prithivMLmods/Magellanic-Opus-14B-Exp
Full fine-tune vs prithivMLmods/Megatron-Corpus-14B-Exp. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/1, Norms 97/97, MLP 144/144, Attention 336/336, lm_head 1/1
4 derivatives checked
- DevQuasar-8/prithivMLmods.Magellanic-Opus-14B-Exp-GGUF: Same weights, converted
- mradermacher/Magellanic-Opus-14B-Exp-i1-GGUF: Quantization
- mradermacher/Magellanic-Opus-14B-Exp-GGUF: Quantization
- MaziyarPanahi/Magellanic-Opus-14B-Exp-GGUF: Same weights, converted