ibm-granite/granite-3.1-3b-a800m-instruct
Full fine-tune vs ibm-granite/granite-3.1-3b-a800m-base. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: other 96/96, Norms 65/65, Attention 128/128
7 derivatives checked
- bartowski/granite-3.1-3b-a800m-instruct-GGUF: Same weights, converted
- Alcoft/granite-3.1-3b-a800m-instruct-GGUF: Same weights, converted
- lmstudio-community/granite-3.1-3b-a800m-instruct-GGUF: Quantization
- tensorblock/granite-3.1-3b-a800m-instruct-GGUF: Quantization
- DevQuasar-6/ibm-granite.granite-3.1-3b-a800m-instruct-GGUF: Quantization
- PJMixers-Dev/Granite-3.1-Earthen-v0.3-3B-A800M: Partial fine-tune
- franciscobdl/Granite-Estigia-MultilanguageV3: LoRA-style
1 other repos hold byte-for-byte identical weight files.