ValiantLabs/Llama3.1-70B-ShiningValiant2
LoRA-style vs meta-llama/Llama-3.1-70B-Instruct. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of q_proj, v_proj changed, in 80 of 80 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/161, MLP 0/240, Attention 160/320, lm_head 0/1
5 derivatives checked
- bartowski/Llama3.1-70B-ShiningValiant2-GGUF: Quantization
- mradermacher/Llama3.1-70B-ShiningValiant2-i1-GGUF: Quantization
- mradermacher/Llama3.1-70B-ShiningValiant2-GGUF: Quantization
- DevQuasar-3/ValiantLabs.Llama3.1-70B-ShiningValiant2-GGUF: Quantization
- sequelbox/Llama3.1-70B-PlumChat: Full fine-tune