Epiculous / Effervescence-27B-GGUF

Taking what worked well with the CrimsonDawn line, Effervescence-27B carries the same two-phase methodology over to a much bigger base, Qwen3.5-27B. This time the instruct stage got a significantly expanded dataset, and the whole thing is trained as an 8-bit LoRA to fit the 27…