ibm-granite/granite-4.0-tiny-preview
Full fine-tune vs ibm-granite/granite-4.0-tiny-base-preview. Fine-tune: 98% of the weights changed, across 40 of 40 layers.
Tensors changed: other 356/372, Norms 117/117, MLP 80/80, Attention 16/16
5 derivatives checked
- bartowski/ibm-granite_granite-4.0-tiny-preview-GGUF: Same weights, converted
- mradermacher/granite-4.0-tiny-preview-GGUF: Same weights, converted
- mradermacher/granite-4.0-tiny-preview-i1-GGUF: Quantization
- DevQuasar/ibm-granite.granite-4.0-tiny-preview-GGUF: Same weights, converted
- wowbager/granite-cermat-cz-7b: Full fine-tune