ArRENCEAI/granite-3.1-2b-instruct-OBLITERATED

Full fine-tune vs ibm-granite/granite-3.1-2b-instruct. Fine-tune: 95% of the weights changed, across 38 of 40 layers.

Tensors changed: Embeddings 1/1, Norms 0/81, MLP 114/120, Attention 152/160

2 derivatives checked

• mradermacher/granite-3.1-2b-instruct-OBLITERATED-i1-GGUF: Quantization
• mradermacher/granite-3.1-2b-instruct-OBLITERATED-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse