+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: Evaluation & Testing (https://www.datasculpting.ai/inference/forumdisplay.php?fid=16)
+--- Forum: Model Comparisons (https://www.datasculpting.ai/inference/forumdisplay.php?fid=67)
+--- Thread: VRAM saving for image generation for Gemma (/showthread.php?tid=716)
Pages:
1
2
VRAM saving for image generation for Gemma - mangopeak72 - 10-03-2026
Spent some time on this today and wanted to compare notes.
I'm comparing two approaches here: keep it simple on M3 Max, or squeeze more out of the stack and deal with the complexity.
Interested in tradeoffs more than fan loyalty.
Re: VRAM saving for image generation for Gemma - circuitwire58 - 10-03-2026
I tested this tonight. My result was mixed: first run looked bad, second run looked normal. I expected the model choice to matter more than it did.
Re: VRAM saving for image generation for Gemma - vectorport - 10-04-2026
Ran a quick check on my box. My result was mixed: first run looked bad, second run looked normal. I expected the model choice to matter more than it did.
Re: VRAM saving for image generation for Gemma - cobalttinker - 10-05-2026
What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at structured output. Once you know the plain path works, then add the faster path back in piece by piece.
Re: VRAM saving for image generation for Gemma - torchtrack24 - 10-05-2026
I'd approach it in a boring order. Before rebuilding anything, I'd test one clean baseline with Mistral and compare from there.
Re: VRAM saving for image generation for Gemma - orbitline - 10-05-2026
A couple things I'd check first. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at speculative decoding. Not glamorous advice, but repeatable baselines save a lot of time.
Re: VRAM saving for image generation for Gemma - stackpilot - 10-05-2026
I hit almost the same issue last night.
Re: VRAM saving for image generation for Gemma - mangoquill - 10-08-2026
That matches what I was seeing. The pattern around batch size 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: VRAM saving for image generation for Gemma - llmarch83 - 10-09-2026
Maybe, but I had the opposite result. I wouldn't optimize around that first. On my side the bigger swing came from the runtime and not the model. Could be a workload difference, but I wouldn't generalize too fast from one setup.
Re: VRAM saving for image generation for Gemma - mangovault67 - 10-09-2026
That matches what I was seeing. I don't think your conclusion is far off. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.