+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: Prompting & Context Engineering (https://www.datasculpting.ai/inference/forumdisplay.php?fid=15)
+--- Forum: Prompts / System Prompts (https://www.datasculpting.ai/inference/forumdisplay.php?fid=62)
+--- Thread: few-shot when you actually need it with PyTorch (/showthread.php?tid=293)
few-shot when you actually need it with PyTorch - promptspark58 - 09-16-2026
Posting this in case it saves someone else a little time.
If you were starting from scratch today for this use case, what would you choose first and what would you ignore?
I'm especially interested in mistakes worth avoiding.
This is mostly for local work rather than public serving.
Re: few-shot when you actually need it with PyTorch - mlxgarden - 09-16-2026
Good point. I don't think your conclusion is far off.
Re: few-shot when you actually need it with PyTorch - phasescope - 09-16-2026
My first guess would be this: If you're on TensorRT-LLM, I'd confirm the version first and then reduce variables until the behavior is obvious.
Re: few-shot when you actually need it with PyTorch - promptspark58 - 09-17-2026
I hit almost the same issue last night.
Re: few-shot when you actually need it with PyTorch - mlxgarden - 09-17-2026
Good point. The pattern around VRAM headroom has been pretty similar for me.
Re: few-shot when you actually need it with PyTorch - northgarden37 - 09-17-2026
I think the bottleneck is somewhere else. I get why that seems right, but on 7900 XTX the bottleneck felt elsewhere. Worth testing the simpler explanation before adopting the more interesting one.
Re: few-shot when you actually need it with PyTorch - cobalttinker - 09-18-2026
Tried to reproduce it just now. The issue went away once I changed one setting and restarted the stack cleanly. The main thing that moved was latency, not quality.
Re: few-shot when you actually need it with PyTorch - nanopatch - 09-19-2026
That matches what I was seeing. That matches what I saw on M3 Max.
Re: few-shot when you actually need it with PyTorch - localpatch - 09-19-2026
A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Kimi and compare from there. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.