bartowski

/

nvidia_Nemotron-Cascade-14B-Thinking-GGUF

like 2

Just FYI: this model extends quite nicely with YaRN, I did a check on a 70k file and it produced pretty coherent descriptions of the contents.

Command to extend: llama-server --rope-scaling yarn --rope-scale 4 --yarn-orig-ctx 32768 --override-kv qwen3.context_length=int:131072