Nitral-Archive/Pinecone-Rune-12b-Token-Surgery-Chatml

Base model: Pinecone-Rune-12b

Donor model: Nitral-AI/Irixxed-Magcap-12B-0.1a

Arcees tokensurgery using (-v -k 64 --cosine-similarity --cuda --low-cpu-memory) on a colab l4 for brevity.

Example Notebook using l4: https://huggingface.co/Nitral-AI/Pinecone-Rune-12b-Token-Surgery-Chatml/tree/main/TokenSurgeon-Example