Hcompany/NeoMME-260M-Retriever
Full fine-tune vs Hcompany/NeoMME-260M. Fine-tune: 100% of the weights changed, across 17 of 17 layers.
Tensors changed: Embeddings 8/8, other 17/17, MLP 34/34, Attention 102/102
2 derivatives checked
- Hcompany/NeoMME-260M-Retriever-ST-late: Same weights, repacked
- Hcompany/NeoMME-260M-Retriever-ST-dense: Same weights, repacked
1 other repos hold byte-for-byte identical weight files.