ibm-granite/granite-vision-3.1-2b-preview
Full fine-tune (fewer layers) vs ibm-granite/granite-3.1-2b-instruct. ibm-granite/granite-3.1-2b-instruct matches part of ibm-granite/granite-vision-3.1-2b-preview (361 of its 815 tensors; the rest of ibm-granite/granite-vision-3.1-2b-preview has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 40 of 40 layers.
Tensors changed: Norms 81/81, MLP 120/120, Attention 160/160
1 derivatives checked
- DevQuasar-5/ibm-granite.granite-vision-3.1-2b-preview-GGUF: Same weights, converted (fewer layers)