curiositytech/MARS-v0.2
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: Norms 0/65, MLP 0/96, Attention 64/128
2 derivatives checked
- mradermacher/MARS-v0.2-i1-GGUF: Quantization
- mradermacher/MARS-v0.2-GGUF: Same weights, converted