CultriX/Qwen2.5-14B-ReasoningMerge
Full fine-tune vs RDson/WomboCombo-R1-Coder-14B-Preview. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Norms 97/97, MLP 144/144, Attention 336/336
4 derivatives checked
- mradermacher/Qwen2.5-14B-ReasoningMerge-i1-GGUF: Quantization
- mradermacher/Qwen2.5-14B-ReasoningMerge-GGUF: Quantization
- CultriX/Qwen2.5-14B-GeneralReasoning: Full fine-tune
- CultriX/Qwen2.5-14B-Verged: Full fine-tune