mattshumer/ref_70_e3
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 down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 80 of 80 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/161, MLP 240/240, Attention 320/320
8 derivatives checked
- leafspark/Reflection-Llama-3.1-70B-GGUF: Same weights, converted
- unsloth/Reflection-Llama-3.1-70B-GGUF: Quantization
- second-state/Reflection-Llama-3.1-70B-GGUF: Same weights, converted
- MaziyarPanahi/ref_70_e3-GGUF: Quantization
- gaianet/Reflection-Llama-3.1-70B-GGUF: Same weights, converted
- mradermacher/ref_70_e3-i1-GGUF: Quantization
- mradermacher/ref_70_e3-GGUF: Quantization
- jhofseth/Reflection-Llama-3.1-70B-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.