Berkelium-ai/BerkeliumGPT-Coder-3b
Different architecture vs Qwen/Qwen2.5-3B-Instruct. Different architecture: only 56 of 434/338 tensors match by name and shape.
2 derivatives checked
- mradermacher/BerkeliumGPT-Coder-3b-i1-GGUF: Quantization
- mradermacher/BerkeliumGPT-Coder-3b-GGUF: Same weights, converted