This model is an ExecuTorch-compatible, quantized variant of Meta’s Llama 3.2 3B Instruct, using the official SpinQuant INT4/EO8 format provided by Meta for efficient on-device inference.
Based on: Meta Llama 3.2
Quantization method: SpinQuant (INT4 weights / EO8 activations) by Meta
For more information, refer to the
Llama 3.2 Model Card
.