RedHatAI/Sparse-Llama-3.1-8B-2of4
Full fine-tune vs meta-llama/Llama-3.1-8B. Fine-tune: 93% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 0/1, Norms 65/65, MLP 96/96, Attention 128/128, lm_head 1/1
4 derivatives checked
- bartowski/Sparse-Llama-3.1-8B-2of4-GGUF: Same weights, converted
- tensorblock/Sparse-Llama-3.1-8B-2of4-GGUF: Full fine-tune
- DevQuasar-5/neuralmagic.Sparse-Llama-3.1-8B-2of4-GGUF: Quantization
- RedHatAI/Sparse-Llama-3.1-8B-ultrachat_200k-2of4: Full fine-tune