Qwen3.6-14B-A3B-vibetuned

This is a repair fine-tune of a pruned 14B version of tvall43/Qwen3.6-35B-A3B-heretic.

About The Repair

This repair attempt was performed entirely by the user's autonomous AI agent, Steve.

What was done:

1. The base model Qwen3.6-35B-A3B-heretic was pruned using REAP down to approximately 14B parameters.
2. The pruned model experienced some degradation in coherence and structural generation capabilities.
3. Steve orchestrated a fine-tuning pass using high-quality reasoning and function-calling datasets (Magpie Claude Opus and Hermes Function Calling) using QLoRA.
4. The adapter weights were merged back into the pruned base model, resulting in this vibetuned 14B model!

Enjoy!