Similar to the llama engine, quantizing the KV cache requires flash attention to be enabled through the Ollama server.
Attach a file by drag & drop or click to upload
