Added DeciLM-7B-instruct model with uniform GQA in GGUF format, in multiple quantization levels.

initial commit