tinygemma3_cifar

7 of 8 layers cache only a 4,096-token window rather than the full context, on a period of 6. Figures assume the default configuration; --swa-full disables the saving entirely.

A parameters × bits ÷ 8 estimate puts Q8_0 at roughly 0.02 GiB. The real file is 0.04 GiB, because a quantization is a mixture and some tensors are always kept at higher precision. The larger discrepancy is the cache: a flat formula gives 0.50 GiB at 32K context where the real figure is 0.12 GiB, because most of this model's layers cache a fixed window rather than the whole context.

How much VRAM does tinygemma3_cifar need?

Q8_0 is exactly 47,227,552 bytes (0.04 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.

How large is tinygemma3_cifar's KV cache?

0.12 GiB at 32K context with an f16 cache, computed per layer. Quantizing the cache to q8_0 roughly halves it, which is often the difference between a context length fitting and not.

Which quantization of tinygemma3_cifar should I use?

Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.