MaziyarPanahi/Llama-3-16B-Instruct-v0.1
Partial fine-tune (fewer layers) vs MaziyarPanahi/Llama-3-11B-Instruct-v0.1. MaziyarPanahi/Llama-3-11B-Instruct-v0.1 matches part of MaziyarPanahi/Llama-3-16B-Instruct-v0.1 (435 of its 651 tensors; the rest of MaziyarPanahi/Llama-3-16B-Instruct-v0.1 has no counterpart). On the shared part: Partial change: 45% of the weights changed (attention, mlp, norms), in 24 of 48 layers.
Tensors changed: Embeddings 0/1, Norms 48/97, MLP 72/144, Attention 96/192, lm_head 0/1
3 derivatives checked
- MaziyarPanahi/Llama-3-16B-Instruct-v0.1-GGUF: Same weights, converted
- mradermacher/Llama-3-16B-Instruct-v0.1-i1-GGUF: Quantization
- mradermacher/Llama-3-16B-Instruct-v0.1-GGUF: Quantization