cisimcik/cisimcik-4b
Full fine-tune vs cisimcik/cisimcik-base. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/1, Norms 82/105, Attention 190/224, MLP 96/96
1 derivatives checked
- mradermacher/cisimcik-4b-GGUF: Same weights, converted