Syed-Hasan-8503/Exodius-70B
Full fine-tune (fewer layers) vs abacusai/Smaug-72B-v0.1. Syed-Hasan-8503/Exodius-70B matches part of abacusai/Smaug-72B-v0.1 (723 of its 1043 tensors; the rest of abacusai/Smaug-72B-v0.1 has no counterpart). On the shared part: Fine-tune: 95% of the weights changed, across 80 of 80 layers.
Tensors changed: Embeddings 0/1, Norms 16/161, MLP 237/240, Attention 316/320, lm_head 0/1
2 derivatives checked
- mradermacher/Exodius-70B-i1-GGUF: Quantization
- mradermacher/Exodius-70B-GGUF: Quantization