elinas/Llama-3-15B-Instruct-zeroed
Partial fine-tune (fewer layers) vs meta-llama/Meta-Llama-3-8B-Instruct. meta-llama/Meta-Llama-3-8B-Instruct matches part of elinas/Llama-3-15B-Instruct-zeroed (291 of its 579 tensors; the rest of elinas/Llama-3-15B-Instruct-zeroed has no counterpart). On the shared part: Partial change: 22% of the weights changed (attention, mlp, norms), in 8 of 32 layers.
Tensors changed: Embeddings 0/1, Norms 16/65, MLP 24/96, Attention 32/128, lm_head 0/1
6 derivatives checked
- bartowski/L3-Aethora-15B-V2-GGUF: Quantization
- featherless-ai-quants/elinas-Llama-3-15B-Instruct-zeroed-GGUF: Quantization
- bullerwins/L3-Aethora-15B-V2-GGUF: Quantization
- SteelStorage/L3-Aethora-15B-V2: Partial fine-tune
- elinas/Llama-3-15B-Instruct-zeroed-ft: Abliterated
- elinas/Llama-3-15B-Instruct-ft-v2: Full fine-tune