Johnblick187/Qwen3-Omni-Coder
Full fine-tune (fewer layers) vs huihui-ai/Huihui-Qwen3-Coder-30B-A3B-Instruct-abliterated. huihui-ai/Huihui-Qwen3-Coder-30B-A3B-Instruct-abliterated matches part of Johnblick187/Qwen3-Omni-Coder (17545 of its 28010 tensors; the rest of Johnblick187/Qwen3-Omni-Coder has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Norms 185/185, Attention 186/186, MLP 46/46