allura-org/G2-9B-Aletheia-v1
Full fine-tune vs UCLA-AGI/Gemma-2-9B-It-SPPO-Iter3. Fine-tune: 90% of the weights changed, across 42 of 42 layers.
Tensors changed: Embeddings 0/1, Norms 1/169, MLP 126/126, Attention 168/168
4 derivatives checked
- bartowski/G2-9B-Aletheia-v1-GGUF: Same weights, converted
- mradermacher/G2-9B-Aletheia-v1-i1-GGUF: Quantization
- mradermacher/G2-9B-Aletheia-v1-GGUF: Same weights, converted
- mergekit-community/mergekit-slerp-xlblwaw: Full fine-tune