usermma/GRM-Kerlin-3b-Abliterated

Full fine-tune vs OrionLLM/GRM-Kerlin-3b. Fine-tune: 95% of the weights changed, across 34 of 36 layers.

Tensors changed: Embeddings 1/1, Norms 0/73, MLP 102/108, Attention 136/252

2 derivatives checked

• mradermacher/GRM-Kerlin-3b-Abliterated-i1-GGUF: Quantization
• mradermacher/GRM-Kerlin-3b-Abliterated-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse