[nolist_policy]

On the other hand ollama supports iSWA for Gemma 3 while llama.cpp doesn't. iSWA reduces kv cache size to 1/6.

[vlovich123]

What’s iSWA? Can’t find any reference online

[imtringued]

Gemma 3 has some layers with a context size of 1024 tokens and others having full length. You need to read the Gemma technical report.

[nolist_policy]

interleaved sliding window attention