ibm-granite/granite-4.2-3b
Full fine-tune vs ibm-granite/granite-4.1-3b-base. Fine-tune: 100% of the weights changed, across 40 of 40 layers.
Tensors changed: Embeddings 1/1, Norms 78/81, MLP 120/120, Attention 160/160
30 derivatives checked
- ibm-granite/granite-4.2-3b-GGUF: Same weights, converted
- webAI-Official/TwIL-LM3-Pro: LoRA-style
- bartowski/granite-4.2-3b-GGUF: Same weights, converted
- NANI-Nithin/granite-4.2-3b-GGUF: Same weights, converted
- saidutta69/granite-4.2-3b-heretic: Abliterated
- mradermacher/granite-4.2-3b-i1-GGUF: Quantization
- lmstudio-community/granite-4.2-3b-MLX-8bit: Quantization
- ibm-granite/granite-4.2-3b-fp8: Quantization
- ibm-granite/granite-4.2-3b-nvfp4: Quantization
- lmstudio-community/granite-4.2-3b-MLX-4bit: Quantization
- ibm-granite/granite-4.2-3b-mxfp4: Quantization
- lmstudio-community/granite-4.2-3b-GGUF: Quantization
- KittyLM/kittylm-granite-4.2-3b: LoRA-style
- ibm-granite/granite-4.2-3b-q4-mlx: Quantization
- richardyoung/granite-4.2-3b-heretic: Abliterated
- ibm-granite/granite-4.2-3b-q8-mlx: Quantization
- ibm-granite/granite-4.2-3b-bf16-mlx: Same weights, repacked
- mradermacher/granite-4.2-3b-GGUF: Same weights, converted
- quark75/granite-4.2-3b-MXFP4A16-GPTQ: Quantization
- snellchapo/granite-4.2-3b-heretic-abliterated: Abliterated
- tinyopsec/granite-4.2-3b-Heretic: Abliterated
- liodon-ai/granite-4.2-3b-imatrix-GGUF: Quantization
- lmstudio-community/granite-4.2-3b-MLX-6bit: Quantization
- 0xzknw/Granite-4.2-3B-Heretic-NX-PRIME-GGUF: Abliterated
- Neurona/TwIL-LM3-Pro: LoRA-style
- Orvena/granite-4.2-3b-MLX-4bit: Quantization
- salohcin714/granite-4.2-3b-bf16-mlx: Same weights, repacked
- DevQuasar/ibm-granite.granite-4.2-3b-GGUF: Quantization
- quark75/granite-4.2-3b-Q4_K_M-imatrix-GGUF: Quantization
- Jeethu/granite-4.2-3b-PARO: Quantization
2 other repos hold byte-for-byte identical weight files.