XiaodongChen/Llama-3.1-5.4B
Partial fine-tune (fewer layers) vs meta-llama/Llama-3.1-8B. XiaodongChen/Llama-3.1-5.4B matches part of meta-llama/Llama-3.1-8B (183 of its 291 tensors; the rest of meta-llama/Llama-3.1-8B has no counterpart). On the shared part: Partial change: 18% of the weights changed (attention, mlp, norms), in 15 of 20 layers.
Tensors changed: Embeddings 0/1, Norms 12/41, MLP 13/60, Attention 21/80, lm_head 0/1
2 derivatives checked
- mradermacher/Llama-3.1-5.4B-i1-GGUF: Quantization
- mradermacher/Llama-3.1-5.4B-GGUF: Same weights, converted