cjvt/GaMS-27B-Instruct

Full fine-tune vs cjvt/GaMS-27B. Fine-tune: 99% of the weights changed, across 46 of 46 layers.

Tensors changed: Embeddings 1/1, Attention 182/184, Norms 75/185, MLP 136/138

3 derivatives checked

• mradermacher/GaMS-27B-Instruct-i1-GGUF: Quantization
• mradermacher/GaMS-27B-Instruct-GGUF: Quantization
• cjvt/GaMS-27B-Instruct-Nemotron: Full fine-tune

On the Hub · Downloads on Model Pulse