RDson/WomboCombo-R1-Coder-14B-Preview
Full fine-tune vs Qwen/Qwen2.5-Coder-14B. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/1, Norms 97/97, MLP 144/144, Attention 336/336, lm_head 1/1
3 derivatives checked
- mradermacher/WomboCombo-R1-Coder-14B-Preview-i1-GGUF: Quantization
- mradermacher/WomboCombo-R1-Coder-14B-Preview-GGUF: Quantization
- CultriX/Qwen2.5-14B-ReasoningMerge: Full fine-tune
3 other repos hold byte-for-byte identical weight files.