zkxxxx/VibeThinker-3B-heretic
Full fine-tune vs Qwen/Qwen2.5-Coder-3B. Fine-tune: 100% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 1/1, Norms 73/73, MLP 108/108, Attention 252/252
3 derivatives checked
- mradermacher/VibeThinker-3B-heretic-i1-GGUF: Quantization
- mradermacher/VibeThinker-3B-heretic-GGUF: Same weights, converted
- zkxxxx/VibeThinker-3B-heretic-fc: LoRA-style