ilevytate/Fusion-7B-Quintessence
Full fine-tune (fewer layers) vs NousResearch/Genstruct-7B. NousResearch/Genstruct-7B matches part of ilevytate/Fusion-7B-Quintessence (289 of its 867 tensors; the rest of ilevytate/Fusion-7B-Quintessence has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Norms 17/65, MLP 96/96, Attention 128/128
2 derivatives checked
- mradermacher/Fusion-7B-Quintessence-i1-GGUF: Quantization
- mradermacher/Fusion-7B-Quintessence-GGUF: Quantization