mradermacher/Viking-33B-GGUF
Full fine-tune vs LumiOpen/Viking-33B. mradermacher/Viking-33B-GGUF is a GGUF (Q8_0) of different weights. Fine-tune: 100% of the weights changed, across 56 of 56 layers.
Tensors changed: Embeddings 1/1, Norms 112/113, MLP 168/168, Attention 224/224, lm_head 1/1