Back to AIBriefs
LaunchAI Models

Qwable fine-tune replicates Fable 5 reasoning in local Qwen3.6-27B

Qwable fine-tune replicates Fable 5 reasoning in local Qwen3.6-27BDecrypt

The 27B-parameter model, a full fine-tune of Alibaba's Qwen3.6-27B trained on Fable 5 reasoning examples, runs on consumer hardware. An abliterated version removes refusal behavior via weight modification using llama.cpp's cvector-generator.

·
Jun 23, 10:01 PM
Qwable fine-tune replicates Fable 5 reasoning in local Qwen3.6-27B — AIBriefs