ibm-granite/granite-4.0-h-1b
Full fine-tune vs ibm-granite/granite-4.0-h-1b-base. Fine-tune: 100% of the weights changed, across 40 of 40 layers.
Tensors changed: Embeddings 1/1, Norms 117/117, other 252/252, MLP 80/80, Attention 16/16
11 derivatives checked
- unsloth/granite-4.0-h-1b-GGUF: Not comparable yet
- ibm-granite/granite-4.0-h-1b-GGUF: Not comparable yet
- mlx-community/granite-4.0-h-1b-6bit: Quantization
- albapepper/scoracle-articulator-v4-q6: Quantization
- mlx-community/granite-4.0-h-1b-4bit: Quantization
- DevQuasar/ibm-granite.granite-4.0-h-1b-GGUF: Not comparable yet
- glogwa68/granite-4.0-h-1b-DISTILL-glm-4.7-think: Full fine-tune
- glogwa68/granite-4.0-h-1b-DISTILL-glm-4.7: Full fine-tune
- unsloth/granite-4.0-h-1b: Identical copy
- Danielbrdz/Barcenas-h-1b: LoRA-style
- TitleOS/ADBait-1B-FP16: LoRA-style
4 other repos hold byte-for-byte identical weight files.