open-r1/OlympicCoder-7B
Full fine-tune vs Qwen/Qwen2.5-Coder-7B-Instruct. Fine-tune: 93% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 0/1, Norms 57/57, MLP 84/84, Attention 196/196, lm_head 1/1
6 derivatives checked
- bartowski/open-r1_OlympicCoder-7B-GGUF: Same weights, converted
- mradermacher/OlympicCoder-7B-GGUF: Same weights, converted
- DevQuasar-5/open-r1.OlympicCoder-7B-GGUF: Same weights, converted
- tensorblock/OlympicCoder-7B-GGUF: Quantization
- II-Vietnam/R1-Math-Code-Fusion-v2: Full fine-tune
- II-Vietnam/R1-Math-Code-Fusion: Full fine-tune