ibm-granite/granite-7b-instruct
Full fine-tune vs ibm-granite/granite-7b-base. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Norms 65/65, MLP 96/96, Attention 128/128
4 derivatives checked
- mradermacher/granite-7b-instruct-i1-GGUF: Quantization
- mradermacher/granite-7b-instruct-GGUF: Same weights, converted
- tensorblock/granite-7b-instruct-GGUF: Quantization
- DevQuasar-4/granite-7b-instruct-GGUF: Quantization
2 other repos hold byte-for-byte identical weight files.