant choices for 70B local - vrambandit - 09-23-2026
following this because I'm in the same boat.
Re: quant choices for 70B local - paperrunner - 09-24-2026
I'd approach it in a boring order. Before rebuilding anything, I'd test one clean baseline with Qwen and compare from there. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.