ValiantLabs/Llama3.1-8B-Cobalt
LoRA-style vs meta-llama/Llama-3.1-8B-Instruct. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of q_proj, v_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/65, MLP 0/96, Attention 64/128, lm_head 0/1
5 derivatives checked
- bartowski/Llama3.1-8B-Cobalt-GGUF: Quantization
- mradermacher/Llama3.1-8B-Cobalt-i1-GGUF: Quantization
- mradermacher/Llama3.1-8B-Cobalt-GGUF: Same weights, converted
- sequelbox/Llama3.1-8B-PlumMath: LoRA-style
- Stormtrooperaim/Valiant-Vanta-8B-Dark-Fusion: Partial fine-tune