cjvt/GaMS-27B
Full fine-tune vs google/gemma-2-27b. Fine-tune: 100% of the weights changed, across 46 of 46 layers.
Tensors changed: Embeddings 1/1, Attention 184/184, Norms 185/185, MLP 138/138
3 derivatives checked
- mradermacher/GaMS-27B-i1-GGUF: Quantization
- cjvt/GaMS-27B-Instruct: Full fine-tune
- mradermacher/GaMS-27B-GGUF: Quantization