IFM/K2-Think
Full fine-tune vs Qwen/Qwen2.5-32B. Fine-tune: 100% of the weights changed, across 64 of 64 layers.
Tensors changed: Embeddings 1/1, Norms 127/129, MLP 192/192, Attention 448/448, lm_head 1/1
3 derivatives checked
- mradermacher/K2-Think-i1-GGUF: Quantization
- mradermacher/K2-Think-GGUF: Quantization
- sambal/k2-v2-think-dpo-checkpoint-200: LoRA-style
2 other repos hold byte-for-byte identical weight files.